commit 141a8c8c5e3e4e5a223ae57cbcb88ccddef0ac3c Author: ModelHub XC Date: Sun Sep 27 14:32:07 2026 +0800 初始化项目,由ModelHub XC社区提供模型 Model: Neelectric/Ministral-8B-Instruct-2410_SFT_mathsp_ewc_v00.01 Source: Original Platform diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000..52373fe --- /dev/null +++ b/.gitattributes @@ -0,0 +1,36 @@ +*.7z filter=lfs diff=lfs merge=lfs -text +*.arrow filter=lfs diff=lfs merge=lfs -text +*.bin filter=lfs diff=lfs merge=lfs -text +*.bz2 filter=lfs diff=lfs merge=lfs -text +*.ckpt filter=lfs diff=lfs merge=lfs -text +*.ftz filter=lfs diff=lfs merge=lfs -text +*.gz filter=lfs diff=lfs merge=lfs -text +*.h5 filter=lfs diff=lfs merge=lfs -text +*.joblib filter=lfs diff=lfs merge=lfs -text +*.lfs.* filter=lfs diff=lfs merge=lfs -text +*.mlmodel filter=lfs diff=lfs merge=lfs -text +*.model filter=lfs diff=lfs merge=lfs -text +*.msgpack filter=lfs diff=lfs merge=lfs -text +*.npy filter=lfs diff=lfs merge=lfs -text +*.npz filter=lfs diff=lfs merge=lfs -text +*.onnx filter=lfs diff=lfs merge=lfs -text +*.ot filter=lfs diff=lfs merge=lfs -text +*.parquet filter=lfs diff=lfs merge=lfs -text +*.pb filter=lfs diff=lfs merge=lfs -text +*.pickle filter=lfs diff=lfs merge=lfs -text +*.pkl filter=lfs diff=lfs merge=lfs -text +*.pt filter=lfs diff=lfs merge=lfs -text +*.pth filter=lfs diff=lfs merge=lfs -text +*.rar filter=lfs diff=lfs merge=lfs -text +*.safetensors filter=lfs diff=lfs merge=lfs -text +saved_model/**/* filter=lfs diff=lfs merge=lfs -text +*.tar.* filter=lfs diff=lfs merge=lfs -text +*.tar filter=lfs diff=lfs merge=lfs -text +*.tflite filter=lfs diff=lfs merge=lfs -text +*.tgz filter=lfs diff=lfs merge=lfs -text +*.wasm filter=lfs diff=lfs merge=lfs -text +*.xz filter=lfs diff=lfs merge=lfs -text +*.zip filter=lfs diff=lfs merge=lfs -text +*.zst filter=lfs diff=lfs merge=lfs -text +*tfevents* filter=lfs diff=lfs merge=lfs -text +tokenizer.json filter=lfs diff=lfs merge=lfs -text diff --git a/README.md b/README.md new file mode 100644 index 0000000..32f23f2 --- /dev/null +++ b/README.md @@ -0,0 +1,60 @@ +--- +base_model: mistralai/Ministral-8B-Instruct-2410 +datasets: Neelectric/OpenR1-Math-220k_all_Llama3_4096toks +library_name: transformers +model_name: Ministral-8B-Instruct-2410_SFT_mathsp_ewc_v00.01 +tags: +- generated_from_trainer +- open-r1 +- sft +- trl +licence: license +--- + +# Model Card for Ministral-8B-Instruct-2410_SFT_mathsp_ewc_v00.01 + +This model is a fine-tuned version of [mistralai/Ministral-8B-Instruct-2410](https://huggingface.co/mistralai/Ministral-8B-Instruct-2410) on the [Neelectric/OpenR1-Math-220k_all_Llama3_4096toks](https://huggingface.co/datasets/Neelectric/OpenR1-Math-220k_all_Llama3_4096toks) dataset. +It has been trained using [TRL](https://github.com/huggingface/trl). + +## Quick start + +```python +from transformers import pipeline + +question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?" +generator = pipeline("text-generation", model="Neelectric/Ministral-8B-Instruct-2410_SFT_mathsp_ewc_v00.01", device="cuda") +output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0] +print(output["generated_text"]) +``` + +## Training procedure + +[Visualize in Weights & Biases](https://wandb.ai/neelectric/open-r1_math/runs/1hud2ke1) + + + +This model was trained with SFT. + +### Framework versions + +- TRL: 1.1.0.dev0 +- Transformers: 4.57.6 +- Pytorch: 2.9.0 +- Datasets: 5.0.0 +- Tokenizers: 0.22.2 + +## Citations + + + +Cite TRL as: + +```bibtex +@software{vonwerra2020trl, + title = {{TRL: Transformers Reinforcement Learning}}, + author = {von Werra, Leandro and Belkada, Younes and Tunstall, Lewis and Beeching, Edward and Thrush, Tristan and Lambert, Nathan and Huang, Shengyi and Rasul, Kashif and Gallouédec, Quentin}, + license = {Apache-2.0}, + url = {https://github.com/huggingface/trl}, + year = {2020} +} +``` \ No newline at end of file diff --git a/all_results.json b/all_results.json new file mode 100644 index 0000000..bf02d98 --- /dev/null +++ b/all_results.json @@ -0,0 +1,11 @@ +{ + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "total_flos": 5.751448627479105e+19, + "train_loss": 0.3408017477382914, + "train_runtime": 50265.1743, + "train_samples": 125770, + "train_samples_per_second": 7.506, + "train_steps_per_second": 0.469 +} \ No newline at end of file diff --git a/chat_template.jinja b/chat_template.jinja new file mode 100644 index 0000000..315517a --- /dev/null +++ b/chat_template.jinja @@ -0,0 +1,92 @@ +{%- if messages[0]["role"] == "system" %} + {%- set system_message = messages[0]["content"] %} + {%- set loop_messages = messages[1:] %} +{%- else %} + {%- set system_message = "You are a helpful AI Assistant that provides well-reasoned and detailed responses. You first think about the reasoning process as an internal monologue and then provide the user with the answer. Respond in the following format: \n...\n\n\n...\n" %} + {%- set loop_messages = messages %} +{%- endif %} +{%- if not tools is defined %} + {%- set tools = none %} +{%- endif %} +{%- set user_messages = loop_messages | selectattr("role", "equalto", "user") | list %} + +{#- This block checks for alternating user/assistant messages, skipping tool calling messages #} +{%- set ns = namespace() %} +{%- set ns.index = 0 %} +{%- for message in loop_messages %} + {%- if not (message.role == "tool" or message.role == "tool_results" or (message.tool_calls is defined and message.tool_calls is not none)) %} + {%- if (message["role"] == "user") != (ns.index % 2 == 0) %} + {{- raise_exception("After the optional system message, conversation roles must alternate user/assistant/user/assistant/...") }} + {%- endif %} + {%- set ns.index = ns.index + 1 %} + {%- endif %} +{%- endfor %} + +{{- bos_token }} +{%- for message in loop_messages %} + {%- if message["role"] == "user" %} + {%- if tools is not none and (message == user_messages[-1]) %} + {{- "[AVAILABLE_TOOLS][" }} + {%- for tool in tools %} + {%- set tool = tool.function %} + {{- '{"type": "function", "function": {' }} + {%- for key, val in tool.items() if key != "return" %} + {%- if val is string %} + {{- '"' + key + '": "' + val + '"' }} + {%- else %} + {{- '"' + key + '": ' + val|tojson }} + {%- endif %} + {%- if not loop.last %} + {{- ", " }} + {%- endif %} + {%- endfor %} + {{- "}}" }} + {%- if not loop.last %} + {{- ", " }} + {%- else %} + {{- "]" }} + {%- endif %} + {%- endfor %} + {{- "[/AVAILABLE_TOOLS]" }} + {%- endif %} + {%- if message == user_messages[-1] and system_message is defined %} + {{- "[INST]" + system_message + "\n\n" + message["content"] + "[/INST]" }} + {%- else %} + {{- "[INST]" + message["content"] + "[/INST]" }} + {%- endif %} + {%- elif (message.tool_calls is defined and message.tool_calls is not none) %} + {%- generation %} + {{- "[TOOL_CALLS][" }} + {%- for tool_call in message.tool_calls %} + {%- set out = tool_call.function|tojson %} + {{- out[:-1] }} + {%- if not tool_call.id is defined or tool_call.id|length != 9 %} + {{- raise_exception("Tool call IDs should be alphanumeric strings with length 9!") }} + {%- endif %} + {{- ', "id": "' + tool_call.id + '"}' }} + {%- if not loop.last %} + {{- ", " }} + {%- else %} + {{- "]" + eos_token }} + {%- endif %} + {%- endfor %} + {%- endgeneration %} + {%- elif message["role"] == "assistant" %} + {%- generation %} + {{- message["content"] + eos_token}} + {%- endgeneration %} + {%- elif message["role"] == "tool_results" or message["role"] == "tool" %} + {%- if message.content is defined and message.content.content is defined %} + {%- set content = message.content.content %} + {%- else %} + {%- set content = message.content %} + {%- endif %} + {{- '[TOOL_RESULTS]{"content": ' + content|string + ", " }} + {%- if not message.tool_call_id is defined or message.tool_call_id|length != 9 %} + {{- raise_exception("Tool call IDs should be alphanumeric strings with length 9!") }} + {%- endif %} + {{- '"call_id": "' + message.tool_call_id + '"}[/TOOL_RESULTS]' }} + {%- else %} + {{- raise_exception("Only user and assistant roles are supported, with the exception of an initial optional system message!") }} + {%- endif %} +{%- endfor %} \ No newline at end of file diff --git a/config.json b/config.json new file mode 100644 index 0000000..033f6c6 --- /dev/null +++ b/config.json @@ -0,0 +1,65 @@ +{ + "architectures": [ + "MinistralForCausalLM" + ], + "attention_dropout": 0.0, + "bos_token_id": 1, + "dtype": "bfloat16", + "eos_token_id": 0, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 4096, + "initializer_range": 0.02, + "intermediate_size": 12288, + "layer_types": [ + "full_attention", + "sliding_attention", + "sliding_attention", + "sliding_attention", + "full_attention", + "sliding_attention", + "sliding_attention", + "sliding_attention", + "full_attention", + "sliding_attention", + "sliding_attention", + "sliding_attention", + "full_attention", + "sliding_attention", + "sliding_attention", + "sliding_attention", + "full_attention", + "sliding_attention", + "sliding_attention", + "sliding_attention", + "full_attention", + "sliding_attention", + "sliding_attention", + "sliding_attention", + "full_attention", + "sliding_attention", + "sliding_attention", + "sliding_attention", + "full_attention", + "sliding_attention", + "sliding_attention", + "sliding_attention", + "full_attention", + "sliding_attention", + "sliding_attention", + "sliding_attention" + ], + "max_position_embeddings": 32768, + "model_type": "ministral", + "num_attention_heads": 32, + "num_hidden_layers": 36, + "num_key_value_heads": 8, + "pad_token_id": 2, + "rms_norm_eps": 1e-05, + "rope_theta": 100000000.0, + "sliding_window": 32768, + "tie_word_embeddings": false, + "transformers_version": "4.57.6", + "use_cache": true, + "vocab_size": 131072 +} diff --git a/generation_config.json b/generation_config.json new file mode 100644 index 0000000..19a44f5 --- /dev/null +++ b/generation_config.json @@ -0,0 +1,7 @@ +{ + "_from_model_config": true, + "bos_token_id": 1, + "eos_token_id": 0, + "pad_token_id": 2, + "transformers_version": "4.57.6" +} diff --git a/model-00001-of-00004.safetensors b/model-00001-of-00004.safetensors new file mode 100644 index 0000000..ee77fc6 --- /dev/null +++ b/model-00001-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5dd591b8a1b59408db42e726e72095246506c4fb98cf0f1cf325516b8738cb7b +size 4983007904 diff --git a/model-00002-of-00004.safetensors b/model-00002-of-00004.safetensors new file mode 100644 index 0000000..e754d0b --- /dev/null +++ b/model-00002-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b27413cf2c325e7fe7a149ab56c1a8002a8c59bf432f76f404bed665283d4ad4 +size 4999836776 diff --git a/model-00003-of-00004.safetensors b/model-00003-of-00004.safetensors new file mode 100644 index 0000000..cbd6f68 --- /dev/null +++ b/model-00003-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e66ffb880be38d21b9fcc867223d0b36a8f866db051e61b41e46f1fc7581cbb3 +size 4983067960 diff --git a/model-00004-of-00004.safetensors b/model-00004-of-00004.safetensors new file mode 100644 index 0000000..51cee61 --- /dev/null +++ b/model-00004-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bafd846c475c8f8d283e39db9fcda14a8699604ba14c5b533d40089c0f9f3f25 +size 1073741952 diff --git a/model.safetensors.index.json b/model.safetensors.index.json new file mode 100644 index 0000000..01b203f --- /dev/null +++ b/model.safetensors.index.json @@ -0,0 +1,335 @@ +{ + "metadata": { + "total_parameters": 8019808256, + "total_size": 16039616512 + }, + "weight_map": { + "lm_head.weight": "model-00004-of-00004.safetensors", + "model.embed_tokens.weight": "model-00001-of-00004.safetensors", + "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.20.input_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.20.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.21.input_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.22.input_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.32.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.32.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.32.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.33.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.33.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.33.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.33.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.33.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.33.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.33.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.33.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.33.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.34.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.34.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.34.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.34.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.34.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.34.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.34.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.34.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.34.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.35.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.35.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.35.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.35.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.35.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.35.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.35.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.35.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.35.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.7.input_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.8.input_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.8.mlp.down_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.8.mlp.up_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.9.input_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00004.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00004.safetensors", + "model.norm.weight": "model-00003-of-00004.safetensors" + } +} diff --git a/special_tokens_map.json b/special_tokens_map.json new file mode 100644 index 0000000..2310ad9 --- /dev/null +++ b/special_tokens_map.json @@ -0,0 +1,18 @@ +{ + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": "", + "pad_token": "", + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/tokenizer.json b/tokenizer.json new file mode 100644 index 0000000..ff212c4 --- /dev/null +++ b/tokenizer.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7edbeaf20dd7f571b5dd1c54d9ace4f9b6299127cc7ba2afb14a6d51a4a79a4 +size 17078136 diff --git a/tokenizer_config.json b/tokenizer_config.json new file mode 100644 index 0000000..6708225 --- /dev/null +++ b/tokenizer_config.json @@ -0,0 +1,8017 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "add_prefix_space": null, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "3": { + "content": "[INST]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "4": { + "content": "[/INST]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "5": { + "content": "[AVAILABLE_TOOLS]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "6": { + "content": "[/AVAILABLE_TOOLS]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "7": { + "content": "[TOOL_RESULTS]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "8": { + "content": "[/TOOL_RESULTS]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "9": { + "content": "[TOOL_CALLS]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "10": { + "content": "[IMG]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "11": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "12": { + "content": "[IMG_BREAK]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "13": { + "content": "[IMG_END]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "14": { + "content": "[PREFIX]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "15": { + "content": "[MIDDLE]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "16": { + "content": "[SUFFIX]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "17": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "18": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "19": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "20": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "21": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "22": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "23": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "24": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "25": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "26": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "27": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "28": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "29": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "30": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "31": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "32": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "33": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "34": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "35": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "36": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "37": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "38": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "39": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "40": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "41": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "42": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "43": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "44": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "45": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "46": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "47": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "48": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "49": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "50": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "51": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "52": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "53": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "54": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "55": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "56": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "57": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "58": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "59": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "60": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "61": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "62": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "63": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "64": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "65": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "66": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "67": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "68": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "69": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "70": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "71": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "72": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "73": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "74": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "75": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "76": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "77": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "78": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "79": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "80": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "81": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "82": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "83": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "84": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "85": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "86": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "87": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "88": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "89": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "90": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "91": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "92": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "93": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "94": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "95": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "96": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "97": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "98": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "99": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "100": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "101": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "102": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "103": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "104": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "105": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "106": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "107": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "108": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "109": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "110": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "111": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "112": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "113": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "114": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "115": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "116": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "117": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "118": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "119": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "120": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "121": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "122": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "123": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "124": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "125": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "126": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "127": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "129": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "130": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "131": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "132": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "133": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "134": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "135": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "136": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "137": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "138": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "139": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "140": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "141": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "142": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "143": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "144": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "145": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "146": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "147": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "148": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "149": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "150": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "152": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "153": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "154": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "155": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "156": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "157": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "158": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "159": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "160": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "161": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "162": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "163": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "164": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "165": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "166": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "167": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "168": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "169": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "170": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "171": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "172": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "173": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "174": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "175": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "176": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "177": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "178": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "179": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "180": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "181": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "182": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "183": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "184": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "185": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "186": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "187": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "188": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "189": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "190": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "191": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "192": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "193": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "194": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "195": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "196": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "197": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "198": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "199": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "200": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "201": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "202": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "203": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "204": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "205": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "206": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "207": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "208": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "209": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "210": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "211": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "212": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "213": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "214": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "215": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "216": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "217": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "218": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "219": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "220": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "221": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "222": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "223": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "224": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "225": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "226": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "227": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "228": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "229": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "230": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "231": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "232": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "233": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "234": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "235": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "236": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "237": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "238": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "239": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "240": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "241": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "242": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "243": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "244": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "245": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "246": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "247": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "248": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "249": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "250": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "251": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "252": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "253": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "254": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "255": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "256": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "257": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "258": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "259": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "260": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "261": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "262": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "263": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "264": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "265": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "266": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "267": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "268": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "269": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "270": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "271": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "272": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "273": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "274": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "275": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "276": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "277": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "278": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "279": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "280": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "281": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "282": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "283": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "284": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "285": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "286": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "287": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "288": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "289": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "290": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "291": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "292": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "293": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "294": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "295": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "296": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "297": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "298": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "299": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "300": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "301": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "302": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "303": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "304": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "305": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "306": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "307": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "308": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "309": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "310": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "311": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "312": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "313": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "314": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "315": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "316": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "317": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "318": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "319": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "320": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "321": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "322": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "323": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "324": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "325": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "326": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "327": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "328": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "329": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "330": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "331": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "332": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "333": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "334": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "335": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "336": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "337": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "338": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "339": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "340": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "341": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "342": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "343": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "344": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "345": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "346": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "347": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "348": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "349": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "350": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "351": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "352": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "353": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "354": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "355": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "356": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "357": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "358": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "359": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "360": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "361": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "362": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "363": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "364": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "365": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "366": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "367": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "368": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "369": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "370": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "371": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "372": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "373": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "374": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "375": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "376": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "377": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "378": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "379": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "380": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "381": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "382": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "383": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "384": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "385": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "386": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "387": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "388": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "389": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "390": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "391": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "392": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "393": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "394": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "395": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "396": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "397": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "398": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "399": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "400": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "401": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "402": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "403": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "404": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "405": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "406": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "407": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "408": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "409": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "410": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "411": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "412": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "413": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "414": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "415": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "416": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "417": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "418": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "419": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "420": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "421": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "422": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "423": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "424": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "425": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "426": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "427": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "428": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "429": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "430": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "431": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "432": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "433": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "434": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "435": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "436": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "437": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "438": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "439": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "440": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "441": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "442": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "443": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "444": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "445": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "446": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "447": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "448": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "449": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "450": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "451": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "452": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "453": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "454": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "455": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "456": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "457": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "458": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "459": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "460": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "461": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "462": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "463": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "464": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "465": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "466": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "467": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "468": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "469": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "470": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "471": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "472": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "473": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "474": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "475": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "476": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "477": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "478": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "479": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "480": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "481": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "482": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "483": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "484": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "485": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "486": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "487": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "488": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "489": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "490": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "491": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "492": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "493": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "494": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "495": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "496": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "497": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "498": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "499": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "500": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "501": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "502": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "503": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "504": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "505": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "506": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "507": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "508": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "509": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "510": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "511": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "512": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "513": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "514": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "515": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "516": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "517": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "518": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "519": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "520": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "521": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "522": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "523": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "524": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "525": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "526": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "527": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "528": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "529": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "530": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "531": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "532": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "533": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "534": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "535": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "536": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "537": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "538": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "539": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "540": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "541": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "542": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "543": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "544": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "545": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "546": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "547": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "548": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "549": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "550": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "551": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "552": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "553": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "554": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "555": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "556": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "557": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "558": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "559": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "560": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "561": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "562": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "563": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "564": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "565": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "566": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "567": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "568": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "569": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "570": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "571": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "572": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "573": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "574": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "575": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "576": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "577": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "578": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "579": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "580": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "581": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "582": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "583": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "584": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "585": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "586": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "587": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "588": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "589": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "590": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "591": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "592": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "593": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "594": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "595": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "596": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "597": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "598": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "599": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "600": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "601": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "602": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "603": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "604": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "605": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "606": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "607": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "608": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "609": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "610": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "611": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "612": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "613": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "614": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "615": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "616": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "617": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "618": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "619": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "620": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "621": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "622": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "623": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "624": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "625": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "626": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "627": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "628": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "629": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "630": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "631": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "632": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "633": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "634": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "635": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "636": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "637": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "638": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "639": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "640": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "641": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "642": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "643": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "644": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "645": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "646": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "647": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "648": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "649": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "650": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "651": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "652": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "653": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "654": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "655": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "656": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "657": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "658": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "659": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "660": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "661": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "662": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "663": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "664": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "665": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "666": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "667": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "668": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "669": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "670": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "671": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "672": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "673": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "674": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "675": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "676": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "677": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "678": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "679": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "680": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "681": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "682": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "683": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "684": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "685": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "686": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "687": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "688": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "689": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "690": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "691": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "692": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "693": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "694": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "695": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "696": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "697": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "698": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "699": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "700": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "701": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "702": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "703": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "704": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "705": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "706": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "707": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "708": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "709": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "710": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "711": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "712": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "713": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "714": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "715": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "716": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "717": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "718": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "719": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "720": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "721": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "722": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "723": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "724": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "725": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "726": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "727": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "728": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "729": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "730": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "731": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "732": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "733": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "734": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "735": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "736": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "737": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "738": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "739": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "740": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "741": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "742": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "743": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "744": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "745": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "746": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "747": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "748": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "749": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "750": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "751": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "752": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "753": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "754": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "755": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "756": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "757": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "758": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "759": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "760": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "761": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "762": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "763": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "764": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "765": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "766": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "767": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "768": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "769": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "770": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "771": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "772": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "773": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "774": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "775": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "776": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "777": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "778": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "779": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "780": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "781": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "782": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "783": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "784": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "785": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "786": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "787": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "788": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "789": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "790": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "791": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "792": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "793": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "794": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "795": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "796": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "797": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "798": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "799": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "800": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "801": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "802": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "803": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "804": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "805": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "806": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "807": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "808": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "809": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "810": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "811": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "812": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "813": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "814": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "815": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "816": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "817": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "818": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "819": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "820": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "821": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "822": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "823": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "824": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "825": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "826": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "827": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "828": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "829": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "830": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "831": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "832": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "833": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "834": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "835": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "836": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "837": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "838": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "839": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "840": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "841": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "842": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "843": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "844": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "845": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "846": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "847": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "848": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "849": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "850": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "851": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "852": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "853": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "854": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "855": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "856": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "857": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "858": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "859": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "860": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "861": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "862": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "863": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "864": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "865": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "866": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "867": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "868": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "869": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "870": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "871": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "872": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "873": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "874": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "875": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "876": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "877": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "878": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "879": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "880": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "881": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "882": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "883": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "884": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "885": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "886": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "887": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "888": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "889": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "890": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "891": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "892": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "893": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "894": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "895": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "896": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "897": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "898": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "899": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "900": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "901": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "902": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "903": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "904": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "905": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "906": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "907": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "908": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "909": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "910": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "911": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "912": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "913": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "914": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "915": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "916": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "917": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "918": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "919": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "920": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "921": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "922": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "923": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "924": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "925": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "926": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "927": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "928": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "929": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "930": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "931": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "932": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "933": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "934": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "935": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "936": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "937": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "938": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "939": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "940": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "941": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "942": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "943": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "944": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "945": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "946": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "947": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "948": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "949": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "950": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "951": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "952": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "953": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "954": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "955": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "956": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "957": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "958": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "959": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "960": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "961": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "962": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "963": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "964": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "965": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "966": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "967": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "968": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "969": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "970": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "971": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "972": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "973": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "974": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "975": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "976": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "977": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "978": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "979": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "980": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "981": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "982": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "983": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "984": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "985": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "986": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "987": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "988": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "989": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "990": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "991": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "992": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "993": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "994": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "995": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "996": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "997": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "998": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "999": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "bos_token": "", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "extra_special_tokens": {}, + "legacy": true, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "tokenizer_class": "LlamaTokenizerFast", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/train_results.json b/train_results.json new file mode 100644 index 0000000..bf02d98 --- /dev/null +++ b/train_results.json @@ -0,0 +1,11 @@ +{ + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "total_flos": 5.751448627479105e+19, + "train_loss": 0.3408017477382914, + "train_runtime": 50265.1743, + "train_samples": 125770, + "train_samples_per_second": 7.506, + "train_steps_per_second": 0.469 +} \ No newline at end of file diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000..9a94225 --- /dev/null +++ b/trainer_state.json @@ -0,0 +1,283042 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 3.0, + "eval_steps": 500, + "global_step": 23583, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.00012721027859051011, + "ewc_loss": 0.0, + "ewc_loss_diag": 0.0, + "ewc_loss_parallel": 0.0, + "grad_norm": 8.018087387084961, + "learning_rate": 0.0, + "loss": 0.7412, + "mean_token_accuracy": 0.7922049164772034, + "num_tokens": 38786.0, + "step": 1 + }, + { + "epoch": 0.00025442055718102023, + "ewc_loss": 0.0, + "ewc_loss_diag": 0.0, + "ewc_loss_parallel": 0.0, + "grad_norm": 7.368358612060547, + "learning_rate": 4.2390843577787196e-10, + "loss": 0.7841, + "mean_token_accuracy": 0.7776151895523071, + "num_tokens": 80672.0, + "step": 2 + }, + { + "epoch": 0.0003816308357715303, + "ewc_loss": 6.2727600891321345e-15, + "ewc_loss_diag": 9.690056916589196e-19, + "ewc_loss_parallel": 6.179952383167375e-18, + "grad_norm": 8.411513328552246, + "learning_rate": 8.478168715557439e-10, + "loss": 0.6622, + "mean_token_accuracy": 0.8157488107681274, + "num_tokens": 120184.0, + "step": 3 + }, + { + "epoch": 0.0005088411143620405, + "ewc_loss": 2.327027459614328e-13, + "ewc_loss_diag": 2.5587171270657905e-17, + "ewc_loss_parallel": 2.3071822230491534e-16, + "grad_norm": 8.208902359008789, + "learning_rate": 1.271725307333616e-09, + "loss": 0.7541, + "mean_token_accuracy": 0.7935553789138794, + "num_tokens": 152191.0, + "step": 4 + }, + { + "epoch": 0.0006360513929525506, + "ewc_loss": 1.9184653865522705e-12, + "ewc_loss_diag": 1.9168694409543718e-16, + "ewc_loss_parallel": 1.9012569296705806e-15, + "grad_norm": 6.45544958114624, + "learning_rate": 1.6956337431114878e-09, + "loss": 0.7316, + "mean_token_accuracy": 0.793449878692627, + "num_tokens": 196322.0, + "step": 5 + }, + { + "epoch": 0.0007632616715430606, + "ewc_loss": 1.0061285138363019e-11, + "ewc_loss_diag": 1.2490009027033011e-15, + "ewc_loss_parallel": 9.936496070395151e-15, + "grad_norm": 8.997086524963379, + "learning_rate": 2.1195421788893596e-09, + "loss": 0.7481, + "mean_token_accuracy": 0.7930362224578857, + "num_tokens": 230662.0, + "step": 6 + }, + { + "epoch": 0.0008904719501335708, + "ewc_loss": 2.05773176276125e-11, + "ewc_loss_diag": 2.0816681711721685e-15, + "ewc_loss_parallel": 2.0317081350640365e-14, + "grad_norm": 7.013577461242676, + "learning_rate": 2.543450614667232e-09, + "loss": 0.7255, + "mean_token_accuracy": 0.7985407710075378, + "num_tokens": 269859.0, + "step": 7 + }, + { + "epoch": 0.001017682228724081, + "ewc_loss": 8.049028110690415e-11, + "ewc_loss_diag": 1.1657341758564144e-14, + "ewc_loss_parallel": 7.904787935331115e-14, + "grad_norm": 7.530651569366455, + "learning_rate": 2.967359050445104e-09, + "loss": 0.689, + "mean_token_accuracy": 0.8059301376342773, + "num_tokens": 305413.0, + "step": 8 + }, + { + "epoch": 0.001144892507314591, + "ewc_loss": 1.318767317570746e-10, + "ewc_loss_diag": 1.63202784619898e-14, + "ewc_loss_parallel": 1.2967404927621828e-13, + "grad_norm": 6.8073225021362305, + "learning_rate": 3.3912674862229757e-09, + "loss": 0.728, + "mean_token_accuracy": 0.7959203124046326, + "num_tokens": 349032.0, + "step": 9 + }, + { + "epoch": 0.0012721027859051012, + "ewc_loss": 2.2100721253082156e-10, + "ewc_loss_diag": 3.952393967665557e-14, + "ewc_loss_parallel": 2.1760371282653068e-13, + "grad_norm": 8.444629669189453, + "learning_rate": 3.815175922000847e-09, + "loss": 0.7754, + "mean_token_accuracy": 0.7844054102897644, + "num_tokens": 383205.0, + "step": 10 + }, + { + "epoch": 0.0013993130644956112, + "ewc_loss": 7.057678885757923e-10, + "ewc_loss_diag": 8.79296635503124e-14, + "ewc_loss_parallel": 6.998845947236987e-13, + "grad_norm": 8.04902458190918, + "learning_rate": 4.239084357778719e-09, + "loss": 0.7493, + "mean_token_accuracy": 0.7881491184234619, + "num_tokens": 424559.0, + "step": 11 + }, + { + "epoch": 0.0015265233430861213, + "ewc_loss": 9.677023626863956e-10, + "ewc_loss_diag": 1.1546319456101628e-13, + "ewc_loss_parallel": 9.521272659185342e-13, + "grad_norm": 8.727396965026855, + "learning_rate": 4.662992793556591e-09, + "loss": 0.7621, + "mean_token_accuracy": 0.7904402017593384, + "num_tokens": 457609.0, + "step": 12 + }, + { + "epoch": 0.0016537336216766315, + "ewc_loss": 1.2441887520253658e-09, + "ewc_loss_diag": 1.4566126083082054e-13, + "ewc_loss_parallel": 1.2292389328649733e-12, + "grad_norm": 8.569257736206055, + "learning_rate": 5.086901229334464e-09, + "loss": 0.8269, + "mean_token_accuracy": 0.7692526578903198, + "num_tokens": 489456.0, + "step": 13 + }, + { + "epoch": 0.0017809439002671415, + "ewc_loss": 3.2159732654690742e-09, + "ewc_loss_diag": 5.648814749292796e-13, + "ewc_loss_parallel": 3.154809746774845e-12, + "grad_norm": 8.702582359313965, + "learning_rate": 5.510809665112336e-09, + "loss": 0.7266, + "mean_token_accuracy": 0.7987388372421265, + "num_tokens": 535340.0, + "step": 14 + }, + { + "epoch": 0.0019081541788576518, + "ewc_loss": 5.558831617236137e-09, + "ewc_loss_diag": 7.709388682997087e-13, + "ewc_loss_parallel": 5.4853899200679734e-12, + "grad_norm": 7.4282331466674805, + "learning_rate": 5.934718100890208e-09, + "loss": 0.6838, + "mean_token_accuracy": 0.8097742199897766, + "num_tokens": 575698.0, + "step": 15 + }, + { + "epoch": 0.002035364457448162, + "ewc_loss": 6.810296326875687e-09, + "ewc_loss_diag": 8.917311333789257e-13, + "ewc_loss_parallel": 6.73594513500575e-12, + "grad_norm": 7.952362537384033, + "learning_rate": 6.3586265366680796e-09, + "loss": 0.7839, + "mean_token_accuracy": 0.7791250944137573, + "num_tokens": 611403.0, + "step": 16 + }, + { + "epoch": 0.002162574736038672, + "ewc_loss": 8.614733815193176e-09, + "ewc_loss_diag": 1.0871303857129533e-12, + "ewc_loss_parallel": 8.469669410260394e-12, + "grad_norm": 7.411961078643799, + "learning_rate": 6.782534972445951e-09, + "loss": 0.7522, + "mean_token_accuracy": 0.7871662974357605, + "num_tokens": 648714.0, + "step": 17 + }, + { + "epoch": 0.002289785014629182, + "ewc_loss": 9.778887033462524e-09, + "ewc_loss_diag": 1.2434497875801753e-12, + "ewc_loss_parallel": 9.663381206337363e-12, + "grad_norm": 7.757039546966553, + "learning_rate": 7.206443408223823e-09, + "loss": 0.8009, + "mean_token_accuracy": 0.7765024900436401, + "num_tokens": 689272.0, + "step": 18 + }, + { + "epoch": 0.0024169952932196924, + "ewc_loss": 2.153683453798294e-08, + "ewc_loss_diag": 4.206412995699793e-12, + "ewc_loss_parallel": 2.1032064978498966e-11, + "grad_norm": 7.132312774658203, + "learning_rate": 7.630351844001695e-09, + "loss": 0.7134, + "mean_token_accuracy": 0.7966217994689941, + "num_tokens": 724064.0, + "step": 19 + }, + { + "epoch": 0.0025442055718102024, + "ewc_loss": 3.5390257835388184e-08, + "ewc_loss_diag": 5.7411853049416095e-12, + "ewc_loss_parallel": 3.501554601825774e-11, + "grad_norm": 7.37175178527832, + "learning_rate": 8.054260279779567e-09, + "loss": 0.7231, + "mean_token_accuracy": 0.7974886298179626, + "num_tokens": 765715.0, + "step": 20 + }, + { + "epoch": 0.0026714158504007124, + "ewc_loss": 4.516914486885071e-08, + "ewc_loss_diag": 6.59383658785373e-12, + "ewc_loss_parallel": 4.4565240386873484e-11, + "grad_norm": 8.500150680541992, + "learning_rate": 8.478168715557438e-09, + "loss": 0.7756, + "mean_token_accuracy": 0.7857774496078491, + "num_tokens": 800775.0, + "step": 21 + }, + { + "epoch": 0.0027986261289912225, + "ewc_loss": 5.145557224750519e-08, + "ewc_loss_diag": 7.275957614183426e-12, + "ewc_loss_parallel": 5.070432962384075e-11, + "grad_norm": 7.7503180503845215, + "learning_rate": 8.902077151335311e-09, + "loss": 0.7568, + "mean_token_accuracy": 0.7907545566558838, + "num_tokens": 835271.0, + "step": 22 + }, + { + "epoch": 0.0029258364075817325, + "ewc_loss": 6.193295121192932e-08, + "ewc_loss_diag": 8.697043085703626e-12, + "ewc_loss_parallel": 6.093614501878619e-11, + "grad_norm": 9.52897834777832, + "learning_rate": 9.325985587113182e-09, + "loss": 0.715, + "mean_token_accuracy": 0.7998312711715698, + "num_tokens": 872011.0, + "step": 23 + }, + { + "epoch": 0.0030530466861722425, + "ewc_loss": 6.891787052154541e-08, + "ewc_loss_diag": 9.549694368615746e-12, + "ewc_loss_parallel": 6.821210263296962e-11, + "grad_norm": 9.502488136291504, + "learning_rate": 9.749894022891054e-09, + "loss": 0.7692, + "mean_token_accuracy": 0.7824834585189819, + "num_tokens": 903630.0, + "step": 24 + }, + { + "epoch": 0.003180256964762753, + "ewc_loss": 7.59027898311615e-08, + "ewc_loss_diag": 1.0402345651527867e-11, + "ewc_loss_parallel": 7.503331289626658e-11, + "grad_norm": 7.623968124389648, + "learning_rate": 1.0173802458668929e-08, + "loss": 0.7938, + "mean_token_accuracy": 0.774856448173523, + "num_tokens": 946964.0, + "step": 25 + }, + { + "epoch": 0.003307467243353263, + "ewc_loss": 1.3690441846847534e-07, + "ewc_loss_diag": 2.8990143619012088e-11, + "ewc_loss_parallel": 1.3369572116062045e-10, + "grad_norm": 10.177714347839355, + "learning_rate": 1.05977108944468e-08, + "loss": 0.7657, + "mean_token_accuracy": 0.7830474376678467, + "num_tokens": 986103.0, + "step": 26 + }, + { + "epoch": 0.003434677521943773, + "ewc_loss": 2.1979212760925293e-07, + "ewc_loss_diag": 3.865352482534945e-11, + "ewc_loss_parallel": 2.1645973902195692e-10, + "grad_norm": 6.952630519866943, + "learning_rate": 1.1021619330224672e-08, + "loss": 0.7121, + "mean_token_accuracy": 0.801318883895874, + "num_tokens": 1024451.0, + "step": 27 + }, + { + "epoch": 0.003561887800534283, + "ewc_loss": 2.812594175338745e-07, + "ewc_loss_diag": 4.524736141320318e-11, + "ewc_loss_parallel": 2.7830537874251604e-10, + "grad_norm": 8.463916778564453, + "learning_rate": 1.1445527766002543e-08, + "loss": 0.8121, + "mean_token_accuracy": 0.7722006440162659, + "num_tokens": 1062954.0, + "step": 28 + }, + { + "epoch": 0.003689098079124793, + "ewc_loss": 3.371387720108032e-07, + "ewc_loss_diag": 5.002220859751105e-11, + "ewc_loss_parallel": 3.310560714453459e-10, + "grad_norm": 8.315230369567871, + "learning_rate": 1.1869436201780416e-08, + "loss": 0.7114, + "mean_token_accuracy": 0.8011436462402344, + "num_tokens": 1104809.0, + "step": 29 + }, + { + "epoch": 0.0038163083577153036, + "ewc_loss": 3.688037395477295e-07, + "ewc_loss_diag": 5.3432813729159534e-11, + "ewc_loss_parallel": 3.637978807091713e-10, + "grad_norm": 7.620183944702148, + "learning_rate": 1.2293344637558287e-08, + "loss": 0.7449, + "mean_token_accuracy": 0.7956128716468811, + "num_tokens": 1144106.0, + "step": 30 + }, + { + "epoch": 0.003943518636305814, + "ewc_loss": 4.0605664253234863e-07, + "ewc_loss_diag": 6.093614501878619e-11, + "ewc_loss_parallel": 4.001776687800884e-10, + "grad_norm": 8.256895065307617, + "learning_rate": 1.2717253073336159e-08, + "loss": 0.7232, + "mean_token_accuracy": 0.8032463192939758, + "num_tokens": 1181459.0, + "step": 31 + }, + { + "epoch": 0.004070728914896324, + "ewc_loss": 4.5634806156158447e-07, + "ewc_loss_diag": 6.775735528208315e-11, + "ewc_loss_parallel": 4.4929038267582655e-10, + "grad_norm": 7.212483882904053, + "learning_rate": 1.314116150911403e-08, + "loss": 0.7602, + "mean_token_accuracy": 0.7913727760314941, + "num_tokens": 1221712.0, + "step": 32 + }, + { + "epoch": 0.004197939193486834, + "ewc_loss": 4.954636096954346e-07, + "ewc_loss_diag": 7.321432349272072e-11, + "ewc_loss_parallel": 4.874891601502895e-10, + "grad_norm": 8.557799339294434, + "learning_rate": 1.3565069944891903e-08, + "loss": 0.7467, + "mean_token_accuracy": 0.7959757447242737, + "num_tokens": 1262557.0, + "step": 33 + }, + { + "epoch": 0.004325149472077344, + "ewc_loss": 5.327165126800537e-07, + "ewc_loss_diag": 7.776179700158536e-11, + "ewc_loss_parallel": 5.238689482212067e-10, + "grad_norm": 7.02394437789917, + "learning_rate": 1.3988978380669775e-08, + "loss": 0.6923, + "mean_token_accuracy": 0.805105447769165, + "num_tokens": 1306841.0, + "step": 34 + }, + { + "epoch": 0.004452359750667854, + "ewc_loss": 5.662441253662109e-07, + "ewc_loss_diag": 8.685674401931465e-11, + "ewc_loss_parallel": 5.602487362921238e-10, + "grad_norm": 8.334759712219238, + "learning_rate": 1.4412886816447646e-08, + "loss": 0.7224, + "mean_token_accuracy": 0.7987914681434631, + "num_tokens": 1345345.0, + "step": 35 + }, + { + "epoch": 0.004579570029258364, + "ewc_loss": 9.015202522277832e-07, + "ewc_loss_diag": 1.9917933968827128e-10, + "ewc_loss_parallel": 8.840288501232862e-10, + "grad_norm": 10.485506057739258, + "learning_rate": 1.4836795252225519e-08, + "loss": 0.7759, + "mean_token_accuracy": 0.7865338325500488, + "num_tokens": 1384344.0, + "step": 36 + }, + { + "epoch": 0.004706780307848874, + "ewc_loss": 1.3560056686401367e-06, + "ewc_loss_diag": 2.6921043172478676e-10, + "ewc_loss_parallel": 1.3242242857813835e-09, + "grad_norm": 8.690103530883789, + "learning_rate": 1.526070368800339e-08, + "loss": 0.7476, + "mean_token_accuracy": 0.7932512760162354, + "num_tokens": 1422071.0, + "step": 37 + }, + { + "epoch": 0.004833990586439385, + "ewc_loss": 1.780688762664795e-06, + "ewc_loss_diag": 3.092281986027956e-10, + "ewc_loss_parallel": 1.7535057850182056e-09, + "grad_norm": 11.37038803100586, + "learning_rate": 1.5684612123781262e-08, + "loss": 0.7378, + "mean_token_accuracy": 0.7950788140296936, + "num_tokens": 1466821.0, + "step": 38 + }, + { + "epoch": 0.004961200865029895, + "ewc_loss": 2.0563602447509766e-06, + "ewc_loss_diag": 3.383320290595293e-10, + "ewc_loss_parallel": 2.0227162167429924e-09, + "grad_norm": 6.92936897277832, + "learning_rate": 1.6108520559559135e-08, + "loss": 0.6899, + "mean_token_accuracy": 0.8072113990783691, + "num_tokens": 1503091.0, + "step": 39 + }, + { + "epoch": 0.005088411143620405, + "ewc_loss": 2.2798776626586914e-06, + "ewc_loss_diag": 3.67435859516263e-10, + "ewc_loss_parallel": 2.255546860396862e-09, + "grad_norm": 7.102014064788818, + "learning_rate": 1.6532428995337004e-08, + "loss": 0.7431, + "mean_token_accuracy": 0.7907979488372803, + "num_tokens": 1543027.0, + "step": 40 + }, + { + "epoch": 0.005215621422210915, + "ewc_loss": 2.4884939193725586e-06, + "ewc_loss_diag": 3.838067641481757e-10, + "ewc_loss_parallel": 2.444721758365631e-09, + "grad_norm": 6.846263885498047, + "learning_rate": 1.6956337431114877e-08, + "loss": 0.7403, + "mean_token_accuracy": 0.7942314147949219, + "num_tokens": 1581925.0, + "step": 41 + }, + { + "epoch": 0.005342831700801425, + "ewc_loss": 2.637505531311035e-06, + "ewc_loss_diag": 4.0381564758718014e-10, + "ewc_loss_parallel": 2.5902409106492996e-09, + "grad_norm": 8.295856475830078, + "learning_rate": 1.738024586689275e-08, + "loss": 0.6982, + "mean_token_accuracy": 0.805553674697876, + "num_tokens": 1616614.0, + "step": 42 + }, + { + "epoch": 0.005470041979391935, + "ewc_loss": 2.771615982055664e-06, + "ewc_loss_diag": 4.2746250983327627e-10, + "ewc_loss_parallel": 2.7212081477046013e-09, + "grad_norm": 6.922977447509766, + "learning_rate": 1.7804154302670622e-08, + "loss": 0.7325, + "mean_token_accuracy": 0.7938005328178406, + "num_tokens": 1654043.0, + "step": 43 + }, + { + "epoch": 0.005597252257982445, + "ewc_loss": 2.9206275939941406e-06, + "ewc_loss_diag": 4.69299266114831e-10, + "ewc_loss_parallel": 2.8812792152166367e-09, + "grad_norm": 7.203141212463379, + "learning_rate": 1.8228062738448494e-08, + "loss": 0.7591, + "mean_token_accuracy": 0.7867714166641235, + "num_tokens": 1695778.0, + "step": 44 + }, + { + "epoch": 0.005724462536572955, + "ewc_loss": 3.069639205932617e-06, + "ewc_loss_diag": 5.093170329928398e-10, + "ewc_loss_parallel": 3.026798367500305e-09, + "grad_norm": 6.827267169952393, + "learning_rate": 1.8651971174226364e-08, + "loss": 0.726, + "mean_token_accuracy": 0.7952036261558533, + "num_tokens": 1732935.0, + "step": 45 + }, + { + "epoch": 0.005851672815163465, + "ewc_loss": 3.3080577850341797e-06, + "ewc_loss_diag": 5.384208634495735e-10, + "ewc_loss_parallel": 3.245077095925808e-09, + "grad_norm": 7.546327590942383, + "learning_rate": 1.9075879610004236e-08, + "loss": 0.6811, + "mean_token_accuracy": 0.8070294857025146, + "num_tokens": 1777062.0, + "step": 46 + }, + { + "epoch": 0.005978883093753975, + "ewc_loss": 3.591179847717285e-06, + "ewc_loss_diag": 5.711626727133989e-10, + "ewc_loss_parallel": 3.536115400493145e-09, + "grad_norm": 7.367828369140625, + "learning_rate": 1.949978804578211e-08, + "loss": 0.7426, + "mean_token_accuracy": 0.7919274568557739, + "num_tokens": 1813489.0, + "step": 47 + }, + { + "epoch": 0.006106093372344485, + "ewc_loss": 3.7401914596557617e-06, + "ewc_loss_diag": 5.893525667488575e-10, + "ewc_loss_parallel": 3.6816345527768135e-09, + "grad_norm": 7.086641788482666, + "learning_rate": 1.9923696481559985e-08, + "loss": 0.7594, + "mean_token_accuracy": 0.7831497192382812, + "num_tokens": 1852221.0, + "step": 48 + }, + { + "epoch": 0.006233303650934996, + "ewc_loss": 3.9637088775634766e-06, + "ewc_loss_diag": 6.475602276623249e-10, + "ewc_loss_parallel": 3.899913281202316e-09, + "grad_norm": 8.609175682067871, + "learning_rate": 2.0347604917337857e-08, + "loss": 0.6945, + "mean_token_accuracy": 0.8029326796531677, + "num_tokens": 1896155.0, + "step": 49 + }, + { + "epoch": 0.006360513929525506, + "ewc_loss": 5.304813385009766e-06, + "ewc_loss_diag": 1.0550138540565968e-09, + "ewc_loss_parallel": 5.180481821298599e-09, + "grad_norm": 6.574238300323486, + "learning_rate": 2.0771513353115727e-08, + "loss": 0.6548, + "mean_token_accuracy": 0.8165422677993774, + "num_tokens": 1933443.0, + "step": 50 + }, + { + "epoch": 0.006487724208116016, + "ewc_loss": 7.748603820800781e-06, + "ewc_loss_diag": 1.382431946694851e-09, + "ewc_loss_parallel": 7.62520357966423e-09, + "grad_norm": 8.646349906921387, + "learning_rate": 2.11954217888936e-08, + "loss": 0.7598, + "mean_token_accuracy": 0.7879875898361206, + "num_tokens": 1972520.0, + "step": 51 + }, + { + "epoch": 0.006614934486706526, + "ewc_loss": 9.715557098388672e-06, + "ewc_loss_diag": 1.6007106751203537e-09, + "ewc_loss_parallel": 9.546056389808655e-09, + "grad_norm": 9.168891906738281, + "learning_rate": 2.1619330224671472e-08, + "loss": 0.721, + "mean_token_accuracy": 0.7981687784194946, + "num_tokens": 2016041.0, + "step": 52 + }, + { + "epoch": 0.006742144765297036, + "ewc_loss": 1.150369644165039e-05, + "ewc_loss_diag": 1.7898855730891228e-09, + "ewc_loss_parallel": 1.1350493878126144e-08, + "grad_norm": 7.691535472869873, + "learning_rate": 2.2043238660449344e-08, + "loss": 0.7323, + "mean_token_accuracy": 0.7932584285736084, + "num_tokens": 2049352.0, + "step": 53 + }, + { + "epoch": 0.006869355043887546, + "ewc_loss": 1.3053417205810547e-05, + "ewc_loss_diag": 1.9354047253727913e-09, + "ewc_loss_parallel": 1.2863893061876297e-08, + "grad_norm": 7.448302268981934, + "learning_rate": 2.2467147096227214e-08, + "loss": 0.781, + "mean_token_accuracy": 0.7806377410888672, + "num_tokens": 2089922.0, + "step": 54 + }, + { + "epoch": 0.006996565322478056, + "ewc_loss": 1.3887882232666016e-05, + "ewc_loss_diag": 2.066371962428093e-09, + "ewc_loss_parallel": 1.3737007975578308e-08, + "grad_norm": 6.561429977416992, + "learning_rate": 2.2891055532005086e-08, + "loss": 0.7782, + "mean_token_accuracy": 0.7831435203552246, + "num_tokens": 2132350.0, + "step": 55 + }, + { + "epoch": 0.007123775601068566, + "ewc_loss": 1.4901161193847656e-05, + "ewc_loss_diag": 2.2118911147117615e-09, + "ewc_loss_parallel": 1.4668330550193787e-08, + "grad_norm": 8.010149955749512, + "learning_rate": 2.331496396778296e-08, + "loss": 0.7738, + "mean_token_accuracy": 0.7850101590156555, + "num_tokens": 2170149.0, + "step": 56 + }, + { + "epoch": 0.007250985879659076, + "ewc_loss": 1.537799835205078e-05, + "ewc_loss_diag": 2.2846506908535957e-09, + "ewc_loss_parallel": 1.5133991837501526e-08, + "grad_norm": 7.068899154663086, + "learning_rate": 2.373887240356083e-08, + "loss": 0.7072, + "mean_token_accuracy": 0.8000180721282959, + "num_tokens": 2215386.0, + "step": 57 + }, + { + "epoch": 0.007378196158249586, + "ewc_loss": 1.5735626220703125e-05, + "ewc_loss_diag": 2.371962182223797e-09, + "ewc_loss_parallel": 1.548323780298233e-08, + "grad_norm": 7.1047043800354, + "learning_rate": 2.4162780839338704e-08, + "loss": 0.6737, + "mean_token_accuracy": 0.8115779161453247, + "num_tokens": 2256925.0, + "step": 58 + }, + { + "epoch": 0.007505406436840096, + "ewc_loss": 1.609325408935547e-05, + "ewc_loss_diag": 2.444721758365631e-09, + "ewc_loss_parallel": 1.5832483768463135e-08, + "grad_norm": 8.029025077819824, + "learning_rate": 2.4586689275116573e-08, + "loss": 0.7824, + "mean_token_accuracy": 0.7823240756988525, + "num_tokens": 2290305.0, + "step": 59 + }, + { + "epoch": 0.007632616715430607, + "ewc_loss": 1.6450881958007812e-05, + "ewc_loss_diag": 2.5756889954209328e-09, + "ewc_loss_parallel": 1.618172973394394e-08, + "grad_norm": 6.535834312438965, + "learning_rate": 2.5010597710894446e-08, + "loss": 0.7326, + "mean_token_accuracy": 0.7902361154556274, + "num_tokens": 2332029.0, + "step": 60 + }, + { + "epoch": 0.007759826994021117, + "ewc_loss": 1.6689300537109375e-05, + "ewc_loss_diag": 2.837623469531536e-09, + "ewc_loss_parallel": 1.641456037759781e-08, + "grad_norm": 7.102357387542725, + "learning_rate": 2.5434506146672318e-08, + "loss": 0.7151, + "mean_token_accuracy": 0.7988458871841431, + "num_tokens": 2375604.0, + "step": 61 + }, + { + "epoch": 0.007887037272611627, + "ewc_loss": 1.704692840576172e-05, + "ewc_loss_diag": 2.9976945370435715e-09, + "ewc_loss_parallel": 1.664739102125168e-08, + "grad_norm": 6.843193531036377, + "learning_rate": 2.585841458245019e-08, + "loss": 0.7679, + "mean_token_accuracy": 0.7851330637931824, + "num_tokens": 2413583.0, + "step": 62 + }, + { + "epoch": 0.008014247551202136, + "ewc_loss": 1.7404556274414062e-05, + "ewc_loss_diag": 3.2014213502407074e-09, + "ewc_loss_parallel": 1.6996636986732483e-08, + "grad_norm": 5.873024940490723, + "learning_rate": 2.628232301822806e-08, + "loss": 0.6897, + "mean_token_accuracy": 0.8066288232803345, + "num_tokens": 2456170.0, + "step": 63 + }, + { + "epoch": 0.008141457829792647, + "ewc_loss": 1.7642974853515625e-05, + "ewc_loss_diag": 3.3032847568392754e-09, + "ewc_loss_parallel": 1.7229467630386353e-08, + "grad_norm": 6.6844000816345215, + "learning_rate": 2.6706231454005933e-08, + "loss": 0.7337, + "mean_token_accuracy": 0.794657826423645, + "num_tokens": 2500303.0, + "step": 64 + }, + { + "epoch": 0.008268668108383158, + "ewc_loss": 1.800060272216797e-05, + "ewc_loss_diag": 3.434251993894577e-09, + "ewc_loss_parallel": 1.7695128917694092e-08, + "grad_norm": 8.860391616821289, + "learning_rate": 2.7130139889783805e-08, + "loss": 0.8124, + "mean_token_accuracy": 0.7734394669532776, + "num_tokens": 2530232.0, + "step": 65 + }, + { + "epoch": 0.008395878386973667, + "ewc_loss": 1.823902130126953e-05, + "ewc_loss_diag": 3.6088749766349792e-09, + "ewc_loss_parallel": 1.792795956134796e-08, + "grad_norm": 6.8680033683776855, + "learning_rate": 2.7554048325561678e-08, + "loss": 0.7319, + "mean_token_accuracy": 0.7914441823959351, + "num_tokens": 2575652.0, + "step": 66 + }, + { + "epoch": 0.008523088665564178, + "ewc_loss": 1.895427703857422e-05, + "ewc_loss_diag": 3.754394128918648e-09, + "ewc_loss_parallel": 1.862645149230957e-08, + "grad_norm": 6.918760776519775, + "learning_rate": 2.797795676133955e-08, + "loss": 0.7411, + "mean_token_accuracy": 0.7912634611129761, + "num_tokens": 2621129.0, + "step": 67 + }, + { + "epoch": 0.008650298944154687, + "ewc_loss": 2.0503997802734375e-05, + "ewc_loss_diag": 3.987224772572517e-09, + "ewc_loss_parallel": 2.0139850676059723e-08, + "grad_norm": 7.277801513671875, + "learning_rate": 2.840186519711742e-08, + "loss": 0.7557, + "mean_token_accuracy": 0.788272500038147, + "num_tokens": 2659663.0, + "step": 68 + }, + { + "epoch": 0.008777509222745198, + "ewc_loss": 2.491474151611328e-05, + "ewc_loss_diag": 4.423782229423523e-09, + "ewc_loss_parallel": 2.444721758365631e-08, + "grad_norm": 10.133346557617188, + "learning_rate": 2.8825773632895292e-08, + "loss": 0.7574, + "mean_token_accuracy": 0.7881383895874023, + "num_tokens": 2695841.0, + "step": 69 + }, + { + "epoch": 0.008904719501335707, + "ewc_loss": 3.361701965332031e-05, + "ewc_loss_diag": 5.3551048040390015e-09, + "ewc_loss_parallel": 3.306195139884949e-08, + "grad_norm": 6.674553394317627, + "learning_rate": 2.9249682068673165e-08, + "loss": 0.6796, + "mean_token_accuracy": 0.8109325170516968, + "num_tokens": 2737777.0, + "step": 70 + }, + { + "epoch": 0.009031929779926218, + "ewc_loss": 4.482269287109375e-05, + "ewc_loss_diag": 6.461050361394882e-09, + "ewc_loss_parallel": 4.400499165058136e-08, + "grad_norm": 7.574827671051025, + "learning_rate": 2.9673590504451037e-08, + "loss": 0.7338, + "mean_token_accuracy": 0.7923822402954102, + "num_tokens": 2772914.0, + "step": 71 + }, + { + "epoch": 0.009159140058516728, + "ewc_loss": 5.3882598876953125e-05, + "ewc_loss_diag": 7.101334631443024e-09, + "ewc_loss_parallel": 5.3085386753082275e-08, + "grad_norm": 6.723392963409424, + "learning_rate": 3.0097498940228907e-08, + "loss": 0.6835, + "mean_token_accuracy": 0.8095634579658508, + "num_tokens": 2811601.0, + "step": 72 + }, + { + "epoch": 0.009286350337107238, + "ewc_loss": 6.151199340820312e-05, + "ewc_loss_diag": 7.683411240577698e-09, + "ewc_loss_parallel": 6.05359673500061e-08, + "grad_norm": 7.5216145515441895, + "learning_rate": 3.052140737600678e-08, + "loss": 0.777, + "mean_token_accuracy": 0.7870553731918335, + "num_tokens": 2847656.0, + "step": 73 + }, + { + "epoch": 0.009413560615697748, + "ewc_loss": 6.437301635742188e-05, + "ewc_loss_diag": 8.32369551062584e-09, + "ewc_loss_parallel": 6.332993507385254e-08, + "grad_norm": 7.518307209014893, + "learning_rate": 3.094531581178465e-08, + "loss": 0.7932, + "mean_token_accuracy": 0.7796745896339417, + "num_tokens": 2882931.0, + "step": 74 + }, + { + "epoch": 0.009540770894288259, + "ewc_loss": 6.818771362304688e-05, + "ewc_loss_diag": 8.731149137020111e-09, + "ewc_loss_parallel": 6.705522537231445e-08, + "grad_norm": 8.933274269104004, + "learning_rate": 3.1369224247562524e-08, + "loss": 0.6952, + "mean_token_accuracy": 0.8037612438201904, + "num_tokens": 2931457.0, + "step": 75 + }, + { + "epoch": 0.00966798117287877, + "ewc_loss": 7.200241088867188e-05, + "ewc_loss_diag": 9.255018085241318e-09, + "ewc_loss_parallel": 7.12461769580841e-08, + "grad_norm": 7.059638500213623, + "learning_rate": 3.17931326833404e-08, + "loss": 0.7299, + "mean_token_accuracy": 0.7921668887138367, + "num_tokens": 2966284.0, + "step": 76 + }, + { + "epoch": 0.009795191451469279, + "ewc_loss": 7.43865966796875e-05, + "ewc_loss_diag": 9.604264050722122e-09, + "ewc_loss_parallel": 7.35744833946228e-08, + "grad_norm": 5.590839385986328, + "learning_rate": 3.221704111911827e-08, + "loss": 0.6401, + "mean_token_accuracy": 0.8172075748443604, + "num_tokens": 3012070.0, + "step": 77 + }, + { + "epoch": 0.00992240173005979, + "ewc_loss": 7.43865966796875e-05, + "ewc_loss_diag": 1.0128132998943329e-08, + "ewc_loss_parallel": 7.35744833946228e-08, + "grad_norm": 7.652857303619385, + "learning_rate": 3.264094955489614e-08, + "loss": 0.7453, + "mean_token_accuracy": 0.7935401201248169, + "num_tokens": 3043231.0, + "step": 78 + }, + { + "epoch": 0.010049612008650299, + "ewc_loss": 7.62939453125e-05, + "ewc_loss_diag": 1.05355866253376e-08, + "ewc_loss_parallel": 7.543712854385376e-08, + "grad_norm": 11.2106351852417, + "learning_rate": 3.306485799067401e-08, + "loss": 0.683, + "mean_token_accuracy": 0.8143540620803833, + "num_tokens": 3076648.0, + "step": 79 + }, + { + "epoch": 0.01017682228724081, + "ewc_loss": 7.963180541992188e-05, + "ewc_loss_diag": 1.0826624929904938e-08, + "ewc_loss_parallel": 7.869675755500793e-08, + "grad_norm": 7.413086891174316, + "learning_rate": 3.348876642645188e-08, + "loss": 0.6572, + "mean_token_accuracy": 0.8124239444732666, + "num_tokens": 3119180.0, + "step": 80 + }, + { + "epoch": 0.010304032565831319, + "ewc_loss": 8.153915405273438e-05, + "ewc_loss_diag": 1.1292286217212677e-08, + "ewc_loss_parallel": 8.055940270423889e-08, + "grad_norm": 8.66599178314209, + "learning_rate": 3.391267486222975e-08, + "loss": 0.749, + "mean_token_accuracy": 0.7868285179138184, + "num_tokens": 3154393.0, + "step": 81 + }, + { + "epoch": 0.01043124284442183, + "ewc_loss": 8.296966552734375e-05, + "ewc_loss_diag": 1.1641532182693481e-08, + "ewc_loss_parallel": 8.195638656616211e-08, + "grad_norm": 6.211172580718994, + "learning_rate": 3.4336583298007626e-08, + "loss": 0.7132, + "mean_token_accuracy": 0.7957860231399536, + "num_tokens": 3194560.0, + "step": 82 + }, + { + "epoch": 0.010558453123012339, + "ewc_loss": 8.106231689453125e-05, + "ewc_loss_diag": 1.1932570487260818e-08, + "ewc_loss_parallel": 8.009374141693115e-08, + "grad_norm": 8.81539249420166, + "learning_rate": 3.47604917337855e-08, + "loss": 0.7537, + "mean_token_accuracy": 0.7869526147842407, + "num_tokens": 3228718.0, + "step": 83 + }, + { + "epoch": 0.01068566340160285, + "ewc_loss": 8.106231689453125e-05, + "ewc_loss_diag": 1.2514647096395493e-08, + "ewc_loss_parallel": 7.962808012962341e-08, + "grad_norm": 7.600975513458252, + "learning_rate": 3.518440016956337e-08, + "loss": 0.7341, + "mean_token_accuracy": 0.7914618253707886, + "num_tokens": 3267480.0, + "step": 84 + }, + { + "epoch": 0.010812873680193359, + "ewc_loss": 7.963180541992188e-05, + "ewc_loss_diag": 1.3271346688270569e-08, + "ewc_loss_parallel": 7.82310962677002e-08, + "grad_norm": 6.4632344245910645, + "learning_rate": 3.5608308605341244e-08, + "loss": 0.7741, + "mean_token_accuracy": 0.7809546589851379, + "num_tokens": 3307376.0, + "step": 85 + }, + { + "epoch": 0.01094008395878387, + "ewc_loss": 7.867813110351562e-05, + "ewc_loss_diag": 1.4086253941059113e-08, + "ewc_loss_parallel": 7.729977369308472e-08, + "grad_norm": 8.02495002746582, + "learning_rate": 3.6032217041119116e-08, + "loss": 0.7592, + "mean_token_accuracy": 0.7850685119628906, + "num_tokens": 3350367.0, + "step": 86 + }, + { + "epoch": 0.01106729423737438, + "ewc_loss": 7.867813110351562e-05, + "ewc_loss_diag": 1.5133991837501526e-08, + "ewc_loss_parallel": 7.729977369308472e-08, + "grad_norm": 6.255082130432129, + "learning_rate": 3.645612547689699e-08, + "loss": 0.7352, + "mean_token_accuracy": 0.7948575615882874, + "num_tokens": 3392757.0, + "step": 87 + }, + { + "epoch": 0.01119450451596489, + "ewc_loss": 7.867813110351562e-05, + "ewc_loss_diag": 1.5599653124809265e-08, + "ewc_loss_parallel": 7.729977369308472e-08, + "grad_norm": 7.338225841522217, + "learning_rate": 3.6880033912674855e-08, + "loss": 0.7382, + "mean_token_accuracy": 0.7916621565818787, + "num_tokens": 3433563.0, + "step": 88 + }, + { + "epoch": 0.0113217147945554, + "ewc_loss": 7.867813110351562e-05, + "ewc_loss_diag": 1.6298145055770874e-08, + "ewc_loss_parallel": 7.729977369308472e-08, + "grad_norm": 6.910682201385498, + "learning_rate": 3.730394234845273e-08, + "loss": 0.7146, + "mean_token_accuracy": 0.8001234531402588, + "num_tokens": 3472300.0, + "step": 89 + }, + { + "epoch": 0.01144892507314591, + "ewc_loss": 7.963180541992188e-05, + "ewc_loss_diag": 1.6996636986732483e-08, + "ewc_loss_parallel": 7.776543498039246e-08, + "grad_norm": 6.401655673980713, + "learning_rate": 3.77278507842306e-08, + "loss": 0.791, + "mean_token_accuracy": 0.7758180499076843, + "num_tokens": 3509036.0, + "step": 90 + }, + { + "epoch": 0.01157613535173642, + "ewc_loss": 7.963180541992188e-05, + "ewc_loss_diag": 1.7462298274040222e-08, + "ewc_loss_parallel": 7.776543498039246e-08, + "grad_norm": 6.322074890136719, + "learning_rate": 3.815175922000847e-08, + "loss": 0.7551, + "mean_token_accuracy": 0.7893143892288208, + "num_tokens": 3548002.0, + "step": 91 + }, + { + "epoch": 0.01170334563032693, + "ewc_loss": 8.058547973632812e-05, + "ewc_loss_diag": 1.8044374883174896e-08, + "ewc_loss_parallel": 7.869675755500793e-08, + "grad_norm": 7.006691932678223, + "learning_rate": 3.8575667655786345e-08, + "loss": 0.721, + "mean_token_accuracy": 0.7972205877304077, + "num_tokens": 3585379.0, + "step": 92 + }, + { + "epoch": 0.01183055590891744, + "ewc_loss": 8.20159912109375e-05, + "ewc_loss_diag": 1.8742866814136505e-08, + "ewc_loss_parallel": 8.009374141693115e-08, + "grad_norm": 6.181943416595459, + "learning_rate": 3.899957609156422e-08, + "loss": 0.6868, + "mean_token_accuracy": 0.8057000637054443, + "num_tokens": 3627684.0, + "step": 93 + }, + { + "epoch": 0.01195776618750795, + "ewc_loss": 8.440017700195312e-05, + "ewc_loss_diag": 1.909211277961731e-08, + "ewc_loss_parallel": 8.242204785346985e-08, + "grad_norm": 6.182516574859619, + "learning_rate": 3.94234845273421e-08, + "loss": 0.7154, + "mean_token_accuracy": 0.797447919845581, + "num_tokens": 3667526.0, + "step": 94 + }, + { + "epoch": 0.012084976466098461, + "ewc_loss": 9.059906005859375e-05, + "ewc_loss_diag": 1.979060471057892e-08, + "ewc_loss_parallel": 8.847564458847046e-08, + "grad_norm": 6.619266033172607, + "learning_rate": 3.984739296311997e-08, + "loss": 0.7154, + "mean_token_accuracy": 0.795172929763794, + "num_tokens": 3703986.0, + "step": 95 + }, + { + "epoch": 0.01221218674468897, + "ewc_loss": 0.000102996826171875, + "ewc_loss_diag": 2.0721927285194397e-08, + "ewc_loss_parallel": 1.0104849934577942e-07, + "grad_norm": 6.8617329597473145, + "learning_rate": 4.027130139889784e-08, + "loss": 0.7527, + "mean_token_accuracy": 0.7879102826118469, + "num_tokens": 3745982.0, + "step": 96 + }, + { + "epoch": 0.012339397023279481, + "ewc_loss": 0.0001316070556640625, + "ewc_loss_diag": 2.2118911147117615e-08, + "ewc_loss_parallel": 1.2945383787155151e-07, + "grad_norm": 7.637326717376709, + "learning_rate": 4.0695209834675715e-08, + "loss": 0.7308, + "mean_token_accuracy": 0.795433759689331, + "num_tokens": 3788497.0, + "step": 97 + }, + { + "epoch": 0.012466607301869992, + "ewc_loss": 0.00018024444580078125, + "ewc_loss_diag": 2.3981556296348572e-08, + "ewc_loss_parallel": 1.7695128917694092e-07, + "grad_norm": 6.316046714782715, + "learning_rate": 4.111911827045358e-08, + "loss": 0.6781, + "mean_token_accuracy": 0.8114596605300903, + "num_tokens": 3829130.0, + "step": 98 + }, + { + "epoch": 0.012593817580460501, + "ewc_loss": 0.000225067138671875, + "ewc_loss_diag": 2.7474015951156616e-08, + "ewc_loss_parallel": 2.2258609533309937e-07, + "grad_norm": 7.5356926918029785, + "learning_rate": 4.154302670623145e-08, + "loss": 0.7004, + "mean_token_accuracy": 0.8023958206176758, + "num_tokens": 3868642.0, + "step": 99 + }, + { + "epoch": 0.012721027859051012, + "ewc_loss": 0.00026702880859375, + "ewc_loss_diag": 3.026798367500305e-08, + "ewc_loss_parallel": 2.6263296604156494e-07, + "grad_norm": 7.368246078491211, + "learning_rate": 4.1966935142009326e-08, + "loss": 0.7192, + "mean_token_accuracy": 0.794708251953125, + "num_tokens": 3902845.0, + "step": 100 + }, + { + "epoch": 0.012848238137641521, + "ewc_loss": 0.000286102294921875, + "ewc_loss_diag": 3.096647560596466e-08, + "ewc_loss_parallel": 2.8312206268310547e-07, + "grad_norm": 5.518518447875977, + "learning_rate": 4.23908435777872e-08, + "loss": 0.6917, + "mean_token_accuracy": 0.8050274848937988, + "num_tokens": 3945777.0, + "step": 101 + }, + { + "epoch": 0.012975448416232032, + "ewc_loss": 0.000293731689453125, + "ewc_loss_diag": 3.259629011154175e-08, + "ewc_loss_parallel": 2.905726432800293e-07, + "grad_norm": 6.091429233551025, + "learning_rate": 4.281475201356507e-08, + "loss": 0.7325, + "mean_token_accuracy": 0.7911089658737183, + "num_tokens": 3983626.0, + "step": 102 + }, + { + "epoch": 0.013102658694822541, + "ewc_loss": 0.0002899169921875, + "ewc_loss_diag": 3.4226104617118835e-08, + "ewc_loss_parallel": 2.868473529815674e-07, + "grad_norm": 7.009044170379639, + "learning_rate": 4.3238660449342943e-08, + "loss": 0.654, + "mean_token_accuracy": 0.8151296973228455, + "num_tokens": 4014959.0, + "step": 103 + }, + { + "epoch": 0.013229868973413052, + "ewc_loss": 0.00030517578125, + "ewc_loss_diag": 3.608874976634979e-08, + "ewc_loss_parallel": 3.0174851417541504e-07, + "grad_norm": 5.480094909667969, + "learning_rate": 4.3662568885120816e-08, + "loss": 0.6859, + "mean_token_accuracy": 0.8029963970184326, + "num_tokens": 4059052.0, + "step": 104 + }, + { + "epoch": 0.013357079252003561, + "ewc_loss": 0.00030517578125, + "ewc_loss_diag": 3.795139491558075e-08, + "ewc_loss_parallel": 3.0174851417541504e-07, + "grad_norm": 5.3130621910095215, + "learning_rate": 4.408647732089869e-08, + "loss": 0.6863, + "mean_token_accuracy": 0.8022798299789429, + "num_tokens": 4102524.0, + "step": 105 + }, + { + "epoch": 0.013484289530594072, + "ewc_loss": 0.0003108978271484375, + "ewc_loss_diag": 3.888271749019623e-08, + "ewc_loss_parallel": 3.073364496231079e-07, + "grad_norm": 5.769085884094238, + "learning_rate": 4.451038575667656e-08, + "loss": 0.7147, + "mean_token_accuracy": 0.7915486097335815, + "num_tokens": 4147427.0, + "step": 106 + }, + { + "epoch": 0.013611499809184581, + "ewc_loss": 0.00031280517578125, + "ewc_loss_diag": 4.0279701352119446e-08, + "ewc_loss_parallel": 3.0919909477233887e-07, + "grad_norm": 5.815685272216797, + "learning_rate": 4.493429419245443e-08, + "loss": 0.7352, + "mean_token_accuracy": 0.7900145053863525, + "num_tokens": 4185194.0, + "step": 107 + }, + { + "epoch": 0.013738710087775092, + "ewc_loss": 0.0003108978271484375, + "ewc_loss_diag": 4.190951585769653e-08, + "ewc_loss_parallel": 3.073364496231079e-07, + "grad_norm": 6.079276084899902, + "learning_rate": 4.53582026282323e-08, + "loss": 0.725, + "mean_token_accuracy": 0.7894178628921509, + "num_tokens": 4220839.0, + "step": 108 + }, + { + "epoch": 0.013865920366365603, + "ewc_loss": 0.00031280517578125, + "ewc_loss_diag": 4.400499165058136e-08, + "ewc_loss_parallel": 3.0919909477233887e-07, + "grad_norm": 6.060924053192139, + "learning_rate": 4.578211106401017e-08, + "loss": 0.6573, + "mean_token_accuracy": 0.8131396770477295, + "num_tokens": 4263441.0, + "step": 109 + }, + { + "epoch": 0.013993130644956112, + "ewc_loss": 0.0003108978271484375, + "ewc_loss_diag": 4.540197551250458e-08, + "ewc_loss_parallel": 3.073364496231079e-07, + "grad_norm": 6.071436405181885, + "learning_rate": 4.6206019499788045e-08, + "loss": 0.6985, + "mean_token_accuracy": 0.7971266508102417, + "num_tokens": 4304422.0, + "step": 110 + }, + { + "epoch": 0.014120340923546623, + "ewc_loss": 0.000308990478515625, + "ewc_loss_diag": 4.6333298087120056e-08, + "ewc_loss_parallel": 3.0547380447387695e-07, + "grad_norm": 5.434844017028809, + "learning_rate": 4.662992793556592e-08, + "loss": 0.6885, + "mean_token_accuracy": 0.8058072924613953, + "num_tokens": 4349590.0, + "step": 111 + }, + { + "epoch": 0.014247551202137132, + "ewc_loss": 0.0003070831298828125, + "ewc_loss_diag": 4.819594323635101e-08, + "ewc_loss_parallel": 3.0174851417541504e-07, + "grad_norm": 5.806923866271973, + "learning_rate": 4.705383637134379e-08, + "loss": 0.663, + "mean_token_accuracy": 0.809118390083313, + "num_tokens": 4387875.0, + "step": 112 + }, + { + "epoch": 0.014374761480727643, + "ewc_loss": 0.000301361083984375, + "ewc_loss_diag": 4.959292709827423e-08, + "ewc_loss_parallel": 2.9616057872772217e-07, + "grad_norm": 5.1780314445495605, + "learning_rate": 4.747774480712166e-08, + "loss": 0.6923, + "mean_token_accuracy": 0.800742506980896, + "num_tokens": 4430940.0, + "step": 113 + }, + { + "epoch": 0.014501971759318152, + "ewc_loss": 0.0002899169921875, + "ewc_loss_diag": 5.145557224750519e-08, + "ewc_loss_parallel": 2.849847078323364e-07, + "grad_norm": 5.077310085296631, + "learning_rate": 4.7901653242899535e-08, + "loss": 0.6385, + "mean_token_accuracy": 0.8158164024353027, + "num_tokens": 4470285.0, + "step": 114 + }, + { + "epoch": 0.014629182037908663, + "ewc_loss": 0.0002841949462890625, + "ewc_loss_diag": 5.3551048040390015e-08, + "ewc_loss_parallel": 2.7939677238464355e-07, + "grad_norm": 5.26384162902832, + "learning_rate": 4.832556167867741e-08, + "loss": 0.6889, + "mean_token_accuracy": 0.805054247379303, + "num_tokens": 4508503.0, + "step": 115 + }, + { + "epoch": 0.014756392316499172, + "ewc_loss": 0.000278472900390625, + "ewc_loss_diag": 5.448237061500549e-08, + "ewc_loss_parallel": 2.7194619178771973e-07, + "grad_norm": 5.490653991699219, + "learning_rate": 4.8749470114455274e-08, + "loss": 0.7123, + "mean_token_accuracy": 0.7945756912231445, + "num_tokens": 4546693.0, + "step": 116 + }, + { + "epoch": 0.014883602595089683, + "ewc_loss": 0.00026702880859375, + "ewc_loss_diag": 5.704350769519806e-08, + "ewc_loss_parallel": 2.60770320892334e-07, + "grad_norm": 5.093722820281982, + "learning_rate": 4.9173378550233146e-08, + "loss": 0.6567, + "mean_token_accuracy": 0.8085443377494812, + "num_tokens": 4585977.0, + "step": 117 + }, + { + "epoch": 0.015010812873680193, + "ewc_loss": 0.0002593994140625, + "ewc_loss_diag": 5.960464477539063e-08, + "ewc_loss_parallel": 2.5331974029541016e-07, + "grad_norm": 5.956377983093262, + "learning_rate": 4.959728698601102e-08, + "loss": 0.684, + "mean_token_accuracy": 0.8062458038330078, + "num_tokens": 4622905.0, + "step": 118 + }, + { + "epoch": 0.015138023152270703, + "ewc_loss": 0.0002574920654296875, + "ewc_loss_diag": 6.146728992462158e-08, + "ewc_loss_parallel": 2.514570951461792e-07, + "grad_norm": 5.3439226150512695, + "learning_rate": 5.002119542178889e-08, + "loss": 0.647, + "mean_token_accuracy": 0.8105518221855164, + "num_tokens": 4663616.0, + "step": 119 + }, + { + "epoch": 0.015265233430861214, + "ewc_loss": 0.0002574920654296875, + "ewc_loss_diag": 6.239861249923706e-08, + "ewc_loss_parallel": 2.514570951461792e-07, + "grad_norm": 5.316130638122559, + "learning_rate": 5.0445103857566764e-08, + "loss": 0.6927, + "mean_token_accuracy": 0.8005872368812561, + "num_tokens": 4703879.0, + "step": 120 + }, + { + "epoch": 0.015392443709451724, + "ewc_loss": 0.0002536773681640625, + "ewc_loss_diag": 6.379559636116028e-08, + "ewc_loss_parallel": 2.477318048477173e-07, + "grad_norm": 4.4908952713012695, + "learning_rate": 5.0869012293344637e-08, + "loss": 0.5987, + "mean_token_accuracy": 0.8250181674957275, + "num_tokens": 4749923.0, + "step": 121 + }, + { + "epoch": 0.015519653988042234, + "ewc_loss": 0.0002498626708984375, + "ewc_loss_diag": 6.612390279769897e-08, + "ewc_loss_parallel": 2.4400651454925537e-07, + "grad_norm": 4.9357709884643555, + "learning_rate": 5.129292072912251e-08, + "loss": 0.6676, + "mean_token_accuracy": 0.8084475994110107, + "num_tokens": 4796025.0, + "step": 122 + }, + { + "epoch": 0.015646864266632744, + "ewc_loss": 0.000247955322265625, + "ewc_loss_diag": 6.845220923423767e-08, + "ewc_loss_parallel": 2.4028122425079346e-07, + "grad_norm": 5.060362815856934, + "learning_rate": 5.171682916490038e-08, + "loss": 0.6832, + "mean_token_accuracy": 0.8006855249404907, + "num_tokens": 4836394.0, + "step": 123 + }, + { + "epoch": 0.015774074545223254, + "ewc_loss": 0.0002460479736328125, + "ewc_loss_diag": 7.031485438346863e-08, + "ewc_loss_parallel": 2.384185791015625e-07, + "grad_norm": 5.050446510314941, + "learning_rate": 5.2140737600678254e-08, + "loss": 0.7108, + "mean_token_accuracy": 0.7916163802146912, + "num_tokens": 4875897.0, + "step": 124 + }, + { + "epoch": 0.015901284823813765, + "ewc_loss": 0.00024318695068359375, + "ewc_loss_diag": 7.171183824539185e-08, + "ewc_loss_parallel": 2.3562461137771606e-07, + "grad_norm": 5.5164289474487305, + "learning_rate": 5.256464603645612e-08, + "loss": 0.6545, + "mean_token_accuracy": 0.8094825148582458, + "num_tokens": 4912603.0, + "step": 125 + }, + { + "epoch": 0.016028495102404273, + "ewc_loss": 0.00024318695068359375, + "ewc_loss_diag": 7.35744833946228e-08, + "ewc_loss_parallel": 2.3562461137771606e-07, + "grad_norm": 5.421999454498291, + "learning_rate": 5.298855447223399e-08, + "loss": 0.7184, + "mean_token_accuracy": 0.7927770614624023, + "num_tokens": 4955248.0, + "step": 126 + }, + { + "epoch": 0.016155705380994784, + "ewc_loss": 0.000244140625, + "ewc_loss_diag": 7.497146725654602e-08, + "ewc_loss_parallel": 2.3748725652694702e-07, + "grad_norm": 5.365941047668457, + "learning_rate": 5.3412462908011865e-08, + "loss": 0.6786, + "mean_token_accuracy": 0.8061628341674805, + "num_tokens": 4993414.0, + "step": 127 + }, + { + "epoch": 0.016282915659585295, + "ewc_loss": 0.0002498626708984375, + "ewc_loss_diag": 7.729977369308472e-08, + "ewc_loss_parallel": 2.421438694000244e-07, + "grad_norm": 5.0275397300720215, + "learning_rate": 5.383637134378974e-08, + "loss": 0.7019, + "mean_token_accuracy": 0.7953125238418579, + "num_tokens": 5028819.0, + "step": 128 + }, + { + "epoch": 0.016410125938175806, + "ewc_loss": 0.00025177001953125, + "ewc_loss_diag": 8.055940270423889e-08, + "ewc_loss_parallel": 2.4400651454925537e-07, + "grad_norm": 4.935311794281006, + "learning_rate": 5.426027977956761e-08, + "loss": 0.6747, + "mean_token_accuracy": 0.8058720827102661, + "num_tokens": 5067874.0, + "step": 129 + }, + { + "epoch": 0.016537336216766316, + "ewc_loss": 0.0002593994140625, + "ewc_loss_diag": 8.475035429000854e-08, + "ewc_loss_parallel": 2.514570951461792e-07, + "grad_norm": 4.802576065063477, + "learning_rate": 5.468418821534548e-08, + "loss": 0.6798, + "mean_token_accuracy": 0.8033565878868103, + "num_tokens": 5104177.0, + "step": 130 + }, + { + "epoch": 0.016664546495356824, + "ewc_loss": 0.0002841949462890625, + "ewc_loss_diag": 8.800998330116272e-08, + "ewc_loss_parallel": 2.738088369369507e-07, + "grad_norm": 5.2881760597229, + "learning_rate": 5.5108096651123356e-08, + "loss": 0.6605, + "mean_token_accuracy": 0.8073439598083496, + "num_tokens": 5143352.0, + "step": 131 + }, + { + "epoch": 0.016791756773947335, + "ewc_loss": 0.0003032684326171875, + "ewc_loss_diag": 9.173527359962463e-08, + "ewc_loss_parallel": 2.942979335784912e-07, + "grad_norm": 5.0672760009765625, + "learning_rate": 5.553200508690123e-08, + "loss": 0.6751, + "mean_token_accuracy": 0.801376461982727, + "num_tokens": 5185182.0, + "step": 132 + }, + { + "epoch": 0.016918967052537846, + "ewc_loss": 0.0003414154052734375, + "ewc_loss_diag": 9.313225746154785e-08, + "ewc_loss_parallel": 3.3155083656311035e-07, + "grad_norm": 5.387881278991699, + "learning_rate": 5.59559135226791e-08, + "loss": 0.7163, + "mean_token_accuracy": 0.7964396476745605, + "num_tokens": 5217960.0, + "step": 133 + }, + { + "epoch": 0.017046177331128357, + "ewc_loss": 0.0003910064697265625, + "ewc_loss_diag": 9.639188647270203e-08, + "ewc_loss_parallel": 3.818422555923462e-07, + "grad_norm": 5.100269794464111, + "learning_rate": 5.637982195845697e-08, + "loss": 0.6707, + "mean_token_accuracy": 0.8054680824279785, + "num_tokens": 5255907.0, + "step": 134 + }, + { + "epoch": 0.017173387609718864, + "ewc_loss": 0.0004482269287109375, + "ewc_loss_diag": 1.0104849934577942e-07, + "ewc_loss_parallel": 4.3958425521850586e-07, + "grad_norm": 4.9148688316345215, + "learning_rate": 5.680373039423484e-08, + "loss": 0.6993, + "mean_token_accuracy": 0.8023097515106201, + "num_tokens": 5293792.0, + "step": 135 + }, + { + "epoch": 0.017300597888309375, + "ewc_loss": 0.000507354736328125, + "ewc_loss_diag": 1.0337680578231812e-07, + "ewc_loss_parallel": 4.954636096954346e-07, + "grad_norm": 5.0265302658081055, + "learning_rate": 5.722763883001271e-08, + "loss": 0.6569, + "mean_token_accuracy": 0.8070249557495117, + "num_tokens": 5332426.0, + "step": 136 + }, + { + "epoch": 0.017427808166899886, + "ewc_loss": 0.000553131103515625, + "ewc_loss_diag": 1.103617250919342e-07, + "ewc_loss_parallel": 5.401670932769775e-07, + "grad_norm": 5.3372368812561035, + "learning_rate": 5.7651547265790585e-08, + "loss": 0.7145, + "mean_token_accuracy": 0.7965527176856995, + "num_tokens": 5366857.0, + "step": 137 + }, + { + "epoch": 0.017555018445490397, + "ewc_loss": 0.000614166259765625, + "ewc_loss_diag": 1.1315569281578064e-07, + "ewc_loss_parallel": 6.034970283508301e-07, + "grad_norm": 4.94066047668457, + "learning_rate": 5.807545570156846e-08, + "loss": 0.6855, + "mean_token_accuracy": 0.8028630018234253, + "num_tokens": 5404003.0, + "step": 138 + }, + { + "epoch": 0.017682228724080904, + "ewc_loss": 0.000659942626953125, + "ewc_loss_diag": 1.1920928955078125e-07, + "ewc_loss_parallel": 6.48200511932373e-07, + "grad_norm": 4.728944301605225, + "learning_rate": 5.849936413734633e-08, + "loss": 0.6312, + "mean_token_accuracy": 0.8151828050613403, + "num_tokens": 5440025.0, + "step": 139 + }, + { + "epoch": 0.017809439002671415, + "ewc_loss": 0.000698089599609375, + "ewc_loss_diag": 1.2479722499847412e-07, + "ewc_loss_parallel": 6.854534149169922e-07, + "grad_norm": 4.547050952911377, + "learning_rate": 5.89232725731242e-08, + "loss": 0.688, + "mean_token_accuracy": 0.8034219741821289, + "num_tokens": 5479167.0, + "step": 140 + }, + { + "epoch": 0.017936649281261926, + "ewc_loss": 0.000713348388671875, + "ewc_loss_diag": 1.2759119272232056e-07, + "ewc_loss_parallel": 7.003545761108398e-07, + "grad_norm": 4.442161560058594, + "learning_rate": 5.9347181008902075e-08, + "loss": 0.6394, + "mean_token_accuracy": 0.8121866583824158, + "num_tokens": 5517958.0, + "step": 141 + }, + { + "epoch": 0.018063859559852437, + "ewc_loss": 0.000732421875, + "ewc_loss_diag": 1.3131648302078247e-07, + "ewc_loss_parallel": 7.227063179016113e-07, + "grad_norm": 4.195199966430664, + "learning_rate": 5.977108944467995e-08, + "loss": 0.6229, + "mean_token_accuracy": 0.8179011940956116, + "num_tokens": 5562683.0, + "step": 142 + }, + { + "epoch": 0.018191069838442948, + "ewc_loss": 0.0007476806640625, + "ewc_loss_diag": 1.3504177331924438e-07, + "ewc_loss_parallel": 7.338821887969971e-07, + "grad_norm": 4.497960567474365, + "learning_rate": 6.019499788045781e-08, + "loss": 0.6247, + "mean_token_accuracy": 0.8159642219543457, + "num_tokens": 5601178.0, + "step": 143 + }, + { + "epoch": 0.018318280117033455, + "ewc_loss": 0.00077056884765625, + "ewc_loss_diag": 1.387670636177063e-07, + "ewc_loss_parallel": 7.562339305877686e-07, + "grad_norm": 4.615710258483887, + "learning_rate": 6.061890631623569e-08, + "loss": 0.6486, + "mean_token_accuracy": 0.8074823021888733, + "num_tokens": 5637230.0, + "step": 144 + }, + { + "epoch": 0.018445490395623966, + "ewc_loss": 0.000782012939453125, + "ewc_loss_diag": 1.434236764907837e-07, + "ewc_loss_parallel": 7.674098014831543e-07, + "grad_norm": 4.280570030212402, + "learning_rate": 6.104281475201356e-08, + "loss": 0.6313, + "mean_token_accuracy": 0.8142982721328735, + "num_tokens": 5675394.0, + "step": 145 + }, + { + "epoch": 0.018572700674214477, + "ewc_loss": 0.00078582763671875, + "ewc_loss_diag": 1.471489667892456e-07, + "ewc_loss_parallel": 7.711350917816162e-07, + "grad_norm": 4.416815757751465, + "learning_rate": 6.146672318779143e-08, + "loss": 0.6651, + "mean_token_accuracy": 0.8034185171127319, + "num_tokens": 5712426.0, + "step": 146 + }, + { + "epoch": 0.018699910952804988, + "ewc_loss": 0.000797271728515625, + "ewc_loss_diag": 1.5087425708770752e-07, + "ewc_loss_parallel": 7.82310962677002e-07, + "grad_norm": 4.090319633483887, + "learning_rate": 6.18906316235693e-08, + "loss": 0.6425, + "mean_token_accuracy": 0.8095189332962036, + "num_tokens": 5757609.0, + "step": 147 + }, + { + "epoch": 0.018827121231395495, + "ewc_loss": 0.000789642333984375, + "ewc_loss_diag": 1.5459954738616943e-07, + "ewc_loss_parallel": 7.748603820800781e-07, + "grad_norm": 4.430126190185547, + "learning_rate": 6.231454005934718e-08, + "loss": 0.6057, + "mean_token_accuracy": 0.82387375831604, + "num_tokens": 5795712.0, + "step": 148 + }, + { + "epoch": 0.018954331509986006, + "ewc_loss": 0.00078582763671875, + "ewc_loss_diag": 1.6111880540847778e-07, + "ewc_loss_parallel": 7.711350917816162e-07, + "grad_norm": 4.365373611450195, + "learning_rate": 6.273844849512505e-08, + "loss": 0.632, + "mean_token_accuracy": 0.8120307922363281, + "num_tokens": 5832052.0, + "step": 149 + }, + { + "epoch": 0.019081541788576517, + "ewc_loss": 0.000782012939453125, + "ewc_loss_diag": 1.6670674085617065e-07, + "ewc_loss_parallel": 7.674098014831543e-07, + "grad_norm": 4.343483924865723, + "learning_rate": 6.316235693090292e-08, + "loss": 0.641, + "mean_token_accuracy": 0.8116627335548401, + "num_tokens": 5870058.0, + "step": 150 + }, + { + "epoch": 0.019208752067167028, + "ewc_loss": 0.0007781982421875, + "ewc_loss_diag": 1.7136335372924805e-07, + "ewc_loss_parallel": 7.599592208862305e-07, + "grad_norm": 4.210840702056885, + "learning_rate": 6.35862653666808e-08, + "loss": 0.6372, + "mean_token_accuracy": 0.8080683946609497, + "num_tokens": 5911423.0, + "step": 151 + }, + { + "epoch": 0.01933596234575754, + "ewc_loss": 0.00077056884765625, + "ewc_loss_diag": 1.7695128917694092e-07, + "ewc_loss_parallel": 7.525086402893066e-07, + "grad_norm": 4.570464134216309, + "learning_rate": 6.401017380245867e-08, + "loss": 0.6491, + "mean_token_accuracy": 0.8089238405227661, + "num_tokens": 5951418.0, + "step": 152 + }, + { + "epoch": 0.019463172624348046, + "ewc_loss": 0.000762939453125, + "ewc_loss_diag": 1.816079020500183e-07, + "ewc_loss_parallel": 7.450580596923828e-07, + "grad_norm": 4.136258602142334, + "learning_rate": 6.443408223823654e-08, + "loss": 0.6619, + "mean_token_accuracy": 0.8033338785171509, + "num_tokens": 5992390.0, + "step": 153 + }, + { + "epoch": 0.019590382902938557, + "ewc_loss": 0.000743865966796875, + "ewc_loss_diag": 1.8533319234848022e-07, + "ewc_loss_parallel": 7.264316082000732e-07, + "grad_norm": 4.658485412597656, + "learning_rate": 6.485799067401441e-08, + "loss": 0.7252, + "mean_token_accuracy": 0.7874494791030884, + "num_tokens": 6035396.0, + "step": 154 + }, + { + "epoch": 0.019717593181529068, + "ewc_loss": 0.000743865966796875, + "ewc_loss_diag": 1.8812716007232666e-07, + "ewc_loss_parallel": 7.264316082000732e-07, + "grad_norm": 4.133712291717529, + "learning_rate": 6.528189910979228e-08, + "loss": 0.6002, + "mean_token_accuracy": 0.8183189630508423, + "num_tokens": 6071581.0, + "step": 155 + }, + { + "epoch": 0.01984480346011958, + "ewc_loss": 0.0007171630859375, + "ewc_loss_diag": 1.8998980522155762e-07, + "ewc_loss_parallel": 6.966292858123779e-07, + "grad_norm": 4.2029290199279785, + "learning_rate": 6.570580754557016e-08, + "loss": 0.6196, + "mean_token_accuracy": 0.8126417398452759, + "num_tokens": 6108767.0, + "step": 156 + }, + { + "epoch": 0.019972013738710086, + "ewc_loss": 0.000698089599609375, + "ewc_loss_diag": 1.9185245037078857e-07, + "ewc_loss_parallel": 6.780028343200684e-07, + "grad_norm": 4.035506248474121, + "learning_rate": 6.612971598134802e-08, + "loss": 0.5676, + "mean_token_accuracy": 0.8294918537139893, + "num_tokens": 6147181.0, + "step": 157 + }, + { + "epoch": 0.020099224017300597, + "ewc_loss": 0.00067138671875, + "ewc_loss_diag": 1.94646418094635e-07, + "ewc_loss_parallel": 6.51925802230835e-07, + "grad_norm": 4.145239353179932, + "learning_rate": 6.655362441712589e-08, + "loss": 0.6115, + "mean_token_accuracy": 0.8180578351020813, + "num_tokens": 6185325.0, + "step": 158 + }, + { + "epoch": 0.020226434295891108, + "ewc_loss": 0.00066375732421875, + "ewc_loss_diag": 2.0116567611694336e-07, + "ewc_loss_parallel": 6.444752216339111e-07, + "grad_norm": 4.130581855773926, + "learning_rate": 6.697753285290376e-08, + "loss": 0.6705, + "mean_token_accuracy": 0.800510823726654, + "num_tokens": 6224080.0, + "step": 159 + }, + { + "epoch": 0.02035364457448162, + "ewc_loss": 0.000637054443359375, + "ewc_loss_diag": 2.0675361156463623e-07, + "ewc_loss_parallel": 6.183981895446777e-07, + "grad_norm": 4.095557689666748, + "learning_rate": 6.740144128868163e-08, + "loss": 0.6268, + "mean_token_accuracy": 0.8141858577728271, + "num_tokens": 6261152.0, + "step": 160 + }, + { + "epoch": 0.020480854853072127, + "ewc_loss": 0.000629425048828125, + "ewc_loss_diag": 2.1047890186309814e-07, + "ewc_loss_parallel": 6.07222318649292e-07, + "grad_norm": 3.7472217082977295, + "learning_rate": 6.78253497244595e-08, + "loss": 0.6144, + "mean_token_accuracy": 0.8167408108711243, + "num_tokens": 6305128.0, + "step": 161 + }, + { + "epoch": 0.020608065131662637, + "ewc_loss": 0.0006103515625, + "ewc_loss_diag": 2.123415470123291e-07, + "ewc_loss_parallel": 5.885958671569824e-07, + "grad_norm": 4.020700931549072, + "learning_rate": 6.824925816023738e-08, + "loss": 0.6173, + "mean_token_accuracy": 0.8161861896514893, + "num_tokens": 6341563.0, + "step": 162 + }, + { + "epoch": 0.02073527541025315, + "ewc_loss": 0.000598907470703125, + "ewc_loss_diag": 2.1606683731079102e-07, + "ewc_loss_parallel": 5.774199962615967e-07, + "grad_norm": 4.396273612976074, + "learning_rate": 6.867316659601525e-08, + "loss": 0.6517, + "mean_token_accuracy": 0.809751033782959, + "num_tokens": 6371766.0, + "step": 163 + }, + { + "epoch": 0.02086248568884366, + "ewc_loss": 0.000598907470703125, + "ewc_loss_diag": 2.1979212760925293e-07, + "ewc_loss_parallel": 5.774199962615967e-07, + "grad_norm": 3.731229305267334, + "learning_rate": 6.909707503179312e-08, + "loss": 0.6004, + "mean_token_accuracy": 0.8215018510818481, + "num_tokens": 6414331.0, + "step": 164 + }, + { + "epoch": 0.02098969596743417, + "ewc_loss": 0.00058746337890625, + "ewc_loss_diag": 2.2165477275848389e-07, + "ewc_loss_parallel": 5.662441253662109e-07, + "grad_norm": 3.931408405303955, + "learning_rate": 6.9520983467571e-08, + "loss": 0.6619, + "mean_token_accuracy": 0.8037334084510803, + "num_tokens": 6455066.0, + "step": 165 + }, + { + "epoch": 0.021116906246024678, + "ewc_loss": 0.000579833984375, + "ewc_loss_diag": 2.253800630569458e-07, + "ewc_loss_parallel": 5.550682544708252e-07, + "grad_norm": 3.876210927963257, + "learning_rate": 6.994489190334887e-08, + "loss": 0.5853, + "mean_token_accuracy": 0.8240945339202881, + "num_tokens": 6492300.0, + "step": 166 + }, + { + "epoch": 0.02124411652461519, + "ewc_loss": 0.000568389892578125, + "ewc_loss_diag": 2.2724270820617676e-07, + "ewc_loss_parallel": 5.438923835754395e-07, + "grad_norm": 3.7603464126586914, + "learning_rate": 7.036880033912674e-08, + "loss": 0.593, + "mean_token_accuracy": 0.820956826210022, + "num_tokens": 6531548.0, + "step": 167 + }, + { + "epoch": 0.0213713268032057, + "ewc_loss": 0.000560760498046875, + "ewc_loss_diag": 2.3096799850463867e-07, + "ewc_loss_parallel": 5.364418029785156e-07, + "grad_norm": 3.8887054920196533, + "learning_rate": 7.079270877490461e-08, + "loss": 0.6189, + "mean_token_accuracy": 0.8147139549255371, + "num_tokens": 6570494.0, + "step": 168 + }, + { + "epoch": 0.02149853708179621, + "ewc_loss": 0.00055694580078125, + "ewc_loss_diag": 2.3748725652694702e-07, + "ewc_loss_parallel": 5.327165126800537e-07, + "grad_norm": 4.017823219299316, + "learning_rate": 7.121661721068249e-08, + "loss": 0.6018, + "mean_token_accuracy": 0.8210353851318359, + "num_tokens": 6602208.0, + "step": 169 + }, + { + "epoch": 0.021625747360386718, + "ewc_loss": 0.00055694580078125, + "ewc_loss_diag": 2.4028122425079346e-07, + "ewc_loss_parallel": 5.327165126800537e-07, + "grad_norm": 3.9107630252838135, + "learning_rate": 7.164052564646036e-08, + "loss": 0.5964, + "mean_token_accuracy": 0.82213294506073, + "num_tokens": 6639592.0, + "step": 170 + }, + { + "epoch": 0.02175295763897723, + "ewc_loss": 0.00055694580078125, + "ewc_loss_diag": 2.4400651454925537e-07, + "ewc_loss_parallel": 5.327165126800537e-07, + "grad_norm": 3.977848768234253, + "learning_rate": 7.206443408223823e-08, + "loss": 0.6206, + "mean_token_accuracy": 0.8156943917274475, + "num_tokens": 6675730.0, + "step": 171 + }, + { + "epoch": 0.02188016791756774, + "ewc_loss": 0.00055694580078125, + "ewc_loss_diag": 2.4959444999694824e-07, + "ewc_loss_parallel": 5.289912223815918e-07, + "grad_norm": 3.854173183441162, + "learning_rate": 7.24883425180161e-08, + "loss": 0.6013, + "mean_token_accuracy": 0.814393937587738, + "num_tokens": 6714336.0, + "step": 172 + }, + { + "epoch": 0.02200737819615825, + "ewc_loss": 0.000560760498046875, + "ewc_loss_diag": 2.551823854446411e-07, + "ewc_loss_parallel": 5.327165126800537e-07, + "grad_norm": 3.691054582595825, + "learning_rate": 7.291225095379398e-08, + "loss": 0.5742, + "mean_token_accuracy": 0.8273074626922607, + "num_tokens": 6757046.0, + "step": 173 + }, + { + "epoch": 0.02213458847474876, + "ewc_loss": 0.00054168701171875, + "ewc_loss_diag": 2.5704503059387207e-07, + "ewc_loss_parallel": 5.140900611877441e-07, + "grad_norm": 3.794858932495117, + "learning_rate": 7.333615938957185e-08, + "loss": 0.6529, + "mean_token_accuracy": 0.8054659366607666, + "num_tokens": 6798928.0, + "step": 174 + }, + { + "epoch": 0.02226179875333927, + "ewc_loss": 0.00054168701171875, + "ewc_loss_diag": 2.5890767574310303e-07, + "ewc_loss_parallel": 5.140900611877441e-07, + "grad_norm": 3.8488128185272217, + "learning_rate": 7.376006782534971e-08, + "loss": 0.5671, + "mean_token_accuracy": 0.8295577764511108, + "num_tokens": 6832561.0, + "step": 175 + }, + { + "epoch": 0.02238900903192978, + "ewc_loss": 0.000560760498046875, + "ewc_loss_diag": 2.60770320892334e-07, + "ewc_loss_parallel": 5.327165126800537e-07, + "grad_norm": 3.65376353263855, + "learning_rate": 7.418397626112758e-08, + "loss": 0.5964, + "mean_token_accuracy": 0.8190698623657227, + "num_tokens": 6875783.0, + "step": 176 + }, + { + "epoch": 0.02251621931052029, + "ewc_loss": 0.00055694580078125, + "ewc_loss_diag": 2.6263296604156494e-07, + "ewc_loss_parallel": 5.289912223815918e-07, + "grad_norm": 3.817045211791992, + "learning_rate": 7.460788469690545e-08, + "loss": 0.601, + "mean_token_accuracy": 0.8209283351898193, + "num_tokens": 6912664.0, + "step": 177 + }, + { + "epoch": 0.0226434295891108, + "ewc_loss": 0.000560760498046875, + "ewc_loss_diag": 2.6635825634002686e-07, + "ewc_loss_parallel": 5.327165126800537e-07, + "grad_norm": 3.9521193504333496, + "learning_rate": 7.503179313268333e-08, + "loss": 0.6334, + "mean_token_accuracy": 0.8093574047088623, + "num_tokens": 6946125.0, + "step": 178 + }, + { + "epoch": 0.02277063986770131, + "ewc_loss": 0.000579833984375, + "ewc_loss_diag": 2.7567148208618164e-07, + "ewc_loss_parallel": 5.513429641723633e-07, + "grad_norm": 3.6528656482696533, + "learning_rate": 7.54557015684612e-08, + "loss": 0.5688, + "mean_token_accuracy": 0.8252483606338501, + "num_tokens": 6985450.0, + "step": 179 + }, + { + "epoch": 0.02289785014629182, + "ewc_loss": 0.000583648681640625, + "ewc_loss_diag": 2.7939677238464355e-07, + "ewc_loss_parallel": 5.587935447692871e-07, + "grad_norm": 3.6292150020599365, + "learning_rate": 7.587961000423907e-08, + "loss": 0.5955, + "mean_token_accuracy": 0.8216966986656189, + "num_tokens": 7024527.0, + "step": 180 + }, + { + "epoch": 0.02302506042488233, + "ewc_loss": 0.00058746337890625, + "ewc_loss_diag": 2.8312206268310547e-07, + "ewc_loss_parallel": 5.62518835067749e-07, + "grad_norm": 3.8261780738830566, + "learning_rate": 7.630351844001694e-08, + "loss": 0.6983, + "mean_token_accuracy": 0.7883652448654175, + "num_tokens": 7071851.0, + "step": 181 + }, + { + "epoch": 0.02315227070347284, + "ewc_loss": 0.000606536865234375, + "ewc_loss_diag": 2.849847078323364e-07, + "ewc_loss_parallel": 5.811452865600586e-07, + "grad_norm": 3.664783239364624, + "learning_rate": 7.672742687579482e-08, + "loss": 0.5926, + "mean_token_accuracy": 0.819928765296936, + "num_tokens": 7111426.0, + "step": 182 + }, + { + "epoch": 0.02327948098206335, + "ewc_loss": 0.0006256103515625, + "ewc_loss_diag": 2.942979335784912e-07, + "ewc_loss_parallel": 5.960464477539062e-07, + "grad_norm": 3.8843953609466553, + "learning_rate": 7.715133531157269e-08, + "loss": 0.5835, + "mean_token_accuracy": 0.8213780522346497, + "num_tokens": 7145966.0, + "step": 183 + }, + { + "epoch": 0.02340669126065386, + "ewc_loss": 0.000637054443359375, + "ewc_loss_diag": 2.998858690261841e-07, + "ewc_loss_parallel": 6.07222318649292e-07, + "grad_norm": 3.6017045974731445, + "learning_rate": 7.757524374735056e-08, + "loss": 0.6378, + "mean_token_accuracy": 0.8053148984909058, + "num_tokens": 7188702.0, + "step": 184 + }, + { + "epoch": 0.02353390153924437, + "ewc_loss": 0.0006561279296875, + "ewc_loss_diag": 3.0547380447387695e-07, + "ewc_loss_parallel": 6.258487701416016e-07, + "grad_norm": 3.5765280723571777, + "learning_rate": 7.799915218312844e-08, + "loss": 0.584, + "mean_token_accuracy": 0.8220288753509521, + "num_tokens": 7227937.0, + "step": 185 + }, + { + "epoch": 0.02366111181783488, + "ewc_loss": 0.000659942626953125, + "ewc_loss_diag": 3.1478703022003174e-07, + "ewc_loss_parallel": 6.295740604400635e-07, + "grad_norm": 3.677997350692749, + "learning_rate": 7.842306061890631e-08, + "loss": 0.5588, + "mean_token_accuracy": 0.8245218396186829, + "num_tokens": 7267600.0, + "step": 186 + }, + { + "epoch": 0.023788322096425393, + "ewc_loss": 0.000682830810546875, + "ewc_loss_diag": 3.2223761081695557e-07, + "ewc_loss_parallel": 6.51925802230835e-07, + "grad_norm": 3.6394410133361816, + "learning_rate": 7.88469690546842e-08, + "loss": 0.6, + "mean_token_accuracy": 0.8168330192565918, + "num_tokens": 7308449.0, + "step": 187 + }, + { + "epoch": 0.0239155323750159, + "ewc_loss": 0.000698089599609375, + "ewc_loss_diag": 3.334134817123413e-07, + "ewc_loss_parallel": 6.631016731262207e-07, + "grad_norm": 3.7631654739379883, + "learning_rate": 7.927087749046207e-08, + "loss": 0.593, + "mean_token_accuracy": 0.8219620585441589, + "num_tokens": 7342950.0, + "step": 188 + }, + { + "epoch": 0.02404274265360641, + "ewc_loss": 0.0007171630859375, + "ewc_loss_diag": 3.427267074584961e-07, + "ewc_loss_parallel": 6.817281246185303e-07, + "grad_norm": 3.7027101516723633, + "learning_rate": 7.969478592623994e-08, + "loss": 0.6422, + "mean_token_accuracy": 0.8035424947738647, + "num_tokens": 7384573.0, + "step": 189 + }, + { + "epoch": 0.024169952932196922, + "ewc_loss": 0.000732421875, + "ewc_loss_diag": 3.46451997756958e-07, + "ewc_loss_parallel": 6.966292858123779e-07, + "grad_norm": 3.416511058807373, + "learning_rate": 8.011869436201781e-08, + "loss": 0.5105, + "mean_token_accuracy": 0.844098687171936, + "num_tokens": 7423915.0, + "step": 190 + }, + { + "epoch": 0.024297163210787433, + "ewc_loss": 0.000736236572265625, + "ewc_loss_diag": 3.557652235031128e-07, + "ewc_loss_parallel": 7.003545761108398e-07, + "grad_norm": 3.7188849449157715, + "learning_rate": 8.054260279779568e-08, + "loss": 0.6225, + "mean_token_accuracy": 0.8132581114768982, + "num_tokens": 7461529.0, + "step": 191 + }, + { + "epoch": 0.02442437348937794, + "ewc_loss": 0.000751495361328125, + "ewc_loss_diag": 3.594905138015747e-07, + "ewc_loss_parallel": 7.152557373046875e-07, + "grad_norm": 3.75697660446167, + "learning_rate": 8.096651123357356e-08, + "loss": 0.6235, + "mean_token_accuracy": 0.8084307312965393, + "num_tokens": 7499671.0, + "step": 192 + }, + { + "epoch": 0.02455158376796845, + "ewc_loss": 0.00078582763671875, + "ewc_loss_diag": 3.688037395477295e-07, + "ewc_loss_parallel": 7.487833499908447e-07, + "grad_norm": 3.652085542678833, + "learning_rate": 8.139041966935143e-08, + "loss": 0.616, + "mean_token_accuracy": 0.811649739742279, + "num_tokens": 7543972.0, + "step": 193 + }, + { + "epoch": 0.024678794046558962, + "ewc_loss": 0.00080108642578125, + "ewc_loss_diag": 3.725290298461914e-07, + "ewc_loss_parallel": 7.636845111846924e-07, + "grad_norm": 3.49947452545166, + "learning_rate": 8.181432810512929e-08, + "loss": 0.5546, + "mean_token_accuracy": 0.8278853893280029, + "num_tokens": 7585156.0, + "step": 194 + }, + { + "epoch": 0.024806004325149473, + "ewc_loss": 0.000804901123046875, + "ewc_loss_diag": 3.7439167499542236e-07, + "ewc_loss_parallel": 7.674098014831543e-07, + "grad_norm": 3.3909101486206055, + "learning_rate": 8.223823654090716e-08, + "loss": 0.5481, + "mean_token_accuracy": 0.8338207602500916, + "num_tokens": 7629499.0, + "step": 195 + }, + { + "epoch": 0.024933214603739984, + "ewc_loss": 0.00080108642578125, + "ewc_loss_diag": 3.818422555923462e-07, + "ewc_loss_parallel": 7.636845111846924e-07, + "grad_norm": 3.3062822818756104, + "learning_rate": 8.266214497668503e-08, + "loss": 0.5209, + "mean_token_accuracy": 0.8407866954803467, + "num_tokens": 7675226.0, + "step": 196 + }, + { + "epoch": 0.02506042488233049, + "ewc_loss": 0.000812530517578125, + "ewc_loss_diag": 3.8929283618927e-07, + "ewc_loss_parallel": 7.748603820800781e-07, + "grad_norm": 3.610539436340332, + "learning_rate": 8.30860534124629e-08, + "loss": 0.5828, + "mean_token_accuracy": 0.8220289349555969, + "num_tokens": 7713173.0, + "step": 197 + }, + { + "epoch": 0.025187635160921002, + "ewc_loss": 0.00083160400390625, + "ewc_loss_diag": 3.948807716369629e-07, + "ewc_loss_parallel": 7.934868335723877e-07, + "grad_norm": 3.7254645824432373, + "learning_rate": 8.350996184824078e-08, + "loss": 0.5637, + "mean_token_accuracy": 0.8258283138275146, + "num_tokens": 7745434.0, + "step": 198 + }, + { + "epoch": 0.025314845439511513, + "ewc_loss": 0.000843048095703125, + "ewc_loss_diag": 4.023313522338867e-07, + "ewc_loss_parallel": 8.009374141693115e-07, + "grad_norm": 3.5273585319519043, + "learning_rate": 8.393387028401865e-08, + "loss": 0.5658, + "mean_token_accuracy": 0.826361894607544, + "num_tokens": 7783416.0, + "step": 199 + }, + { + "epoch": 0.025442055718102024, + "ewc_loss": 0.000850677490234375, + "ewc_loss_diag": 4.0605664253234863e-07, + "ewc_loss_parallel": 8.083879947662354e-07, + "grad_norm": 3.6533007621765137, + "learning_rate": 8.435777871979652e-08, + "loss": 0.6077, + "mean_token_accuracy": 0.8160756826400757, + "num_tokens": 7822354.0, + "step": 200 + }, + { + "epoch": 0.02556926599669253, + "ewc_loss": 0.000873565673828125, + "ewc_loss_diag": 4.1350722312927246e-07, + "ewc_loss_parallel": 8.307397365570068e-07, + "grad_norm": 3.8821699619293213, + "learning_rate": 8.47816871555744e-08, + "loss": 0.5493, + "mean_token_accuracy": 0.8302335143089294, + "num_tokens": 7854065.0, + "step": 201 + }, + { + "epoch": 0.025696476275283042, + "ewc_loss": 0.000896453857421875, + "ewc_loss_diag": 4.1909515857696533e-07, + "ewc_loss_parallel": 8.530914783477783e-07, + "grad_norm": 3.590580463409424, + "learning_rate": 8.520559559135227e-08, + "loss": 0.527, + "mean_token_accuracy": 0.8359649181365967, + "num_tokens": 7891325.0, + "step": 202 + }, + { + "epoch": 0.025823686553873553, + "ewc_loss": 0.000896453857421875, + "ewc_loss_diag": 4.209578037261963e-07, + "ewc_loss_parallel": 8.530914783477783e-07, + "grad_norm": 3.428645610809326, + "learning_rate": 8.562950402713014e-08, + "loss": 0.6056, + "mean_token_accuracy": 0.8168082237243652, + "num_tokens": 7939758.0, + "step": 203 + }, + { + "epoch": 0.025950896832464064, + "ewc_loss": 0.000888824462890625, + "ewc_loss_diag": 4.284083843231201e-07, + "ewc_loss_parallel": 8.456408977508545e-07, + "grad_norm": 3.8268139362335205, + "learning_rate": 8.605341246290801e-08, + "loss": 0.5781, + "mean_token_accuracy": 0.8222751021385193, + "num_tokens": 7971195.0, + "step": 204 + }, + { + "epoch": 0.026078107111054575, + "ewc_loss": 0.000896453857421875, + "ewc_loss_diag": 4.3585896492004395e-07, + "ewc_loss_parallel": 8.530914783477783e-07, + "grad_norm": 3.4470410346984863, + "learning_rate": 8.647732089868589e-08, + "loss": 0.5574, + "mean_token_accuracy": 0.8304322957992554, + "num_tokens": 8012254.0, + "step": 205 + }, + { + "epoch": 0.026205317389645082, + "ewc_loss": 0.00089263916015625, + "ewc_loss_diag": 4.3958425521850586e-07, + "ewc_loss_parallel": 8.493661880493164e-07, + "grad_norm": 3.5167558193206787, + "learning_rate": 8.690122933446376e-08, + "loss": 0.5872, + "mean_token_accuracy": 0.8217768669128418, + "num_tokens": 8056314.0, + "step": 206 + }, + { + "epoch": 0.026332527668235593, + "ewc_loss": 0.00089263916015625, + "ewc_loss_diag": 4.414469003677368e-07, + "ewc_loss_parallel": 8.456408977508545e-07, + "grad_norm": 3.353006601333618, + "learning_rate": 8.732513777024163e-08, + "loss": 0.5067, + "mean_token_accuracy": 0.8463214039802551, + "num_tokens": 8098501.0, + "step": 207 + }, + { + "epoch": 0.026459737946826104, + "ewc_loss": 0.000888824462890625, + "ewc_loss_diag": 4.507601261138916e-07, + "ewc_loss_parallel": 8.419156074523926e-07, + "grad_norm": 3.479749917984009, + "learning_rate": 8.77490462060195e-08, + "loss": 0.5773, + "mean_token_accuracy": 0.8248157501220703, + "num_tokens": 8142085.0, + "step": 208 + }, + { + "epoch": 0.026586948225416615, + "ewc_loss": 0.000896453857421875, + "ewc_loss_diag": 4.600733518600464e-07, + "ewc_loss_parallel": 8.493661880493164e-07, + "grad_norm": 3.5914053916931152, + "learning_rate": 8.817295464179738e-08, + "loss": 0.5357, + "mean_token_accuracy": 0.8316657543182373, + "num_tokens": 8178452.0, + "step": 209 + }, + { + "epoch": 0.026714158504007122, + "ewc_loss": 0.0009002685546875, + "ewc_loss_diag": 4.6193599700927734e-07, + "ewc_loss_parallel": 8.530914783477783e-07, + "grad_norm": 3.5337677001953125, + "learning_rate": 8.859686307757525e-08, + "loss": 0.5498, + "mean_token_accuracy": 0.8313318490982056, + "num_tokens": 8215761.0, + "step": 210 + }, + { + "epoch": 0.026841368782597633, + "ewc_loss": 0.00091552734375, + "ewc_loss_diag": 4.7124922275543213e-07, + "ewc_loss_parallel": 8.67992639541626e-07, + "grad_norm": 3.531503915786743, + "learning_rate": 8.902077151335312e-08, + "loss": 0.577, + "mean_token_accuracy": 0.8216167092323303, + "num_tokens": 8256400.0, + "step": 211 + }, + { + "epoch": 0.026968579061188144, + "ewc_loss": 0.00091552734375, + "ewc_loss_diag": 4.7497451305389404e-07, + "ewc_loss_parallel": 8.67992639541626e-07, + "grad_norm": 4.645802021026611, + "learning_rate": 8.944467994913098e-08, + "loss": 0.6252, + "mean_token_accuracy": 0.8081684708595276, + "num_tokens": 8290615.0, + "step": 212 + }, + { + "epoch": 0.027095789339778655, + "ewc_loss": 0.00096893310546875, + "ewc_loss_diag": 4.805624485015869e-07, + "ewc_loss_parallel": 9.201467037200928e-07, + "grad_norm": 3.6090402603149414, + "learning_rate": 8.986858838490885e-08, + "loss": 0.5003, + "mean_token_accuracy": 0.8397128582000732, + "num_tokens": 8324391.0, + "step": 213 + }, + { + "epoch": 0.027222999618369163, + "ewc_loss": 0.00095367431640625, + "ewc_loss_diag": 4.917383193969727e-07, + "ewc_loss_parallel": 9.052455425262451e-07, + "grad_norm": 3.341270685195923, + "learning_rate": 9.029249682068673e-08, + "loss": 0.5489, + "mean_token_accuracy": 0.8283582329750061, + "num_tokens": 8368995.0, + "step": 214 + }, + { + "epoch": 0.027350209896959674, + "ewc_loss": 0.00092315673828125, + "ewc_loss_diag": 4.991888999938965e-07, + "ewc_loss_parallel": 8.717179298400879e-07, + "grad_norm": 3.6014466285705566, + "learning_rate": 9.07164052564646e-08, + "loss": 0.5671, + "mean_token_accuracy": 0.8228363394737244, + "num_tokens": 8405460.0, + "step": 215 + }, + { + "epoch": 0.027477420175550184, + "ewc_loss": 0.0009002685546875, + "ewc_loss_diag": 5.029141902923584e-07, + "ewc_loss_parallel": 8.493661880493164e-07, + "grad_norm": 3.5080766677856445, + "learning_rate": 9.114031369224247e-08, + "loss": 0.5366, + "mean_token_accuracy": 0.8346326351165771, + "num_tokens": 8443334.0, + "step": 216 + }, + { + "epoch": 0.027604630454140695, + "ewc_loss": 0.0009002685546875, + "ewc_loss_diag": 5.103647708892822e-07, + "ewc_loss_parallel": 8.493661880493164e-07, + "grad_norm": 3.3113019466400146, + "learning_rate": 9.156422212802034e-08, + "loss": 0.4985, + "mean_token_accuracy": 0.8426185250282288, + "num_tokens": 8485462.0, + "step": 217 + }, + { + "epoch": 0.027731840732731206, + "ewc_loss": 0.00089263916015625, + "ewc_loss_diag": 5.140900611877441e-07, + "ewc_loss_parallel": 8.381903171539307e-07, + "grad_norm": 3.659163475036621, + "learning_rate": 9.198813056379822e-08, + "loss": 0.5419, + "mean_token_accuracy": 0.8299498558044434, + "num_tokens": 8520862.0, + "step": 218 + }, + { + "epoch": 0.027859051011321714, + "ewc_loss": 0.000904083251953125, + "ewc_loss_diag": 5.252659320831299e-07, + "ewc_loss_parallel": 8.493661880493164e-07, + "grad_norm": 3.4526233673095703, + "learning_rate": 9.241203899957609e-08, + "loss": 0.5477, + "mean_token_accuracy": 0.8290165662765503, + "num_tokens": 8560565.0, + "step": 219 + }, + { + "epoch": 0.027986261289912225, + "ewc_loss": 0.0009002685546875, + "ewc_loss_diag": 5.289912223815918e-07, + "ewc_loss_parallel": 8.456408977508545e-07, + "grad_norm": 3.8353028297424316, + "learning_rate": 9.283594743535396e-08, + "loss": 0.6537, + "mean_token_accuracy": 0.7984663248062134, + "num_tokens": 8597752.0, + "step": 220 + }, + { + "epoch": 0.028113471568502735, + "ewc_loss": 0.00091552734375, + "ewc_loss_diag": 5.327165126800537e-07, + "ewc_loss_parallel": 8.605420589447021e-07, + "grad_norm": 3.6025872230529785, + "learning_rate": 9.325985587113183e-08, + "loss": 0.5222, + "mean_token_accuracy": 0.8399413824081421, + "num_tokens": 8634975.0, + "step": 221 + }, + { + "epoch": 0.028240681847093246, + "ewc_loss": 0.00091552734375, + "ewc_loss_diag": 5.401670932769775e-07, + "ewc_loss_parallel": 8.605420589447021e-07, + "grad_norm": 3.6726441383361816, + "learning_rate": 9.368376430690971e-08, + "loss": 0.5262, + "mean_token_accuracy": 0.8350435495376587, + "num_tokens": 8667850.0, + "step": 222 + }, + { + "epoch": 0.028367892125683754, + "ewc_loss": 0.00090789794921875, + "ewc_loss_diag": 5.438923835754395e-07, + "ewc_loss_parallel": 8.530914783477783e-07, + "grad_norm": 3.5526459217071533, + "learning_rate": 9.410767274268758e-08, + "loss": 0.5494, + "mean_token_accuracy": 0.8315821290016174, + "num_tokens": 8708860.0, + "step": 223 + }, + { + "epoch": 0.028495102404274265, + "ewc_loss": 0.000904083251953125, + "ewc_loss_diag": 5.438923835754395e-07, + "ewc_loss_parallel": 8.493661880493164e-07, + "grad_norm": 3.5045902729034424, + "learning_rate": 9.453158117846545e-08, + "loss": 0.5376, + "mean_token_accuracy": 0.8330100178718567, + "num_tokens": 8748893.0, + "step": 224 + }, + { + "epoch": 0.028622312682864776, + "ewc_loss": 0.0009002685546875, + "ewc_loss_diag": 5.438923835754395e-07, + "ewc_loss_parallel": 8.456408977508545e-07, + "grad_norm": 3.5479214191436768, + "learning_rate": 9.495548961424333e-08, + "loss": 0.5659, + "mean_token_accuracy": 0.8263484835624695, + "num_tokens": 8786745.0, + "step": 225 + }, + { + "epoch": 0.028749522961455286, + "ewc_loss": 0.00089263916015625, + "ewc_loss_diag": 5.476176738739014e-07, + "ewc_loss_parallel": 8.381903171539307e-07, + "grad_norm": 3.4839189052581787, + "learning_rate": 9.53793980500212e-08, + "loss": 0.4866, + "mean_token_accuracy": 0.8470454216003418, + "num_tokens": 8819618.0, + "step": 226 + }, + { + "epoch": 0.028876733240045797, + "ewc_loss": 0.000885009765625, + "ewc_loss_diag": 5.476176738739014e-07, + "ewc_loss_parallel": 8.307397365570068e-07, + "grad_norm": 3.6206963062286377, + "learning_rate": 9.580330648579907e-08, + "loss": 0.63, + "mean_token_accuracy": 0.8111535310745239, + "num_tokens": 8860421.0, + "step": 227 + }, + { + "epoch": 0.029003943518636305, + "ewc_loss": 0.00089263916015625, + "ewc_loss_diag": 5.476176738739014e-07, + "ewc_loss_parallel": 8.381903171539307e-07, + "grad_norm": 3.4909183979034424, + "learning_rate": 9.622721492157694e-08, + "loss": 0.5224, + "mean_token_accuracy": 0.8374423980712891, + "num_tokens": 8898184.0, + "step": 228 + }, + { + "epoch": 0.029131153797226816, + "ewc_loss": 0.00087738037109375, + "ewc_loss_diag": 5.476176738739014e-07, + "ewc_loss_parallel": 8.23289155960083e-07, + "grad_norm": 3.648021936416626, + "learning_rate": 9.665112335735482e-08, + "loss": 0.5202, + "mean_token_accuracy": 0.8356990814208984, + "num_tokens": 8930802.0, + "step": 229 + }, + { + "epoch": 0.029258364075817327, + "ewc_loss": 0.000881195068359375, + "ewc_loss_diag": 5.513429641723633e-07, + "ewc_loss_parallel": 8.270144462585449e-07, + "grad_norm": 3.513216018676758, + "learning_rate": 9.707503179313267e-08, + "loss": 0.4874, + "mean_token_accuracy": 0.8462557792663574, + "num_tokens": 8964223.0, + "step": 230 + }, + { + "epoch": 0.029385574354407838, + "ewc_loss": 0.00087738037109375, + "ewc_loss_diag": 5.550682544708252e-07, + "ewc_loss_parallel": 8.195638656616211e-07, + "grad_norm": 3.6365182399749756, + "learning_rate": 9.749894022891055e-08, + "loss": 0.5447, + "mean_token_accuracy": 0.828624427318573, + "num_tokens": 9001110.0, + "step": 231 + }, + { + "epoch": 0.029512784632998345, + "ewc_loss": 0.000885009765625, + "ewc_loss_diag": 5.587935447692871e-07, + "ewc_loss_parallel": 8.270144462585449e-07, + "grad_norm": 3.369434118270874, + "learning_rate": 9.792284866468842e-08, + "loss": 0.5474, + "mean_token_accuracy": 0.827142596244812, + "num_tokens": 9045952.0, + "step": 232 + }, + { + "epoch": 0.029639994911588856, + "ewc_loss": 0.000881195068359375, + "ewc_loss_diag": 5.699694156646729e-07, + "ewc_loss_parallel": 8.23289155960083e-07, + "grad_norm": 3.4443750381469727, + "learning_rate": 9.834675710046629e-08, + "loss": 0.5198, + "mean_token_accuracy": 0.8360518217086792, + "num_tokens": 9082291.0, + "step": 233 + }, + { + "epoch": 0.029767205190179367, + "ewc_loss": 0.0008697509765625, + "ewc_loss_diag": 5.736947059631348e-07, + "ewc_loss_parallel": 8.121132850646973e-07, + "grad_norm": 3.546725034713745, + "learning_rate": 9.877066553624416e-08, + "loss": 0.5476, + "mean_token_accuracy": 0.8250584602355957, + "num_tokens": 9120280.0, + "step": 234 + }, + { + "epoch": 0.029894415468769878, + "ewc_loss": 0.000873565673828125, + "ewc_loss_diag": 5.736947059631348e-07, + "ewc_loss_parallel": 8.158385753631592e-07, + "grad_norm": 3.715522289276123, + "learning_rate": 9.919457397202204e-08, + "loss": 0.5553, + "mean_token_accuracy": 0.8279068470001221, + "num_tokens": 9156253.0, + "step": 235 + }, + { + "epoch": 0.030021625747360385, + "ewc_loss": 0.000881195068359375, + "ewc_loss_diag": 5.774199962615967e-07, + "ewc_loss_parallel": 8.23289155960083e-07, + "grad_norm": 4.008385181427002, + "learning_rate": 9.961848240779991e-08, + "loss": 0.6102, + "mean_token_accuracy": 0.8140488266944885, + "num_tokens": 9193220.0, + "step": 236 + }, + { + "epoch": 0.030148836025950896, + "ewc_loss": 0.000896453857421875, + "ewc_loss_diag": 5.811452865600586e-07, + "ewc_loss_parallel": 8.381903171539307e-07, + "grad_norm": 3.6637213230133057, + "learning_rate": 1.0004239084357778e-07, + "loss": 0.533, + "mean_token_accuracy": 0.8327759504318237, + "num_tokens": 9230589.0, + "step": 237 + }, + { + "epoch": 0.030276046304541407, + "ewc_loss": 0.000888824462890625, + "ewc_loss_diag": 5.848705768585205e-07, + "ewc_loss_parallel": 8.307397365570068e-07, + "grad_norm": 3.385979413986206, + "learning_rate": 1.0046629927935566e-07, + "loss": 0.5184, + "mean_token_accuracy": 0.8391153812408447, + "num_tokens": 9270383.0, + "step": 238 + }, + { + "epoch": 0.030403256583131918, + "ewc_loss": 0.000865936279296875, + "ewc_loss_diag": 5.848705768585205e-07, + "ewc_loss_parallel": 8.083879947662354e-07, + "grad_norm": 3.3919973373413086, + "learning_rate": 1.0089020771513353e-07, + "loss": 0.4942, + "mean_token_accuracy": 0.845887303352356, + "num_tokens": 9309402.0, + "step": 239 + }, + { + "epoch": 0.03053046686172243, + "ewc_loss": 0.000858306884765625, + "ewc_loss_diag": 5.885958671569824e-07, + "ewc_loss_parallel": 8.009374141693115e-07, + "grad_norm": 3.5566275119781494, + "learning_rate": 1.013141161509114e-07, + "loss": 0.5273, + "mean_token_accuracy": 0.8347175121307373, + "num_tokens": 9344759.0, + "step": 240 + }, + { + "epoch": 0.030657677140312936, + "ewc_loss": 0.000865936279296875, + "ewc_loss_diag": 5.960464477539062e-07, + "ewc_loss_parallel": 8.046627044677734e-07, + "grad_norm": 3.6317880153656006, + "learning_rate": 1.0173802458668927e-07, + "loss": 0.5412, + "mean_token_accuracy": 0.8277738690376282, + "num_tokens": 9377915.0, + "step": 241 + }, + { + "epoch": 0.030784887418903447, + "ewc_loss": 0.0008697509765625, + "ewc_loss_diag": 5.960464477539062e-07, + "ewc_loss_parallel": 8.083879947662354e-07, + "grad_norm": 3.6657776832580566, + "learning_rate": 1.0216193302246715e-07, + "loss": 0.5752, + "mean_token_accuracy": 0.8157230615615845, + "num_tokens": 9411971.0, + "step": 242 + }, + { + "epoch": 0.030912097697493958, + "ewc_loss": 0.000873565673828125, + "ewc_loss_diag": 5.960464477539062e-07, + "ewc_loss_parallel": 8.121132850646973e-07, + "grad_norm": 3.311704397201538, + "learning_rate": 1.0258584145824502e-07, + "loss": 0.4998, + "mean_token_accuracy": 0.8410053253173828, + "num_tokens": 9453786.0, + "step": 243 + }, + { + "epoch": 0.03103930797608447, + "ewc_loss": 0.000850677490234375, + "ewc_loss_diag": 5.960464477539062e-07, + "ewc_loss_parallel": 7.897615432739258e-07, + "grad_norm": 3.3849527835845947, + "learning_rate": 1.0300974989402289e-07, + "loss": 0.4851, + "mean_token_accuracy": 0.8438498973846436, + "num_tokens": 9489371.0, + "step": 244 + }, + { + "epoch": 0.031166518254674976, + "ewc_loss": 0.000858306884765625, + "ewc_loss_diag": 6.07222318649292e-07, + "ewc_loss_parallel": 7.972121238708496e-07, + "grad_norm": 3.3376715183258057, + "learning_rate": 1.0343365832980076e-07, + "loss": 0.5365, + "mean_token_accuracy": 0.8302391767501831, + "num_tokens": 9532357.0, + "step": 245 + }, + { + "epoch": 0.03129372853326549, + "ewc_loss": 0.0008544921875, + "ewc_loss_diag": 6.221234798431396e-07, + "ewc_loss_parallel": 7.934868335723877e-07, + "grad_norm": 3.3852972984313965, + "learning_rate": 1.0385756676557864e-07, + "loss": 0.5214, + "mean_token_accuracy": 0.8371726870536804, + "num_tokens": 9574260.0, + "step": 246 + }, + { + "epoch": 0.031420938811855995, + "ewc_loss": 0.000865936279296875, + "ewc_loss_diag": 6.332993507385254e-07, + "ewc_loss_parallel": 8.009374141693115e-07, + "grad_norm": 3.383253335952759, + "learning_rate": 1.0428147520135651e-07, + "loss": 0.5092, + "mean_token_accuracy": 0.8410772681236267, + "num_tokens": 9614167.0, + "step": 247 + }, + { + "epoch": 0.03154814909044651, + "ewc_loss": 0.00086212158203125, + "ewc_loss_diag": 6.370246410369873e-07, + "ewc_loss_parallel": 7.972121238708496e-07, + "grad_norm": 3.625704765319824, + "learning_rate": 1.0470538363713437e-07, + "loss": 0.541, + "mean_token_accuracy": 0.8286000490188599, + "num_tokens": 9652790.0, + "step": 248 + }, + { + "epoch": 0.031675359369037016, + "ewc_loss": 0.000865936279296875, + "ewc_loss_diag": 6.444752216339111e-07, + "ewc_loss_parallel": 8.009374141693115e-07, + "grad_norm": 3.5108656883239746, + "learning_rate": 1.0512929207291224e-07, + "loss": 0.5307, + "mean_token_accuracy": 0.8311680555343628, + "num_tokens": 9692712.0, + "step": 249 + }, + { + "epoch": 0.03180256964762753, + "ewc_loss": 0.000865936279296875, + "ewc_loss_diag": 6.444752216339111e-07, + "ewc_loss_parallel": 8.009374141693115e-07, + "grad_norm": 3.667445659637451, + "learning_rate": 1.0555320050869011e-07, + "loss": 0.5185, + "mean_token_accuracy": 0.8330173492431641, + "num_tokens": 9725438.0, + "step": 250 + }, + { + "epoch": 0.03192977992621804, + "ewc_loss": 0.00087738037109375, + "ewc_loss_diag": 6.51925802230835e-07, + "ewc_loss_parallel": 8.121132850646973e-07, + "grad_norm": 3.8334603309631348, + "learning_rate": 1.0597710894446799e-07, + "loss": 0.5744, + "mean_token_accuracy": 0.8153671026229858, + "num_tokens": 9758325.0, + "step": 251 + }, + { + "epoch": 0.032056990204808546, + "ewc_loss": 0.00087738037109375, + "ewc_loss_diag": 6.51925802230835e-07, + "ewc_loss_parallel": 8.121132850646973e-07, + "grad_norm": 3.4063963890075684, + "learning_rate": 1.0640101738024586e-07, + "loss": 0.4971, + "mean_token_accuracy": 0.8437950611114502, + "num_tokens": 9800266.0, + "step": 252 + }, + { + "epoch": 0.03218420048339906, + "ewc_loss": 0.000865936279296875, + "ewc_loss_diag": 6.556510925292969e-07, + "ewc_loss_parallel": 8.009374141693115e-07, + "grad_norm": 3.4986917972564697, + "learning_rate": 1.0682492581602373e-07, + "loss": 0.4705, + "mean_token_accuracy": 0.8510997891426086, + "num_tokens": 9833602.0, + "step": 253 + }, + { + "epoch": 0.03231141076198957, + "ewc_loss": 0.0008697509765625, + "ewc_loss_diag": 6.631016731262207e-07, + "ewc_loss_parallel": 8.046627044677734e-07, + "grad_norm": 3.433974027633667, + "learning_rate": 1.072488342518016e-07, + "loss": 0.4621, + "mean_token_accuracy": 0.8521831035614014, + "num_tokens": 9869874.0, + "step": 254 + }, + { + "epoch": 0.03243862104058008, + "ewc_loss": 0.000865936279296875, + "ewc_loss_diag": 6.631016731262207e-07, + "ewc_loss_parallel": 8.009374141693115e-07, + "grad_norm": 3.3811819553375244, + "learning_rate": 1.0767274268757948e-07, + "loss": 0.5123, + "mean_token_accuracy": 0.8385353088378906, + "num_tokens": 9910724.0, + "step": 255 + }, + { + "epoch": 0.03256583131917059, + "ewc_loss": 0.0008697509765625, + "ewc_loss_diag": 6.742775440216064e-07, + "ewc_loss_parallel": 8.009374141693115e-07, + "grad_norm": 3.414182424545288, + "learning_rate": 1.0809665112335735e-07, + "loss": 0.5467, + "mean_token_accuracy": 0.8305460810661316, + "num_tokens": 9955612.0, + "step": 256 + }, + { + "epoch": 0.0326930415977611, + "ewc_loss": 0.0008697509765625, + "ewc_loss_diag": 6.742775440216064e-07, + "ewc_loss_parallel": 8.009374141693115e-07, + "grad_norm": 3.232944965362549, + "learning_rate": 1.0852055955913522e-07, + "loss": 0.4329, + "mean_token_accuracy": 0.8603044152259827, + "num_tokens": 9994954.0, + "step": 257 + }, + { + "epoch": 0.03282025187635161, + "ewc_loss": 0.000865936279296875, + "ewc_loss_diag": 6.780028343200684e-07, + "ewc_loss_parallel": 7.972121238708496e-07, + "grad_norm": 3.449894428253174, + "learning_rate": 1.089444679949131e-07, + "loss": 0.5576, + "mean_token_accuracy": 0.8243522644042969, + "num_tokens": 10035503.0, + "step": 258 + }, + { + "epoch": 0.03294746215494212, + "ewc_loss": 0.000873565673828125, + "ewc_loss_diag": 6.780028343200684e-07, + "ewc_loss_parallel": 8.046627044677734e-07, + "grad_norm": 3.332015037536621, + "learning_rate": 1.0936837643069097e-07, + "loss": 0.4921, + "mean_token_accuracy": 0.8408888578414917, + "num_tokens": 10078296.0, + "step": 259 + }, + { + "epoch": 0.03307467243353263, + "ewc_loss": 0.000865936279296875, + "ewc_loss_diag": 6.891787052154541e-07, + "ewc_loss_parallel": 7.972121238708496e-07, + "grad_norm": 3.4999442100524902, + "learning_rate": 1.0979228486646884e-07, + "loss": 0.4635, + "mean_token_accuracy": 0.853759765625, + "num_tokens": 10113411.0, + "step": 260 + }, + { + "epoch": 0.03320188271212314, + "ewc_loss": 0.000873565673828125, + "ewc_loss_diag": 6.891787052154541e-07, + "ewc_loss_parallel": 8.046627044677734e-07, + "grad_norm": 3.542860507965088, + "learning_rate": 1.1021619330224671e-07, + "loss": 0.5666, + "mean_token_accuracy": 0.8190630674362183, + "num_tokens": 10155184.0, + "step": 261 + }, + { + "epoch": 0.03332909299071365, + "ewc_loss": 0.00087738037109375, + "ewc_loss_diag": 6.966292858123779e-07, + "ewc_loss_parallel": 8.083879947662354e-07, + "grad_norm": 3.576894998550415, + "learning_rate": 1.1064010173802458e-07, + "loss": 0.5507, + "mean_token_accuracy": 0.827172040939331, + "num_tokens": 10193612.0, + "step": 262 + }, + { + "epoch": 0.03345630326930416, + "ewc_loss": 0.000881195068359375, + "ewc_loss_diag": 7.115304470062256e-07, + "ewc_loss_parallel": 8.083879947662354e-07, + "grad_norm": 3.44023060798645, + "learning_rate": 1.1106401017380246e-07, + "loss": 0.4736, + "mean_token_accuracy": 0.8474130034446716, + "num_tokens": 10231243.0, + "step": 263 + }, + { + "epoch": 0.03358351354789467, + "ewc_loss": 0.000873565673828125, + "ewc_loss_diag": 7.189810276031494e-07, + "ewc_loss_parallel": 8.009374141693115e-07, + "grad_norm": 3.933940887451172, + "learning_rate": 1.1148791860958033e-07, + "loss": 0.561, + "mean_token_accuracy": 0.8182637095451355, + "num_tokens": 10260802.0, + "step": 264 + }, + { + "epoch": 0.03371072382648518, + "ewc_loss": 0.000896453857421875, + "ewc_loss_diag": 7.264316082000732e-07, + "ewc_loss_parallel": 8.23289155960083e-07, + "grad_norm": 3.5165233612060547, + "learning_rate": 1.119118270453582e-07, + "loss": 0.528, + "mean_token_accuracy": 0.8321706652641296, + "num_tokens": 10300737.0, + "step": 265 + }, + { + "epoch": 0.03383793410507569, + "ewc_loss": 0.000881195068359375, + "ewc_loss_diag": 7.301568984985352e-07, + "ewc_loss_parallel": 8.083879947662354e-07, + "grad_norm": 3.480727195739746, + "learning_rate": 1.1233573548113607e-07, + "loss": 0.5351, + "mean_token_accuracy": 0.8312498927116394, + "num_tokens": 10340931.0, + "step": 266 + }, + { + "epoch": 0.0339651443836662, + "ewc_loss": 0.0008697509765625, + "ewc_loss_diag": 7.301568984985352e-07, + "ewc_loss_parallel": 7.972121238708496e-07, + "grad_norm": 3.314594030380249, + "learning_rate": 1.1275964391691393e-07, + "loss": 0.5138, + "mean_token_accuracy": 0.8375027775764465, + "num_tokens": 10385218.0, + "step": 267 + }, + { + "epoch": 0.03409235466225671, + "ewc_loss": 0.000865936279296875, + "ewc_loss_diag": 7.301568984985352e-07, + "ewc_loss_parallel": 7.934868335723877e-07, + "grad_norm": 3.594456195831299, + "learning_rate": 1.131835523526918e-07, + "loss": 0.561, + "mean_token_accuracy": 0.8195515871047974, + "num_tokens": 10425701.0, + "step": 268 + }, + { + "epoch": 0.03421956494084722, + "ewc_loss": 0.000873565673828125, + "ewc_loss_diag": 7.37607479095459e-07, + "ewc_loss_parallel": 8.009374141693115e-07, + "grad_norm": 3.6613504886627197, + "learning_rate": 1.1360746078846968e-07, + "loss": 0.555, + "mean_token_accuracy": 0.8271257877349854, + "num_tokens": 10464091.0, + "step": 269 + }, + { + "epoch": 0.03434677521943773, + "ewc_loss": 0.00089263916015625, + "ewc_loss_diag": 7.413327693939209e-07, + "ewc_loss_parallel": 8.195638656616211e-07, + "grad_norm": 3.630689859390259, + "learning_rate": 1.1403136922424755e-07, + "loss": 0.5033, + "mean_token_accuracy": 0.8372784852981567, + "num_tokens": 10498293.0, + "step": 270 + }, + { + "epoch": 0.03447398549802824, + "ewc_loss": 0.00089263916015625, + "ewc_loss_diag": 7.450580596923828e-07, + "ewc_loss_parallel": 8.158385753631592e-07, + "grad_norm": 3.4756932258605957, + "learning_rate": 1.1445527766002542e-07, + "loss": 0.5134, + "mean_token_accuracy": 0.8325279355049133, + "num_tokens": 10540524.0, + "step": 271 + }, + { + "epoch": 0.03460119577661875, + "ewc_loss": 0.00087738037109375, + "ewc_loss_diag": 7.487833499908447e-07, + "ewc_loss_parallel": 8.009374141693115e-07, + "grad_norm": 3.7329976558685303, + "learning_rate": 1.148791860958033e-07, + "loss": 0.5915, + "mean_token_accuracy": 0.8133760690689087, + "num_tokens": 10577235.0, + "step": 272 + }, + { + "epoch": 0.034728406055209264, + "ewc_loss": 0.000885009765625, + "ewc_loss_diag": 7.525086402893066e-07, + "ewc_loss_parallel": 8.083879947662354e-07, + "grad_norm": 3.4410080909729004, + "learning_rate": 1.1530309453158117e-07, + "loss": 0.5202, + "mean_token_accuracy": 0.8362274169921875, + "num_tokens": 10618536.0, + "step": 273 + }, + { + "epoch": 0.03485561633379977, + "ewc_loss": 0.000881195068359375, + "ewc_loss_diag": 7.525086402893066e-07, + "ewc_loss_parallel": 8.046627044677734e-07, + "grad_norm": 3.5141966342926025, + "learning_rate": 1.1572700296735904e-07, + "loss": 0.5201, + "mean_token_accuracy": 0.8322389125823975, + "num_tokens": 10657348.0, + "step": 274 + }, + { + "epoch": 0.03498282661239028, + "ewc_loss": 0.000885009765625, + "ewc_loss_diag": 7.599592208862305e-07, + "ewc_loss_parallel": 8.083879947662354e-07, + "grad_norm": 3.47796893119812, + "learning_rate": 1.1615091140313691e-07, + "loss": 0.539, + "mean_token_accuracy": 0.8293285965919495, + "num_tokens": 10697496.0, + "step": 275 + }, + { + "epoch": 0.03511003689098079, + "ewc_loss": 0.000881195068359375, + "ewc_loss_diag": 7.599592208862305e-07, + "ewc_loss_parallel": 8.046627044677734e-07, + "grad_norm": 3.3666024208068848, + "learning_rate": 1.1657481983891479e-07, + "loss": 0.4919, + "mean_token_accuracy": 0.8430999517440796, + "num_tokens": 10740302.0, + "step": 276 + }, + { + "epoch": 0.0352372471695713, + "ewc_loss": 0.00087738037109375, + "ewc_loss_diag": 7.636845111846924e-07, + "ewc_loss_parallel": 8.009374141693115e-07, + "grad_norm": 3.428591251373291, + "learning_rate": 1.1699872827469266e-07, + "loss": 0.5019, + "mean_token_accuracy": 0.8403818607330322, + "num_tokens": 10780558.0, + "step": 277 + }, + { + "epoch": 0.03536445744816181, + "ewc_loss": 0.000881195068359375, + "ewc_loss_diag": 7.674098014831543e-07, + "ewc_loss_parallel": 8.046627044677734e-07, + "grad_norm": 3.4546542167663574, + "learning_rate": 1.1742263671047053e-07, + "loss": 0.5165, + "mean_token_accuracy": 0.8346507549285889, + "num_tokens": 10821913.0, + "step": 278 + }, + { + "epoch": 0.03549166772675232, + "ewc_loss": 0.000885009765625, + "ewc_loss_diag": 7.674098014831543e-07, + "ewc_loss_parallel": 8.083879947662354e-07, + "grad_norm": 3.702564001083374, + "learning_rate": 1.178465451462484e-07, + "loss": 0.5963, + "mean_token_accuracy": 0.8102583885192871, + "num_tokens": 10861608.0, + "step": 279 + }, + { + "epoch": 0.03561887800534283, + "ewc_loss": 0.000896453857421875, + "ewc_loss_diag": 7.748603820800781e-07, + "ewc_loss_parallel": 8.195638656616211e-07, + "grad_norm": 3.7156784534454346, + "learning_rate": 1.1827045358202628e-07, + "loss": 0.5359, + "mean_token_accuracy": 0.8343812227249146, + "num_tokens": 10895830.0, + "step": 280 + }, + { + "epoch": 0.035746088283933344, + "ewc_loss": 0.00090789794921875, + "ewc_loss_diag": 7.748603820800781e-07, + "ewc_loss_parallel": 8.307397365570068e-07, + "grad_norm": 3.3977267742156982, + "learning_rate": 1.1869436201780415e-07, + "loss": 0.4642, + "mean_token_accuracy": 0.8534811735153198, + "num_tokens": 10934623.0, + "step": 281 + }, + { + "epoch": 0.03587329856252385, + "ewc_loss": 0.000888824462890625, + "ewc_loss_diag": 7.7858567237854e-07, + "ewc_loss_parallel": 8.121132850646973e-07, + "grad_norm": 3.408665180206299, + "learning_rate": 1.1911827045358202e-07, + "loss": 0.4982, + "mean_token_accuracy": 0.8444656133651733, + "num_tokens": 10975304.0, + "step": 282 + }, + { + "epoch": 0.03600050884111436, + "ewc_loss": 0.00089263916015625, + "ewc_loss_diag": 7.82310962677002e-07, + "ewc_loss_parallel": 8.121132850646973e-07, + "grad_norm": 3.432419776916504, + "learning_rate": 1.195421788893599e-07, + "loss": 0.5525, + "mean_token_accuracy": 0.8260555863380432, + "num_tokens": 11017294.0, + "step": 283 + }, + { + "epoch": 0.036127719119704874, + "ewc_loss": 0.00089263916015625, + "ewc_loss_diag": 7.860362529754639e-07, + "ewc_loss_parallel": 8.121132850646973e-07, + "grad_norm": 3.633502960205078, + "learning_rate": 1.1996608732513778e-07, + "loss": 0.5243, + "mean_token_accuracy": 0.832012951374054, + "num_tokens": 11052687.0, + "step": 284 + }, + { + "epoch": 0.03625492939829538, + "ewc_loss": 0.0009002685546875, + "ewc_loss_diag": 7.972121238708496e-07, + "ewc_loss_parallel": 8.195638656616211e-07, + "grad_norm": 3.4715185165405273, + "learning_rate": 1.2038999576091563e-07, + "loss": 0.4883, + "mean_token_accuracy": 0.8428550958633423, + "num_tokens": 11091892.0, + "step": 285 + }, + { + "epoch": 0.036382139676885895, + "ewc_loss": 0.000904083251953125, + "ewc_loss_diag": 7.972121238708496e-07, + "ewc_loss_parallel": 8.23289155960083e-07, + "grad_norm": 3.34037184715271, + "learning_rate": 1.208139041966935e-07, + "loss": 0.4423, + "mean_token_accuracy": 0.8582640886306763, + "num_tokens": 11132446.0, + "step": 286 + }, + { + "epoch": 0.0365093499554764, + "ewc_loss": 0.000888824462890625, + "ewc_loss_diag": 8.009374141693115e-07, + "ewc_loss_parallel": 8.083879947662354e-07, + "grad_norm": 3.6911745071411133, + "learning_rate": 1.2123781263247137e-07, + "loss": 0.5022, + "mean_token_accuracy": 0.8373855948448181, + "num_tokens": 11166721.0, + "step": 287 + }, + { + "epoch": 0.03663656023406691, + "ewc_loss": 0.00091552734375, + "ewc_loss_diag": 8.046627044677734e-07, + "ewc_loss_parallel": 8.344650268554688e-07, + "grad_norm": 3.883493661880493, + "learning_rate": 1.2166172106824924e-07, + "loss": 0.5601, + "mean_token_accuracy": 0.8284640312194824, + "num_tokens": 11199570.0, + "step": 288 + }, + { + "epoch": 0.036763770512657425, + "ewc_loss": 0.0009307861328125, + "ewc_loss_diag": 8.083879947662354e-07, + "ewc_loss_parallel": 8.493661880493164e-07, + "grad_norm": 3.5835766792297363, + "learning_rate": 1.2208562950402712e-07, + "loss": 0.4757, + "mean_token_accuracy": 0.8458677530288696, + "num_tokens": 11236604.0, + "step": 289 + }, + { + "epoch": 0.03689098079124793, + "ewc_loss": 0.00091552734375, + "ewc_loss_diag": 8.195638656616211e-07, + "ewc_loss_parallel": 8.344650268554688e-07, + "grad_norm": 3.492769479751587, + "learning_rate": 1.22509537939805e-07, + "loss": 0.51, + "mean_token_accuracy": 0.8341008424758911, + "num_tokens": 11275612.0, + "step": 290 + }, + { + "epoch": 0.03701819106983844, + "ewc_loss": 0.00090789794921875, + "ewc_loss_diag": 8.23289155960083e-07, + "ewc_loss_parallel": 8.23289155960083e-07, + "grad_norm": 3.3408992290496826, + "learning_rate": 1.2293344637558286e-07, + "loss": 0.5176, + "mean_token_accuracy": 0.834938645362854, + "num_tokens": 11321574.0, + "step": 291 + }, + { + "epoch": 0.037145401348428954, + "ewc_loss": 0.0009002685546875, + "ewc_loss_diag": 8.23289155960083e-07, + "ewc_loss_parallel": 8.158385753631592e-07, + "grad_norm": 3.4724693298339844, + "learning_rate": 1.2335735481136073e-07, + "loss": 0.491, + "mean_token_accuracy": 0.8395577669143677, + "num_tokens": 11362134.0, + "step": 292 + }, + { + "epoch": 0.03727261162701946, + "ewc_loss": 0.000911712646484375, + "ewc_loss_diag": 8.23289155960083e-07, + "ewc_loss_parallel": 8.270144462585449e-07, + "grad_norm": 3.4438321590423584, + "learning_rate": 1.237812632471386e-07, + "loss": 0.493, + "mean_token_accuracy": 0.8430243730545044, + "num_tokens": 11403024.0, + "step": 293 + }, + { + "epoch": 0.037399821905609976, + "ewc_loss": 0.000911712646484375, + "ewc_loss_diag": 8.381903171539307e-07, + "ewc_loss_parallel": 8.270144462585449e-07, + "grad_norm": 3.471018075942993, + "learning_rate": 1.2420517168291648e-07, + "loss": 0.5168, + "mean_token_accuracy": 0.8318375945091248, + "num_tokens": 11444215.0, + "step": 294 + }, + { + "epoch": 0.03752703218420048, + "ewc_loss": 0.00090789794921875, + "ewc_loss_diag": 8.344650268554688e-07, + "ewc_loss_parallel": 8.23289155960083e-07, + "grad_norm": 3.559561252593994, + "learning_rate": 1.2462908011869435e-07, + "loss": 0.5146, + "mean_token_accuracy": 0.8354125618934631, + "num_tokens": 11481100.0, + "step": 295 + }, + { + "epoch": 0.03765424246279099, + "ewc_loss": 0.000919342041015625, + "ewc_loss_diag": 8.419156074523926e-07, + "ewc_loss_parallel": 8.344650268554688e-07, + "grad_norm": 3.449310302734375, + "learning_rate": 1.2505298855447223e-07, + "loss": 0.4475, + "mean_token_accuracy": 0.8574227094650269, + "num_tokens": 11520037.0, + "step": 296 + }, + { + "epoch": 0.037781452741381505, + "ewc_loss": 0.00092315673828125, + "ewc_loss_diag": 8.530914783477783e-07, + "ewc_loss_parallel": 8.381903171539307e-07, + "grad_norm": 3.694479465484619, + "learning_rate": 1.254768969902501e-07, + "loss": 0.5761, + "mean_token_accuracy": 0.8193695545196533, + "num_tokens": 11556821.0, + "step": 297 + }, + { + "epoch": 0.03790866301997201, + "ewc_loss": 0.0009307861328125, + "ewc_loss_diag": 8.717179298400879e-07, + "ewc_loss_parallel": 8.419156074523926e-07, + "grad_norm": 3.9265613555908203, + "learning_rate": 1.2590080542602797e-07, + "loss": 0.534, + "mean_token_accuracy": 0.8307584524154663, + "num_tokens": 11588478.0, + "step": 298 + }, + { + "epoch": 0.03803587329856253, + "ewc_loss": 0.000949859619140625, + "ewc_loss_diag": 8.754432201385498e-07, + "ewc_loss_parallel": 8.605420589447021e-07, + "grad_norm": 3.5988779067993164, + "learning_rate": 1.2632471386180584e-07, + "loss": 0.5227, + "mean_token_accuracy": 0.8302127122879028, + "num_tokens": 11627536.0, + "step": 299 + }, + { + "epoch": 0.038163083577153034, + "ewc_loss": 0.000946044921875, + "ewc_loss_diag": 8.828938007354736e-07, + "ewc_loss_parallel": 8.568167686462402e-07, + "grad_norm": 3.670030117034912, + "learning_rate": 1.2674862229758372e-07, + "loss": 0.5061, + "mean_token_accuracy": 0.8411861062049866, + "num_tokens": 11663335.0, + "step": 300 + }, + { + "epoch": 0.03829029385574354, + "ewc_loss": 0.000946044921875, + "ewc_loss_diag": 8.828938007354736e-07, + "ewc_loss_parallel": 8.568167686462402e-07, + "grad_norm": 3.472883939743042, + "learning_rate": 1.271725307333616e-07, + "loss": 0.4774, + "mean_token_accuracy": 0.8469760417938232, + "num_tokens": 11705037.0, + "step": 301 + }, + { + "epoch": 0.038417504134334056, + "ewc_loss": 0.0009307861328125, + "ewc_loss_diag": 8.866190910339355e-07, + "ewc_loss_parallel": 8.419156074523926e-07, + "grad_norm": 3.372154951095581, + "learning_rate": 1.2759643916913946e-07, + "loss": 0.4964, + "mean_token_accuracy": 0.8402496576309204, + "num_tokens": 11749198.0, + "step": 302 + }, + { + "epoch": 0.03854471441292456, + "ewc_loss": 0.00092315673828125, + "ewc_loss_diag": 8.940696716308594e-07, + "ewc_loss_parallel": 8.344650268554688e-07, + "grad_norm": 3.7600135803222656, + "learning_rate": 1.2802034760491733e-07, + "loss": 0.5015, + "mean_token_accuracy": 0.8385911583900452, + "num_tokens": 11782908.0, + "step": 303 + }, + { + "epoch": 0.03867192469151508, + "ewc_loss": 0.000949859619140625, + "ewc_loss_diag": 9.015202522277832e-07, + "ewc_loss_parallel": 8.568167686462402e-07, + "grad_norm": 3.5902469158172607, + "learning_rate": 1.284442560406952e-07, + "loss": 0.558, + "mean_token_accuracy": 0.8228708505630493, + "num_tokens": 11825898.0, + "step": 304 + }, + { + "epoch": 0.038799134970105585, + "ewc_loss": 0.000946044921875, + "ewc_loss_diag": 9.052455425262451e-07, + "ewc_loss_parallel": 8.530914783477783e-07, + "grad_norm": 3.8369369506835938, + "learning_rate": 1.2886816447647308e-07, + "loss": 0.582, + "mean_token_accuracy": 0.8127864003181458, + "num_tokens": 11862186.0, + "step": 305 + }, + { + "epoch": 0.03892634524869609, + "ewc_loss": 0.000965118408203125, + "ewc_loss_diag": 9.052455425262451e-07, + "ewc_loss_parallel": 8.717179298400879e-07, + "grad_norm": 3.6662003993988037, + "learning_rate": 1.2929207291225095e-07, + "loss": 0.5078, + "mean_token_accuracy": 0.8413649797439575, + "num_tokens": 11899119.0, + "step": 306 + }, + { + "epoch": 0.03905355552728661, + "ewc_loss": 0.000957489013671875, + "ewc_loss_diag": 9.126961231231689e-07, + "ewc_loss_parallel": 8.642673492431641e-07, + "grad_norm": 3.608558177947998, + "learning_rate": 1.2971598134802882e-07, + "loss": 0.506, + "mean_token_accuracy": 0.8387517333030701, + "num_tokens": 11937470.0, + "step": 307 + }, + { + "epoch": 0.039180765805877114, + "ewc_loss": 0.000965118408203125, + "ewc_loss_diag": 9.201467037200928e-07, + "ewc_loss_parallel": 8.717179298400879e-07, + "grad_norm": 3.72753643989563, + "learning_rate": 1.301398897838067e-07, + "loss": 0.5178, + "mean_token_accuracy": 0.8342729210853577, + "num_tokens": 11974463.0, + "step": 308 + }, + { + "epoch": 0.03930797608446762, + "ewc_loss": 0.0009613037109375, + "ewc_loss_diag": 9.275972843170166e-07, + "ewc_loss_parallel": 8.67992639541626e-07, + "grad_norm": 3.5585153102874756, + "learning_rate": 1.3056379821958457e-07, + "loss": 0.525, + "mean_token_accuracy": 0.8356300592422485, + "num_tokens": 12016064.0, + "step": 309 + }, + { + "epoch": 0.039435186363058136, + "ewc_loss": 0.000949859619140625, + "ewc_loss_diag": 9.275972843170166e-07, + "ewc_loss_parallel": 8.568167686462402e-07, + "grad_norm": 3.496204137802124, + "learning_rate": 1.3098770665536244e-07, + "loss": 0.4878, + "mean_token_accuracy": 0.8444457054138184, + "num_tokens": 12056482.0, + "step": 310 + }, + { + "epoch": 0.039562396641648644, + "ewc_loss": 0.000957489013671875, + "ewc_loss_diag": 9.275972843170166e-07, + "ewc_loss_parallel": 8.642673492431641e-07, + "grad_norm": 4.0925211906433105, + "learning_rate": 1.3141161509114031e-07, + "loss": 0.4833, + "mean_token_accuracy": 0.8421359062194824, + "num_tokens": 12089281.0, + "step": 311 + }, + { + "epoch": 0.03968960692023916, + "ewc_loss": 0.00098419189453125, + "ewc_loss_diag": 9.387731552124023e-07, + "ewc_loss_parallel": 8.940696716308594e-07, + "grad_norm": 3.7150979042053223, + "learning_rate": 1.3183552352691819e-07, + "loss": 0.4909, + "mean_token_accuracy": 0.8461716771125793, + "num_tokens": 12123741.0, + "step": 312 + }, + { + "epoch": 0.039816817198829665, + "ewc_loss": 0.0009765625, + "ewc_loss_diag": 9.424984455108643e-07, + "ewc_loss_parallel": 8.866190910339355e-07, + "grad_norm": 3.582456111907959, + "learning_rate": 1.3225943196269603e-07, + "loss": 0.5127, + "mean_token_accuracy": 0.8395991325378418, + "num_tokens": 12165197.0, + "step": 313 + }, + { + "epoch": 0.03994402747742017, + "ewc_loss": 0.00096893310546875, + "ewc_loss_diag": 9.424984455108643e-07, + "ewc_loss_parallel": 8.754432201385498e-07, + "grad_norm": 3.563283681869507, + "learning_rate": 1.3268334039847393e-07, + "loss": 0.4931, + "mean_token_accuracy": 0.8427180647850037, + "num_tokens": 12203040.0, + "step": 314 + }, + { + "epoch": 0.04007123775601069, + "ewc_loss": 0.00096893310546875, + "ewc_loss_diag": 9.424984455108643e-07, + "ewc_loss_parallel": 8.754432201385498e-07, + "grad_norm": 3.8480775356292725, + "learning_rate": 1.3310724883425178e-07, + "loss": 0.5479, + "mean_token_accuracy": 0.8229357004165649, + "num_tokens": 12239018.0, + "step": 315 + }, + { + "epoch": 0.040198448034601195, + "ewc_loss": 0.00098419189453125, + "ewc_loss_diag": 9.462237358093262e-07, + "ewc_loss_parallel": 8.940696716308594e-07, + "grad_norm": 3.4859066009521484, + "learning_rate": 1.3353115727002968e-07, + "loss": 0.5169, + "mean_token_accuracy": 0.8351192474365234, + "num_tokens": 12283220.0, + "step": 316 + }, + { + "epoch": 0.04032565831319171, + "ewc_loss": 0.00096893310546875, + "ewc_loss_diag": 9.462237358093262e-07, + "ewc_loss_parallel": 8.754432201385498e-07, + "grad_norm": 3.472202777862549, + "learning_rate": 1.3395506570580752e-07, + "loss": 0.5016, + "mean_token_accuracy": 0.8398762345314026, + "num_tokens": 12326409.0, + "step": 317 + }, + { + "epoch": 0.040452868591782216, + "ewc_loss": 0.0009613037109375, + "ewc_loss_diag": 9.424984455108643e-07, + "ewc_loss_parallel": 8.642673492431641e-07, + "grad_norm": 4.155890941619873, + "learning_rate": 1.3437897414158542e-07, + "loss": 0.5485, + "mean_token_accuracy": 0.823409378528595, + "num_tokens": 12355065.0, + "step": 318 + }, + { + "epoch": 0.040580078870372724, + "ewc_loss": 0.0010223388671875, + "ewc_loss_diag": 9.424984455108643e-07, + "ewc_loss_parallel": 9.275972843170166e-07, + "grad_norm": 3.8833720684051514, + "learning_rate": 1.3480288257736327e-07, + "loss": 0.492, + "mean_token_accuracy": 0.8427355885505676, + "num_tokens": 12388931.0, + "step": 319 + }, + { + "epoch": 0.04070728914896324, + "ewc_loss": 0.0010223388671875, + "ewc_loss_diag": 9.424984455108643e-07, + "ewc_loss_parallel": 9.313225746154785e-07, + "grad_norm": 3.5265612602233887, + "learning_rate": 1.3522679101314117e-07, + "loss": 0.4544, + "mean_token_accuracy": 0.8511326313018799, + "num_tokens": 12426415.0, + "step": 320 + }, + { + "epoch": 0.040834499427553746, + "ewc_loss": 0.00098419189453125, + "ewc_loss_diag": 9.499490261077881e-07, + "ewc_loss_parallel": 8.940696716308594e-07, + "grad_norm": 3.606318950653076, + "learning_rate": 1.35650699448919e-07, + "loss": 0.5001, + "mean_token_accuracy": 0.8416547775268555, + "num_tokens": 12465501.0, + "step": 321 + }, + { + "epoch": 0.04096170970614425, + "ewc_loss": 0.00098419189453125, + "ewc_loss_diag": 9.499490261077881e-07, + "ewc_loss_parallel": 8.940696716308594e-07, + "grad_norm": 3.6358511447906494, + "learning_rate": 1.360746078846969e-07, + "loss": 0.481, + "mean_token_accuracy": 0.8437173366546631, + "num_tokens": 12503504.0, + "step": 322 + }, + { + "epoch": 0.04108891998473477, + "ewc_loss": 0.0009918212890625, + "ewc_loss_diag": 9.499490261077881e-07, + "ewc_loss_parallel": 9.015202522277832e-07, + "grad_norm": 3.643712043762207, + "learning_rate": 1.3649851632047476e-07, + "loss": 0.5522, + "mean_token_accuracy": 0.8254066705703735, + "num_tokens": 12544947.0, + "step": 323 + }, + { + "epoch": 0.041216130263325275, + "ewc_loss": 0.0009918212890625, + "ewc_loss_diag": 9.685754776000977e-07, + "ewc_loss_parallel": 8.940696716308594e-07, + "grad_norm": 3.535353422164917, + "learning_rate": 1.3692242475625266e-07, + "loss": 0.4579, + "mean_token_accuracy": 0.8550002574920654, + "num_tokens": 12585410.0, + "step": 324 + }, + { + "epoch": 0.04134334054191579, + "ewc_loss": 0.00099945068359375, + "ewc_loss_diag": 9.685754776000977e-07, + "ewc_loss_parallel": 9.015202522277832e-07, + "grad_norm": 3.5598320960998535, + "learning_rate": 1.373463331920305e-07, + "loss": 0.5071, + "mean_token_accuracy": 0.8429757952690125, + "num_tokens": 12625439.0, + "step": 325 + }, + { + "epoch": 0.0414705508205063, + "ewc_loss": 0.00099945068359375, + "ewc_loss_diag": 9.685754776000977e-07, + "ewc_loss_parallel": 9.015202522277832e-07, + "grad_norm": 3.7536845207214355, + "learning_rate": 1.377702416278084e-07, + "loss": 0.5801, + "mean_token_accuracy": 0.8171820640563965, + "num_tokens": 12665178.0, + "step": 326 + }, + { + "epoch": 0.041597761099096804, + "ewc_loss": 0.0010223388671875, + "ewc_loss_diag": 9.685754776000977e-07, + "ewc_loss_parallel": 9.238719940185547e-07, + "grad_norm": 3.47916579246521, + "learning_rate": 1.3819415006358625e-07, + "loss": 0.4432, + "mean_token_accuracy": 0.8547600507736206, + "num_tokens": 12706673.0, + "step": 327 + }, + { + "epoch": 0.04172497137768732, + "ewc_loss": 0.00099945068359375, + "ewc_loss_diag": 9.834766387939453e-07, + "ewc_loss_parallel": 9.052455425262451e-07, + "grad_norm": 3.3835408687591553, + "learning_rate": 1.3861805849936415e-07, + "loss": 0.4785, + "mean_token_accuracy": 0.8491976261138916, + "num_tokens": 12751678.0, + "step": 328 + }, + { + "epoch": 0.041852181656277826, + "ewc_loss": 0.00099945068359375, + "ewc_loss_diag": 9.834766387939453e-07, + "ewc_loss_parallel": 9.015202522277832e-07, + "grad_norm": 3.716442823410034, + "learning_rate": 1.39041966935142e-07, + "loss": 0.5059, + "mean_token_accuracy": 0.8364790678024292, + "num_tokens": 12788413.0, + "step": 329 + }, + { + "epoch": 0.04197939193486834, + "ewc_loss": 0.00101470947265625, + "ewc_loss_diag": 9.834766387939453e-07, + "ewc_loss_parallel": 9.164214134216309e-07, + "grad_norm": 3.6061007976531982, + "learning_rate": 1.394658753709199e-07, + "loss": 0.5004, + "mean_token_accuracy": 0.8389368057250977, + "num_tokens": 12830796.0, + "step": 330 + }, + { + "epoch": 0.04210660221345885, + "ewc_loss": 0.00101470947265625, + "ewc_loss_diag": 9.834766387939453e-07, + "ewc_loss_parallel": 9.164214134216309e-07, + "grad_norm": 3.7143499851226807, + "learning_rate": 1.3988978380669774e-07, + "loss": 0.5204, + "mean_token_accuracy": 0.8344316482543945, + "num_tokens": 12866776.0, + "step": 331 + }, + { + "epoch": 0.042233812492049355, + "ewc_loss": 0.0010223388671875, + "ewc_loss_diag": 9.834766387939453e-07, + "ewc_loss_parallel": 9.238719940185547e-07, + "grad_norm": 3.609858512878418, + "learning_rate": 1.403136922424756e-07, + "loss": 0.5298, + "mean_token_accuracy": 0.8304008841514587, + "num_tokens": 12909420.0, + "step": 332 + }, + { + "epoch": 0.04236102277063987, + "ewc_loss": 0.0010223388671875, + "ewc_loss_diag": 9.909272193908691e-07, + "ewc_loss_parallel": 9.238719940185547e-07, + "grad_norm": 3.6219820976257324, + "learning_rate": 1.4073760067825348e-07, + "loss": 0.4991, + "mean_token_accuracy": 0.8408874273300171, + "num_tokens": 12947922.0, + "step": 333 + }, + { + "epoch": 0.04248823304923038, + "ewc_loss": 0.00103759765625, + "ewc_loss_diag": 9.98377799987793e-07, + "ewc_loss_parallel": 9.387731552124023e-07, + "grad_norm": 3.545851469039917, + "learning_rate": 1.4116150911403136e-07, + "loss": 0.5236, + "mean_token_accuracy": 0.8331338763237, + "num_tokens": 12990181.0, + "step": 334 + }, + { + "epoch": 0.04261544332782089, + "ewc_loss": 0.00103759765625, + "ewc_loss_diag": 1.0058283805847168e-06, + "ewc_loss_parallel": 9.387731552124023e-07, + "grad_norm": 3.6932153701782227, + "learning_rate": 1.4158541754980923e-07, + "loss": 0.4685, + "mean_token_accuracy": 0.8480479121208191, + "num_tokens": 13027579.0, + "step": 335 + }, + { + "epoch": 0.0427426536064114, + "ewc_loss": 0.00104522705078125, + "ewc_loss_diag": 1.0132789611816406e-06, + "ewc_loss_parallel": 9.462237358093262e-07, + "grad_norm": 3.4543628692626953, + "learning_rate": 1.420093259855871e-07, + "loss": 0.4881, + "mean_token_accuracy": 0.8484666347503662, + "num_tokens": 13070634.0, + "step": 336 + }, + { + "epoch": 0.042869863885001906, + "ewc_loss": 0.00102996826171875, + "ewc_loss_diag": 1.0132789611816406e-06, + "ewc_loss_parallel": 9.313225746154785e-07, + "grad_norm": 3.8454532623291016, + "learning_rate": 1.4243323442136497e-07, + "loss": 0.5255, + "mean_token_accuracy": 0.8333626985549927, + "num_tokens": 13104952.0, + "step": 337 + }, + { + "epoch": 0.04299707416359242, + "ewc_loss": 0.001068115234375, + "ewc_loss_diag": 1.0281801223754883e-06, + "ewc_loss_parallel": 9.611248970031738e-07, + "grad_norm": 3.798233985900879, + "learning_rate": 1.4285714285714285e-07, + "loss": 0.5063, + "mean_token_accuracy": 0.8385630249977112, + "num_tokens": 13139352.0, + "step": 338 + }, + { + "epoch": 0.04312428444218293, + "ewc_loss": 0.00106048583984375, + "ewc_loss_diag": 1.0356307029724121e-06, + "ewc_loss_parallel": 9.5367431640625e-07, + "grad_norm": 3.671539783477783, + "learning_rate": 1.4328105129292072e-07, + "loss": 0.4971, + "mean_token_accuracy": 0.8422337174415588, + "num_tokens": 13179741.0, + "step": 339 + }, + { + "epoch": 0.043251494720773435, + "ewc_loss": 0.001068115234375, + "ewc_loss_diag": 1.0356307029724121e-06, + "ewc_loss_parallel": 9.611248970031738e-07, + "grad_norm": 3.5513319969177246, + "learning_rate": 1.437049597286986e-07, + "loss": 0.4792, + "mean_token_accuracy": 0.8464564085006714, + "num_tokens": 13220123.0, + "step": 340 + }, + { + "epoch": 0.04337870499936395, + "ewc_loss": 0.0010528564453125, + "ewc_loss_diag": 1.0356307029724121e-06, + "ewc_loss_parallel": 9.462237358093262e-07, + "grad_norm": 3.9022555351257324, + "learning_rate": 1.4412886816447646e-07, + "loss": 0.531, + "mean_token_accuracy": 0.8322685360908508, + "num_tokens": 13254540.0, + "step": 341 + }, + { + "epoch": 0.04350591527795446, + "ewc_loss": 0.00107574462890625, + "ewc_loss_diag": 1.043081283569336e-06, + "ewc_loss_parallel": 9.685754776000977e-07, + "grad_norm": 3.874501943588257, + "learning_rate": 1.4455277660025434e-07, + "loss": 0.46, + "mean_token_accuracy": 0.8503576517105103, + "num_tokens": 13287542.0, + "step": 342 + }, + { + "epoch": 0.04363312555654497, + "ewc_loss": 0.00107574462890625, + "ewc_loss_diag": 1.043081283569336e-06, + "ewc_loss_parallel": 9.685754776000977e-07, + "grad_norm": 3.6048898696899414, + "learning_rate": 1.449766850360322e-07, + "loss": 0.5118, + "mean_token_accuracy": 0.8373422622680664, + "num_tokens": 13329808.0, + "step": 343 + }, + { + "epoch": 0.04376033583513548, + "ewc_loss": 0.0010528564453125, + "ewc_loss_diag": 1.043081283569336e-06, + "ewc_loss_parallel": 9.499490261077881e-07, + "grad_norm": 3.8950612545013428, + "learning_rate": 1.4540059347181008e-07, + "loss": 0.4959, + "mean_token_accuracy": 0.8418536186218262, + "num_tokens": 13360173.0, + "step": 344 + }, + { + "epoch": 0.043887546113725986, + "ewc_loss": 0.0010833740234375, + "ewc_loss_diag": 1.0505318641662598e-06, + "ewc_loss_parallel": 9.760260581970215e-07, + "grad_norm": 3.7404251098632812, + "learning_rate": 1.4582450190758795e-07, + "loss": 0.5362, + "mean_token_accuracy": 0.8267394304275513, + "num_tokens": 13404241.0, + "step": 345 + }, + { + "epoch": 0.0440147563923165, + "ewc_loss": 0.0010833740234375, + "ewc_loss_diag": 1.0579824447631836e-06, + "ewc_loss_parallel": 9.760260581970215e-07, + "grad_norm": 3.7618930339813232, + "learning_rate": 1.4624841034336583e-07, + "loss": 0.4902, + "mean_token_accuracy": 0.8425068259239197, + "num_tokens": 13441496.0, + "step": 346 + }, + { + "epoch": 0.04414196667090701, + "ewc_loss": 0.0010833740234375, + "ewc_loss_diag": 1.0505318641662598e-06, + "ewc_loss_parallel": 9.760260581970215e-07, + "grad_norm": 3.542884588241577, + "learning_rate": 1.466723187791437e-07, + "loss": 0.4595, + "mean_token_accuracy": 0.8510885238647461, + "num_tokens": 13484246.0, + "step": 347 + }, + { + "epoch": 0.04426917694949752, + "ewc_loss": 0.0010528564453125, + "ewc_loss_diag": 1.0579824447631836e-06, + "ewc_loss_parallel": 9.462237358093262e-07, + "grad_norm": 3.5828182697296143, + "learning_rate": 1.4709622721492157e-07, + "loss": 0.487, + "mean_token_accuracy": 0.8389452695846558, + "num_tokens": 13523163.0, + "step": 348 + }, + { + "epoch": 0.04439638722808803, + "ewc_loss": 0.00106048583984375, + "ewc_loss_diag": 1.0728836059570312e-06, + "ewc_loss_parallel": 9.499490261077881e-07, + "grad_norm": 3.784395694732666, + "learning_rate": 1.4752013565069942e-07, + "loss": 0.4886, + "mean_token_accuracy": 0.8417184352874756, + "num_tokens": 13559059.0, + "step": 349 + }, + { + "epoch": 0.04452359750667854, + "ewc_loss": 0.0010833740234375, + "ewc_loss_diag": 1.0728836059570312e-06, + "ewc_loss_parallel": 9.760260581970215e-07, + "grad_norm": 3.7845826148986816, + "learning_rate": 1.4794404408647732e-07, + "loss": 0.4905, + "mean_token_accuracy": 0.8426799178123474, + "num_tokens": 13595706.0, + "step": 350 + }, + { + "epoch": 0.04465080778526905, + "ewc_loss": 0.0010986328125, + "ewc_loss_diag": 1.080334186553955e-06, + "ewc_loss_parallel": 9.909272193908691e-07, + "grad_norm": 3.491347074508667, + "learning_rate": 1.4836795252225516e-07, + "loss": 0.4462, + "mean_token_accuracy": 0.8523229360580444, + "num_tokens": 13637586.0, + "step": 351 + }, + { + "epoch": 0.04477801806385956, + "ewc_loss": 0.001068115234375, + "ewc_loss_diag": 1.080334186553955e-06, + "ewc_loss_parallel": 9.611248970031738e-07, + "grad_norm": 3.6900417804718018, + "learning_rate": 1.4879186095803306e-07, + "loss": 0.4284, + "mean_token_accuracy": 0.8572669625282288, + "num_tokens": 13675661.0, + "step": 352 + }, + { + "epoch": 0.04490522834245007, + "ewc_loss": 0.00107574462890625, + "ewc_loss_diag": 1.080334186553955e-06, + "ewc_loss_parallel": 9.685754776000977e-07, + "grad_norm": 3.8108596801757812, + "learning_rate": 1.492157693938109e-07, + "loss": 0.4856, + "mean_token_accuracy": 0.8436234593391418, + "num_tokens": 13713358.0, + "step": 353 + }, + { + "epoch": 0.04503243862104058, + "ewc_loss": 0.0010986328125, + "ewc_loss_diag": 1.080334186553955e-06, + "ewc_loss_parallel": 9.909272193908691e-07, + "grad_norm": 3.5331480503082275, + "learning_rate": 1.496396778295888e-07, + "loss": 0.4504, + "mean_token_accuracy": 0.8566924333572388, + "num_tokens": 13756814.0, + "step": 354 + }, + { + "epoch": 0.04515964889963109, + "ewc_loss": 0.00107574462890625, + "ewc_loss_diag": 1.080334186553955e-06, + "ewc_loss_parallel": 9.685754776000977e-07, + "grad_norm": 3.8043506145477295, + "learning_rate": 1.5006358626536665e-07, + "loss": 0.5249, + "mean_token_accuracy": 0.833698034286499, + "num_tokens": 13793137.0, + "step": 355 + }, + { + "epoch": 0.0452868591782216, + "ewc_loss": 0.0010986328125, + "ewc_loss_diag": 1.080334186553955e-06, + "ewc_loss_parallel": 9.909272193908691e-07, + "grad_norm": 3.679568290710449, + "learning_rate": 1.5048749470114455e-07, + "loss": 0.5208, + "mean_token_accuracy": 0.8337980508804321, + "num_tokens": 13834371.0, + "step": 356 + }, + { + "epoch": 0.04541406945681211, + "ewc_loss": 0.00110626220703125, + "ewc_loss_diag": 1.080334186553955e-06, + "ewc_loss_parallel": 9.98377799987793e-07, + "grad_norm": 3.5693297386169434, + "learning_rate": 1.509114031369224e-07, + "loss": 0.4502, + "mean_token_accuracy": 0.8534172177314758, + "num_tokens": 13875083.0, + "step": 357 + }, + { + "epoch": 0.04554127973540262, + "ewc_loss": 0.00109100341796875, + "ewc_loss_diag": 1.080334186553955e-06, + "ewc_loss_parallel": 9.834766387939453e-07, + "grad_norm": 3.7111902236938477, + "learning_rate": 1.513353115727003e-07, + "loss": 0.506, + "mean_token_accuracy": 0.8345836997032166, + "num_tokens": 13913132.0, + "step": 358 + }, + { + "epoch": 0.04566849001399313, + "ewc_loss": 0.00110626220703125, + "ewc_loss_diag": 1.087784767150879e-06, + "ewc_loss_parallel": 9.98377799987793e-07, + "grad_norm": 3.6766107082366943, + "learning_rate": 1.5175922000847814e-07, + "loss": 0.4603, + "mean_token_accuracy": 0.8523820638656616, + "num_tokens": 13952783.0, + "step": 359 + }, + { + "epoch": 0.04579570029258364, + "ewc_loss": 0.0011138916015625, + "ewc_loss_diag": 1.087784767150879e-06, + "ewc_loss_parallel": 1.0058283805847168e-06, + "grad_norm": 3.5885426998138428, + "learning_rate": 1.5218312844425604e-07, + "loss": 0.4856, + "mean_token_accuracy": 0.8375035524368286, + "num_tokens": 13994963.0, + "step": 360 + }, + { + "epoch": 0.045922910571174154, + "ewc_loss": 0.00110626220703125, + "ewc_loss_diag": 1.087784767150879e-06, + "ewc_loss_parallel": 9.98377799987793e-07, + "grad_norm": 3.647913932800293, + "learning_rate": 1.526070368800339e-07, + "loss": 0.429, + "mean_token_accuracy": 0.8586368560791016, + "num_tokens": 14030745.0, + "step": 361 + }, + { + "epoch": 0.04605012084976466, + "ewc_loss": 0.0011138916015625, + "ewc_loss_diag": 1.0952353477478027e-06, + "ewc_loss_parallel": 1.0058283805847168e-06, + "grad_norm": 3.560420513153076, + "learning_rate": 1.530309453158118e-07, + "loss": 0.4405, + "mean_token_accuracy": 0.8583661913871765, + "num_tokens": 14073044.0, + "step": 362 + }, + { + "epoch": 0.04617733112835517, + "ewc_loss": 0.00110626220703125, + "ewc_loss_diag": 1.0952353477478027e-06, + "ewc_loss_parallel": 9.98377799987793e-07, + "grad_norm": 3.834416151046753, + "learning_rate": 1.5345485375158963e-07, + "loss": 0.523, + "mean_token_accuracy": 0.8339343070983887, + "num_tokens": 14112006.0, + "step": 363 + }, + { + "epoch": 0.04630454140694568, + "ewc_loss": 0.0011138916015625, + "ewc_loss_diag": 1.0952353477478027e-06, + "ewc_loss_parallel": 1.0058283805847168e-06, + "grad_norm": 3.3520727157592773, + "learning_rate": 1.5387876218736753e-07, + "loss": 0.4042, + "mean_token_accuracy": 0.8685063123703003, + "num_tokens": 14156207.0, + "step": 364 + }, + { + "epoch": 0.04643175168553619, + "ewc_loss": 0.00109100341796875, + "ewc_loss_diag": 1.0952353477478027e-06, + "ewc_loss_parallel": 9.834766387939453e-07, + "grad_norm": 3.695834159851074, + "learning_rate": 1.5430267062314538e-07, + "loss": 0.5047, + "mean_token_accuracy": 0.8400265574455261, + "num_tokens": 14196167.0, + "step": 365 + }, + { + "epoch": 0.0465589619641267, + "ewc_loss": 0.0011138916015625, + "ewc_loss_diag": 1.0952353477478027e-06, + "ewc_loss_parallel": 1.0058283805847168e-06, + "grad_norm": 3.9514000415802, + "learning_rate": 1.5472657905892328e-07, + "loss": 0.5068, + "mean_token_accuracy": 0.8383995294570923, + "num_tokens": 14232702.0, + "step": 366 + }, + { + "epoch": 0.04668617224271721, + "ewc_loss": 0.00113677978515625, + "ewc_loss_diag": 1.0952353477478027e-06, + "ewc_loss_parallel": 1.0281801223754883e-06, + "grad_norm": 3.935946464538574, + "learning_rate": 1.5515048749470113e-07, + "loss": 0.509, + "mean_token_accuracy": 0.8342475891113281, + "num_tokens": 14266627.0, + "step": 367 + }, + { + "epoch": 0.04681338252130772, + "ewc_loss": 0.001129150390625, + "ewc_loss_diag": 1.0952353477478027e-06, + "ewc_loss_parallel": 1.0207295417785645e-06, + "grad_norm": 3.788057565689087, + "learning_rate": 1.55574395930479e-07, + "loss": 0.5051, + "mean_token_accuracy": 0.8378737568855286, + "num_tokens": 14305113.0, + "step": 368 + }, + { + "epoch": 0.046940592799898234, + "ewc_loss": 0.0011138916015625, + "ewc_loss_diag": 1.0952353477478027e-06, + "ewc_loss_parallel": 1.0058283805847168e-06, + "grad_norm": 3.598259210586548, + "learning_rate": 1.5599830436625687e-07, + "loss": 0.4993, + "mean_token_accuracy": 0.8404502868652344, + "num_tokens": 14346396.0, + "step": 369 + }, + { + "epoch": 0.04706780307848874, + "ewc_loss": 0.0011138916015625, + "ewc_loss_diag": 1.0952353477478027e-06, + "ewc_loss_parallel": 1.0058283805847168e-06, + "grad_norm": 3.682872772216797, + "learning_rate": 1.5642221280203474e-07, + "loss": 0.5007, + "mean_token_accuracy": 0.8402897119522095, + "num_tokens": 14385930.0, + "step": 370 + }, + { + "epoch": 0.04719501335707925, + "ewc_loss": 0.00112152099609375, + "ewc_loss_diag": 1.0952353477478027e-06, + "ewc_loss_parallel": 1.0132789611816406e-06, + "grad_norm": 3.7360193729400635, + "learning_rate": 1.5684612123781262e-07, + "loss": 0.435, + "mean_token_accuracy": 0.8591440916061401, + "num_tokens": 14421019.0, + "step": 371 + }, + { + "epoch": 0.04732222363566976, + "ewc_loss": 0.00113677978515625, + "ewc_loss_diag": 1.1026859283447266e-06, + "ewc_loss_parallel": 1.0281801223754883e-06, + "grad_norm": 3.831270933151245, + "learning_rate": 1.572700296735905e-07, + "loss": 0.4951, + "mean_token_accuracy": 0.8446890711784363, + "num_tokens": 14457636.0, + "step": 372 + }, + { + "epoch": 0.04744943391426027, + "ewc_loss": 0.001129150390625, + "ewc_loss_diag": 1.1026859283447266e-06, + "ewc_loss_parallel": 1.0207295417785645e-06, + "grad_norm": 4.187805652618408, + "learning_rate": 1.576939381093684e-07, + "loss": 0.5084, + "mean_token_accuracy": 0.8345235586166382, + "num_tokens": 14487183.0, + "step": 373 + }, + { + "epoch": 0.047576644192850785, + "ewc_loss": 0.0011749267578125, + "ewc_loss_diag": 1.1026859283447266e-06, + "ewc_loss_parallel": 1.0654330253601074e-06, + "grad_norm": 3.7536094188690186, + "learning_rate": 1.5811784654514623e-07, + "loss": 0.4295, + "mean_token_accuracy": 0.8624488115310669, + "num_tokens": 14521554.0, + "step": 374 + }, + { + "epoch": 0.04770385447144129, + "ewc_loss": 0.0011444091796875, + "ewc_loss_diag": 1.1101365089416504e-06, + "ewc_loss_parallel": 1.0281801223754883e-06, + "grad_norm": 3.6639397144317627, + "learning_rate": 1.5854175498092413e-07, + "loss": 0.5458, + "mean_token_accuracy": 0.8238439559936523, + "num_tokens": 14566196.0, + "step": 375 + }, + { + "epoch": 0.0478310647500318, + "ewc_loss": 0.0011444091796875, + "ewc_loss_diag": 1.1101365089416504e-06, + "ewc_loss_parallel": 1.0281801223754883e-06, + "grad_norm": 3.818878650665283, + "learning_rate": 1.5896566341670198e-07, + "loss": 0.4895, + "mean_token_accuracy": 0.8418817520141602, + "num_tokens": 14606064.0, + "step": 376 + }, + { + "epoch": 0.047958275028622314, + "ewc_loss": 0.00115966796875, + "ewc_loss_diag": 1.1101365089416504e-06, + "ewc_loss_parallel": 1.043081283569336e-06, + "grad_norm": 3.8468971252441406, + "learning_rate": 1.5938957185247988e-07, + "loss": 0.5276, + "mean_token_accuracy": 0.8298799991607666, + "num_tokens": 14645636.0, + "step": 377 + }, + { + "epoch": 0.04808548530721282, + "ewc_loss": 0.00115966796875, + "ewc_loss_diag": 1.1101365089416504e-06, + "ewc_loss_parallel": 1.043081283569336e-06, + "grad_norm": 3.866426944732666, + "learning_rate": 1.5981348028825772e-07, + "loss": 0.514, + "mean_token_accuracy": 0.8355537056922913, + "num_tokens": 14682125.0, + "step": 378 + }, + { + "epoch": 0.048212695585803336, + "ewc_loss": 0.00115966796875, + "ewc_loss_diag": 1.1101365089416504e-06, + "ewc_loss_parallel": 1.043081283569336e-06, + "grad_norm": 4.122020244598389, + "learning_rate": 1.6023738872403562e-07, + "loss": 0.5818, + "mean_token_accuracy": 0.8116506338119507, + "num_tokens": 14714921.0, + "step": 379 + }, + { + "epoch": 0.048339905864393844, + "ewc_loss": 0.00118255615234375, + "ewc_loss_diag": 1.1175870895385742e-06, + "ewc_loss_parallel": 1.0654330253601074e-06, + "grad_norm": 3.7829363346099854, + "learning_rate": 1.6066129715981347e-07, + "loss": 0.4424, + "mean_token_accuracy": 0.8554588556289673, + "num_tokens": 14753308.0, + "step": 380 + }, + { + "epoch": 0.04846711614298435, + "ewc_loss": 0.00115966796875, + "ewc_loss_diag": 1.1175870895385742e-06, + "ewc_loss_parallel": 1.043081283569336e-06, + "grad_norm": 3.699164390563965, + "learning_rate": 1.6108520559559137e-07, + "loss": 0.4574, + "mean_token_accuracy": 0.8509781956672668, + "num_tokens": 14792841.0, + "step": 381 + }, + { + "epoch": 0.048594326421574865, + "ewc_loss": 0.0011444091796875, + "ewc_loss_diag": 1.1175870895385742e-06, + "ewc_loss_parallel": 1.0281801223754883e-06, + "grad_norm": 4.26801872253418, + "learning_rate": 1.6150911403136921e-07, + "loss": 0.5166, + "mean_token_accuracy": 0.8376909494400024, + "num_tokens": 14822027.0, + "step": 382 + }, + { + "epoch": 0.04872153670016537, + "ewc_loss": 0.001190185546875, + "ewc_loss_diag": 1.1175870895385742e-06, + "ewc_loss_parallel": 1.0728836059570312e-06, + "grad_norm": 3.672732353210449, + "learning_rate": 1.619330224671471e-07, + "loss": 0.5117, + "mean_token_accuracy": 0.8375192880630493, + "num_tokens": 14865325.0, + "step": 383 + }, + { + "epoch": 0.04884874697875588, + "ewc_loss": 0.00116729736328125, + "ewc_loss_diag": 1.125037670135498e-06, + "ewc_loss_parallel": 1.0505318641662598e-06, + "grad_norm": 3.7738771438598633, + "learning_rate": 1.6235693090292496e-07, + "loss": 0.513, + "mean_token_accuracy": 0.8385077714920044, + "num_tokens": 14903379.0, + "step": 384 + }, + { + "epoch": 0.048975957257346395, + "ewc_loss": 0.00116729736328125, + "ewc_loss_diag": 1.1324882507324219e-06, + "ewc_loss_parallel": 1.0505318641662598e-06, + "grad_norm": 4.095177173614502, + "learning_rate": 1.6278083933870286e-07, + "loss": 0.5529, + "mean_token_accuracy": 0.823262095451355, + "num_tokens": 14937326.0, + "step": 385 + }, + { + "epoch": 0.0491031675359369, + "ewc_loss": 0.00119781494140625, + "ewc_loss_diag": 1.1324882507324219e-06, + "ewc_loss_parallel": 1.080334186553955e-06, + "grad_norm": 4.050222396850586, + "learning_rate": 1.632047477744807e-07, + "loss": 0.4937, + "mean_token_accuracy": 0.8444909453392029, + "num_tokens": 14973709.0, + "step": 386 + }, + { + "epoch": 0.049230377814527417, + "ewc_loss": 0.00119781494140625, + "ewc_loss_diag": 1.1399388313293457e-06, + "ewc_loss_parallel": 1.080334186553955e-06, + "grad_norm": 3.6940698623657227, + "learning_rate": 1.6362865621025858e-07, + "loss": 0.4966, + "mean_token_accuracy": 0.8390597105026245, + "num_tokens": 15012950.0, + "step": 387 + }, + { + "epoch": 0.049357588093117924, + "ewc_loss": 0.00115966796875, + "ewc_loss_diag": 1.1399388313293457e-06, + "ewc_loss_parallel": 1.043081283569336e-06, + "grad_norm": 3.862267017364502, + "learning_rate": 1.6405256464603645e-07, + "loss": 0.4598, + "mean_token_accuracy": 0.8517712354660034, + "num_tokens": 15049597.0, + "step": 388 + }, + { + "epoch": 0.04948479837170843, + "ewc_loss": 0.00119781494140625, + "ewc_loss_diag": 1.1548399925231934e-06, + "ewc_loss_parallel": 1.080334186553955e-06, + "grad_norm": 3.9675374031066895, + "learning_rate": 1.6447647308181432e-07, + "loss": 0.4953, + "mean_token_accuracy": 0.8376218676567078, + "num_tokens": 15083728.0, + "step": 389 + }, + { + "epoch": 0.049612008650298946, + "ewc_loss": 0.00119781494140625, + "ewc_loss_diag": 1.1622905731201172e-06, + "ewc_loss_parallel": 1.080334186553955e-06, + "grad_norm": 3.812793016433716, + "learning_rate": 1.649003815175922e-07, + "loss": 0.5423, + "mean_token_accuracy": 0.8278506994247437, + "num_tokens": 15123655.0, + "step": 390 + }, + { + "epoch": 0.04973921892888945, + "ewc_loss": 0.001190185546875, + "ewc_loss_diag": 1.169741153717041e-06, + "ewc_loss_parallel": 1.0728836059570312e-06, + "grad_norm": 3.6926193237304688, + "learning_rate": 1.6532428995337007e-07, + "loss": 0.4473, + "mean_token_accuracy": 0.8528658151626587, + "num_tokens": 15163387.0, + "step": 391 + }, + { + "epoch": 0.04986642920747997, + "ewc_loss": 0.00116729736328125, + "ewc_loss_diag": 1.169741153717041e-06, + "ewc_loss_parallel": 1.0505318641662598e-06, + "grad_norm": 3.735187530517578, + "learning_rate": 1.6574819838914794e-07, + "loss": 0.4363, + "mean_token_accuracy": 0.8618905544281006, + "num_tokens": 15201409.0, + "step": 392 + }, + { + "epoch": 0.049993639486070475, + "ewc_loss": 0.00118255615234375, + "ewc_loss_diag": 1.169741153717041e-06, + "ewc_loss_parallel": 1.0654330253601074e-06, + "grad_norm": 3.751554250717163, + "learning_rate": 1.661721068249258e-07, + "loss": 0.4785, + "mean_token_accuracy": 0.8418678641319275, + "num_tokens": 15240108.0, + "step": 393 + }, + { + "epoch": 0.05012084976466098, + "ewc_loss": 0.00119781494140625, + "ewc_loss_diag": 1.169741153717041e-06, + "ewc_loss_parallel": 1.080334186553955e-06, + "grad_norm": 3.821803331375122, + "learning_rate": 1.6659601526070368e-07, + "loss": 0.5156, + "mean_token_accuracy": 0.8341964483261108, + "num_tokens": 15281583.0, + "step": 394 + }, + { + "epoch": 0.0502480600432515, + "ewc_loss": 0.00119781494140625, + "ewc_loss_diag": 1.1771917343139648e-06, + "ewc_loss_parallel": 1.080334186553955e-06, + "grad_norm": 3.6834778785705566, + "learning_rate": 1.6701992369648156e-07, + "loss": 0.4913, + "mean_token_accuracy": 0.8404163122177124, + "num_tokens": 15323574.0, + "step": 395 + }, + { + "epoch": 0.050375270321842004, + "ewc_loss": 0.00119781494140625, + "ewc_loss_diag": 1.1771917343139648e-06, + "ewc_loss_parallel": 1.080334186553955e-06, + "grad_norm": 3.7731542587280273, + "learning_rate": 1.6744383213225943e-07, + "loss": 0.4579, + "mean_token_accuracy": 0.8475236296653748, + "num_tokens": 15361350.0, + "step": 396 + }, + { + "epoch": 0.05050248060043251, + "ewc_loss": 0.0012054443359375, + "ewc_loss_diag": 1.1846423149108887e-06, + "ewc_loss_parallel": 1.087784767150879e-06, + "grad_norm": 3.7395851612091064, + "learning_rate": 1.678677405680373e-07, + "loss": 0.4332, + "mean_token_accuracy": 0.8593769669532776, + "num_tokens": 15398340.0, + "step": 397 + }, + { + "epoch": 0.050629690879023026, + "ewc_loss": 0.001220703125, + "ewc_loss_diag": 1.1920928955078125e-06, + "ewc_loss_parallel": 1.0952353477478027e-06, + "grad_norm": 3.856776237487793, + "learning_rate": 1.6829164900381518e-07, + "loss": 0.4596, + "mean_token_accuracy": 0.8483152389526367, + "num_tokens": 15434590.0, + "step": 398 + }, + { + "epoch": 0.05075690115761353, + "ewc_loss": 0.00122833251953125, + "ewc_loss_diag": 1.1920928955078125e-06, + "ewc_loss_parallel": 1.1026859283447266e-06, + "grad_norm": 3.5731163024902344, + "learning_rate": 1.6871555743959305e-07, + "loss": 0.4444, + "mean_token_accuracy": 0.854543924331665, + "num_tokens": 15477928.0, + "step": 399 + }, + { + "epoch": 0.05088411143620405, + "ewc_loss": 0.00121307373046875, + "ewc_loss_diag": 1.2069940567016602e-06, + "ewc_loss_parallel": 1.087784767150879e-06, + "grad_norm": 3.5670790672302246, + "learning_rate": 1.6913946587537092e-07, + "loss": 0.4629, + "mean_token_accuracy": 0.8507289886474609, + "num_tokens": 15520636.0, + "step": 400 + }, + { + "epoch": 0.051011321714794555, + "ewc_loss": 0.001220703125, + "ewc_loss_diag": 1.2069940567016602e-06, + "ewc_loss_parallel": 1.0952353477478027e-06, + "grad_norm": 3.748612642288208, + "learning_rate": 1.695633743111488e-07, + "loss": 0.4661, + "mean_token_accuracy": 0.8478440046310425, + "num_tokens": 15558637.0, + "step": 401 + }, + { + "epoch": 0.05113853199338506, + "ewc_loss": 0.00122833251953125, + "ewc_loss_diag": 1.2069940567016602e-06, + "ewc_loss_parallel": 1.1026859283447266e-06, + "grad_norm": 3.92277193069458, + "learning_rate": 1.6998728274692667e-07, + "loss": 0.4456, + "mean_token_accuracy": 0.857032835483551, + "num_tokens": 15591674.0, + "step": 402 + }, + { + "epoch": 0.05126574227197558, + "ewc_loss": 0.00124359130859375, + "ewc_loss_diag": 1.2069940567016602e-06, + "ewc_loss_parallel": 1.125037670135498e-06, + "grad_norm": 3.647819995880127, + "learning_rate": 1.7041119118270454e-07, + "loss": 0.45, + "mean_token_accuracy": 0.8547827005386353, + "num_tokens": 15633765.0, + "step": 403 + }, + { + "epoch": 0.051392952550566084, + "ewc_loss": 0.00122833251953125, + "ewc_loss_diag": 1.2069940567016602e-06, + "ewc_loss_parallel": 1.1026859283447266e-06, + "grad_norm": 3.990173101425171, + "learning_rate": 1.7083509961848238e-07, + "loss": 0.4988, + "mean_token_accuracy": 0.8381385803222656, + "num_tokens": 15668127.0, + "step": 404 + }, + { + "epoch": 0.0515201628291566, + "ewc_loss": 0.00124359130859375, + "ewc_loss_diag": 1.2069940567016602e-06, + "ewc_loss_parallel": 1.125037670135498e-06, + "grad_norm": 3.681596040725708, + "learning_rate": 1.7125900805426028e-07, + "loss": 0.5182, + "mean_token_accuracy": 0.8319604396820068, + "num_tokens": 15715683.0, + "step": 405 + }, + { + "epoch": 0.051647373107747106, + "ewc_loss": 0.0012359619140625, + "ewc_loss_diag": 1.2069940567016602e-06, + "ewc_loss_parallel": 1.1101365089416504e-06, + "grad_norm": 3.7942988872528076, + "learning_rate": 1.7168291649003813e-07, + "loss": 0.4573, + "mean_token_accuracy": 0.8483836650848389, + "num_tokens": 15749504.0, + "step": 406 + }, + { + "epoch": 0.051774583386337614, + "ewc_loss": 0.0012359619140625, + "ewc_loss_diag": 1.2069940567016602e-06, + "ewc_loss_parallel": 1.1101365089416504e-06, + "grad_norm": 3.795988082885742, + "learning_rate": 1.7210682492581603e-07, + "loss": 0.486, + "mean_token_accuracy": 0.8446340560913086, + "num_tokens": 15787930.0, + "step": 407 + }, + { + "epoch": 0.05190179366492813, + "ewc_loss": 0.00124359130859375, + "ewc_loss_diag": 1.2069940567016602e-06, + "ewc_loss_parallel": 1.125037670135498e-06, + "grad_norm": 3.6949591636657715, + "learning_rate": 1.7253073336159387e-07, + "loss": 0.4511, + "mean_token_accuracy": 0.8524710536003113, + "num_tokens": 15828649.0, + "step": 408 + }, + { + "epoch": 0.052029003943518635, + "ewc_loss": 0.0012359619140625, + "ewc_loss_diag": 1.2069940567016602e-06, + "ewc_loss_parallel": 1.1101365089416504e-06, + "grad_norm": 3.764030694961548, + "learning_rate": 1.7295464179737177e-07, + "loss": 0.4753, + "mean_token_accuracy": 0.8435363173484802, + "num_tokens": 15869403.0, + "step": 409 + }, + { + "epoch": 0.05215621422210915, + "ewc_loss": 0.0012359619140625, + "ewc_loss_diag": 1.214444637298584e-06, + "ewc_loss_parallel": 1.1175870895385742e-06, + "grad_norm": 3.7085561752319336, + "learning_rate": 1.7337855023314962e-07, + "loss": 0.426, + "mean_token_accuracy": 0.860944390296936, + "num_tokens": 15909375.0, + "step": 410 + }, + { + "epoch": 0.05228342450069966, + "ewc_loss": 0.0012359619140625, + "ewc_loss_diag": 1.214444637298584e-06, + "ewc_loss_parallel": 1.1175870895385742e-06, + "grad_norm": 3.8198113441467285, + "learning_rate": 1.7380245866892752e-07, + "loss": 0.4518, + "mean_token_accuracy": 0.8557797074317932, + "num_tokens": 15944989.0, + "step": 411 + }, + { + "epoch": 0.052410634779290165, + "ewc_loss": 0.00124359130859375, + "ewc_loss_diag": 1.2218952178955078e-06, + "ewc_loss_parallel": 1.125037670135498e-06, + "grad_norm": 3.8075647354125977, + "learning_rate": 1.7422636710470536e-07, + "loss": 0.4872, + "mean_token_accuracy": 0.8430906534194946, + "num_tokens": 15983185.0, + "step": 412 + }, + { + "epoch": 0.05253784505788068, + "ewc_loss": 0.00124359130859375, + "ewc_loss_diag": 1.2218952178955078e-06, + "ewc_loss_parallel": 1.125037670135498e-06, + "grad_norm": 3.7893919944763184, + "learning_rate": 1.7465027554048326e-07, + "loss": 0.5008, + "mean_token_accuracy": 0.8385083079338074, + "num_tokens": 16023938.0, + "step": 413 + }, + { + "epoch": 0.05266505533647119, + "ewc_loss": 0.001251220703125, + "ewc_loss_diag": 1.2218952178955078e-06, + "ewc_loss_parallel": 1.1324882507324219e-06, + "grad_norm": 3.9930295944213867, + "learning_rate": 1.750741839762611e-07, + "loss": 0.4673, + "mean_token_accuracy": 0.8486819267272949, + "num_tokens": 16056761.0, + "step": 414 + }, + { + "epoch": 0.052792265615061694, + "ewc_loss": 0.00125885009765625, + "ewc_loss_diag": 1.2367963790893555e-06, + "ewc_loss_parallel": 1.1399388313293457e-06, + "grad_norm": 3.7444589138031006, + "learning_rate": 1.75498092412039e-07, + "loss": 0.4355, + "mean_token_accuracy": 0.8571338653564453, + "num_tokens": 16093670.0, + "step": 415 + }, + { + "epoch": 0.05291947589365221, + "ewc_loss": 0.00124359130859375, + "ewc_loss_diag": 1.2442469596862793e-06, + "ewc_loss_parallel": 1.125037670135498e-06, + "grad_norm": 4.03150749206543, + "learning_rate": 1.7592200084781686e-07, + "loss": 0.5185, + "mean_token_accuracy": 0.8328496217727661, + "num_tokens": 16129264.0, + "step": 416 + }, + { + "epoch": 0.053046686172242716, + "ewc_loss": 0.00127410888671875, + "ewc_loss_diag": 1.2442469596862793e-06, + "ewc_loss_parallel": 1.1548399925231934e-06, + "grad_norm": 3.7555859088897705, + "learning_rate": 1.7634590928359475e-07, + "loss": 0.4803, + "mean_token_accuracy": 0.845869779586792, + "num_tokens": 16169911.0, + "step": 417 + }, + { + "epoch": 0.05317389645083323, + "ewc_loss": 0.001251220703125, + "ewc_loss_diag": 1.2442469596862793e-06, + "ewc_loss_parallel": 1.1324882507324219e-06, + "grad_norm": 3.8623671531677246, + "learning_rate": 1.767698177193726e-07, + "loss": 0.477, + "mean_token_accuracy": 0.8464954495429993, + "num_tokens": 16206954.0, + "step": 418 + }, + { + "epoch": 0.05330110672942374, + "ewc_loss": 0.001251220703125, + "ewc_loss_diag": 1.2516975402832031e-06, + "ewc_loss_parallel": 1.1324882507324219e-06, + "grad_norm": 3.8235654830932617, + "learning_rate": 1.771937261551505e-07, + "loss": 0.4444, + "mean_token_accuracy": 0.8581211566925049, + "num_tokens": 16243094.0, + "step": 419 + }, + { + "epoch": 0.053428317008014245, + "ewc_loss": 0.001251220703125, + "ewc_loss_diag": 1.2516975402832031e-06, + "ewc_loss_parallel": 1.1324882507324219e-06, + "grad_norm": 3.7533910274505615, + "learning_rate": 1.7761763459092835e-07, + "loss": 0.4779, + "mean_token_accuracy": 0.8462436199188232, + "num_tokens": 16283619.0, + "step": 420 + }, + { + "epoch": 0.05355552728660476, + "ewc_loss": 0.00125885009765625, + "ewc_loss_diag": 1.2665987014770508e-06, + "ewc_loss_parallel": 1.1324882507324219e-06, + "grad_norm": 3.876894235610962, + "learning_rate": 1.7804154302670624e-07, + "loss": 0.4597, + "mean_token_accuracy": 0.8508046865463257, + "num_tokens": 16318901.0, + "step": 421 + }, + { + "epoch": 0.05368273756519527, + "ewc_loss": 0.00128173828125, + "ewc_loss_diag": 1.2665987014770508e-06, + "ewc_loss_parallel": 1.1548399925231934e-06, + "grad_norm": 3.711076259613037, + "learning_rate": 1.784654514624841e-07, + "loss": 0.4409, + "mean_token_accuracy": 0.8544082045555115, + "num_tokens": 16358890.0, + "step": 422 + }, + { + "epoch": 0.05380994784378578, + "ewc_loss": 0.00125885009765625, + "ewc_loss_diag": 1.2740492820739746e-06, + "ewc_loss_parallel": 1.1324882507324219e-06, + "grad_norm": 3.8533365726470947, + "learning_rate": 1.7888935989826196e-07, + "loss": 0.4801, + "mean_token_accuracy": 0.8404191732406616, + "num_tokens": 16395510.0, + "step": 423 + }, + { + "epoch": 0.05393715812237629, + "ewc_loss": 0.00128173828125, + "ewc_loss_diag": 1.2889504432678223e-06, + "ewc_loss_parallel": 1.1548399925231934e-06, + "grad_norm": 3.8655710220336914, + "learning_rate": 1.7931326833403984e-07, + "loss": 0.4786, + "mean_token_accuracy": 0.8446407318115234, + "num_tokens": 16434214.0, + "step": 424 + }, + { + "epoch": 0.054064368400966796, + "ewc_loss": 0.00128173828125, + "ewc_loss_diag": 1.2889504432678223e-06, + "ewc_loss_parallel": 1.1548399925231934e-06, + "grad_norm": 3.6068742275238037, + "learning_rate": 1.797371767698177e-07, + "loss": 0.4321, + "mean_token_accuracy": 0.8611724972724915, + "num_tokens": 16479012.0, + "step": 425 + }, + { + "epoch": 0.05419157867955731, + "ewc_loss": 0.0012664794921875, + "ewc_loss_diag": 1.296401023864746e-06, + "ewc_loss_parallel": 1.1399388313293457e-06, + "grad_norm": 3.595456838607788, + "learning_rate": 1.8016108520559558e-07, + "loss": 0.4552, + "mean_token_accuracy": 0.8520863056182861, + "num_tokens": 16521897.0, + "step": 426 + }, + { + "epoch": 0.05431878895814782, + "ewc_loss": 0.00128173828125, + "ewc_loss_diag": 1.296401023864746e-06, + "ewc_loss_parallel": 1.1548399925231934e-06, + "grad_norm": 3.9191806316375732, + "learning_rate": 1.8058499364137345e-07, + "loss": 0.5112, + "mean_token_accuracy": 0.8399369716644287, + "num_tokens": 16560594.0, + "step": 427 + }, + { + "epoch": 0.054445999236738325, + "ewc_loss": 0.0012969970703125, + "ewc_loss_diag": 1.296401023864746e-06, + "ewc_loss_parallel": 1.169741153717041e-06, + "grad_norm": 3.8036043643951416, + "learning_rate": 1.8100890207715133e-07, + "loss": 0.5387, + "mean_token_accuracy": 0.8269795179367065, + "num_tokens": 16603347.0, + "step": 428 + }, + { + "epoch": 0.05457320951532884, + "ewc_loss": 0.00128936767578125, + "ewc_loss_diag": 1.30385160446167e-06, + "ewc_loss_parallel": 1.1622905731201172e-06, + "grad_norm": 3.8653602600097656, + "learning_rate": 1.814328105129292e-07, + "loss": 0.4925, + "mean_token_accuracy": 0.8396530151367188, + "num_tokens": 16640721.0, + "step": 429 + }, + { + "epoch": 0.05470041979391935, + "ewc_loss": 0.0012969970703125, + "ewc_loss_diag": 1.30385160446167e-06, + "ewc_loss_parallel": 1.169741153717041e-06, + "grad_norm": 4.294247627258301, + "learning_rate": 1.8185671894870707e-07, + "loss": 0.4807, + "mean_token_accuracy": 0.8433119654655457, + "num_tokens": 16668710.0, + "step": 430 + }, + { + "epoch": 0.05482763007250986, + "ewc_loss": 0.00131988525390625, + "ewc_loss_diag": 1.3113021850585938e-06, + "ewc_loss_parallel": 1.1920928955078125e-06, + "grad_norm": 3.8716814517974854, + "learning_rate": 1.8228062738448494e-07, + "loss": 0.5126, + "mean_token_accuracy": 0.833415150642395, + "num_tokens": 16709771.0, + "step": 431 + }, + { + "epoch": 0.05495484035110037, + "ewc_loss": 0.00130462646484375, + "ewc_loss_diag": 1.3113021850585938e-06, + "ewc_loss_parallel": 1.1771917343139648e-06, + "grad_norm": 3.89448881149292, + "learning_rate": 1.8270453582026282e-07, + "loss": 0.4918, + "mean_token_accuracy": 0.8408985137939453, + "num_tokens": 16746643.0, + "step": 432 + }, + { + "epoch": 0.055082050629690876, + "ewc_loss": 0.00130462646484375, + "ewc_loss_diag": 1.3113021850585938e-06, + "ewc_loss_parallel": 1.1771917343139648e-06, + "grad_norm": 3.792510509490967, + "learning_rate": 1.831284442560407e-07, + "loss": 0.4565, + "mean_token_accuracy": 0.8509416580200195, + "num_tokens": 16787259.0, + "step": 433 + }, + { + "epoch": 0.05520926090828139, + "ewc_loss": 0.00128936767578125, + "ewc_loss_diag": 1.3113021850585938e-06, + "ewc_loss_parallel": 1.1622905731201172e-06, + "grad_norm": 3.7401862144470215, + "learning_rate": 1.8355235269181856e-07, + "loss": 0.4513, + "mean_token_accuracy": 0.8524606227874756, + "num_tokens": 16831478.0, + "step": 434 + }, + { + "epoch": 0.0553364711868719, + "ewc_loss": 0.00130462646484375, + "ewc_loss_diag": 1.3113021850585938e-06, + "ewc_loss_parallel": 1.1771917343139648e-06, + "grad_norm": 4.009796619415283, + "learning_rate": 1.8397626112759643e-07, + "loss": 0.4592, + "mean_token_accuracy": 0.8517729043960571, + "num_tokens": 16864209.0, + "step": 435 + }, + { + "epoch": 0.05546368146546241, + "ewc_loss": 0.0013275146484375, + "ewc_loss_diag": 1.3113021850585938e-06, + "ewc_loss_parallel": 1.1995434761047363e-06, + "grad_norm": 3.8291189670562744, + "learning_rate": 1.844001695633743e-07, + "loss": 0.4324, + "mean_token_accuracy": 0.8564390540122986, + "num_tokens": 16900269.0, + "step": 436 + }, + { + "epoch": 0.05559089174405292, + "ewc_loss": 0.001312255859375, + "ewc_loss_diag": 1.3187527656555176e-06, + "ewc_loss_parallel": 1.1846423149108887e-06, + "grad_norm": 3.759695529937744, + "learning_rate": 1.8482407799915218e-07, + "loss": 0.4328, + "mean_token_accuracy": 0.8599977493286133, + "num_tokens": 16937499.0, + "step": 437 + }, + { + "epoch": 0.05571810202264343, + "ewc_loss": 0.00130462646484375, + "ewc_loss_diag": 1.3187527656555176e-06, + "ewc_loss_parallel": 1.1771917343139648e-06, + "grad_norm": 4.105775356292725, + "learning_rate": 1.8524798643493005e-07, + "loss": 0.5211, + "mean_token_accuracy": 0.8301665782928467, + "num_tokens": 16972126.0, + "step": 438 + }, + { + "epoch": 0.05584531230123394, + "ewc_loss": 0.0013427734375, + "ewc_loss_diag": 1.3187527656555176e-06, + "ewc_loss_parallel": 1.214444637298584e-06, + "grad_norm": 3.978860378265381, + "learning_rate": 1.8567189487070792e-07, + "loss": 0.5103, + "mean_token_accuracy": 0.8377097249031067, + "num_tokens": 17011017.0, + "step": 439 + }, + { + "epoch": 0.05597252257982445, + "ewc_loss": 0.00133514404296875, + "ewc_loss_diag": 1.3187527656555176e-06, + "ewc_loss_parallel": 1.2069940567016602e-06, + "grad_norm": 3.7052886486053467, + "learning_rate": 1.8609580330648577e-07, + "loss": 0.4237, + "mean_token_accuracy": 0.8626124858856201, + "num_tokens": 17050432.0, + "step": 440 + }, + { + "epoch": 0.05609973285841496, + "ewc_loss": 0.00130462646484375, + "ewc_loss_diag": 1.3187527656555176e-06, + "ewc_loss_parallel": 1.1771917343139648e-06, + "grad_norm": 3.754077434539795, + "learning_rate": 1.8651971174226367e-07, + "loss": 0.4813, + "mean_token_accuracy": 0.8452396988868713, + "num_tokens": 17090390.0, + "step": 441 + }, + { + "epoch": 0.05622694313700547, + "ewc_loss": 0.00131988525390625, + "ewc_loss_diag": 1.3187527656555176e-06, + "ewc_loss_parallel": 1.1920928955078125e-06, + "grad_norm": 3.8653974533081055, + "learning_rate": 1.8694362017804152e-07, + "loss": 0.4738, + "mean_token_accuracy": 0.8467912673950195, + "num_tokens": 17131440.0, + "step": 442 + }, + { + "epoch": 0.05635415341559598, + "ewc_loss": 0.0013275146484375, + "ewc_loss_diag": 1.3187527656555176e-06, + "ewc_loss_parallel": 1.1995434761047363e-06, + "grad_norm": 3.7117116451263428, + "learning_rate": 1.8736752861381941e-07, + "loss": 0.4183, + "mean_token_accuracy": 0.864020824432373, + "num_tokens": 17171618.0, + "step": 443 + }, + { + "epoch": 0.05648136369418649, + "ewc_loss": 0.001312255859375, + "ewc_loss_diag": 1.3187527656555176e-06, + "ewc_loss_parallel": 1.1846423149108887e-06, + "grad_norm": 3.9129762649536133, + "learning_rate": 1.8779143704959726e-07, + "loss": 0.435, + "mean_token_accuracy": 0.858250081539154, + "num_tokens": 17204540.0, + "step": 444 + }, + { + "epoch": 0.056608573972777, + "ewc_loss": 0.0013427734375, + "ewc_loss_diag": 1.3336539268493652e-06, + "ewc_loss_parallel": 1.214444637298584e-06, + "grad_norm": 4.180079936981201, + "learning_rate": 1.8821534548537516e-07, + "loss": 0.4887, + "mean_token_accuracy": 0.8412876129150391, + "num_tokens": 17238263.0, + "step": 445 + }, + { + "epoch": 0.05673578425136751, + "ewc_loss": 0.0013885498046875, + "ewc_loss_diag": 1.3262033462524414e-06, + "ewc_loss_parallel": 1.259148120880127e-06, + "grad_norm": 3.712688684463501, + "learning_rate": 1.88639253921153e-07, + "loss": 0.4455, + "mean_token_accuracy": 0.855290412902832, + "num_tokens": 17282642.0, + "step": 446 + }, + { + "epoch": 0.05686299452995802, + "ewc_loss": 0.00133514404296875, + "ewc_loss_diag": 1.3262033462524414e-06, + "ewc_loss_parallel": 1.2069940567016602e-06, + "grad_norm": 3.660234212875366, + "learning_rate": 1.890631623569309e-07, + "loss": 0.4584, + "mean_token_accuracy": 0.8484693169593811, + "num_tokens": 17326204.0, + "step": 447 + }, + { + "epoch": 0.05699020480854853, + "ewc_loss": 0.0013427734375, + "ewc_loss_diag": 1.3336539268493652e-06, + "ewc_loss_parallel": 1.214444637298584e-06, + "grad_norm": 4.0977067947387695, + "learning_rate": 1.8948707079270875e-07, + "loss": 0.494, + "mean_token_accuracy": 0.8362369537353516, + "num_tokens": 17359275.0, + "step": 448 + }, + { + "epoch": 0.057117415087139044, + "ewc_loss": 0.0013885498046875, + "ewc_loss_diag": 1.3336539268493652e-06, + "ewc_loss_parallel": 1.259148120880127e-06, + "grad_norm": 3.8152015209198, + "learning_rate": 1.8991097922848665e-07, + "loss": 0.4277, + "mean_token_accuracy": 0.860493540763855, + "num_tokens": 17397945.0, + "step": 449 + }, + { + "epoch": 0.05724462536572955, + "ewc_loss": 0.00138092041015625, + "ewc_loss_diag": 1.341104507446289e-06, + "ewc_loss_parallel": 1.2442469596862793e-06, + "grad_norm": 3.8787593841552734, + "learning_rate": 1.903348876642645e-07, + "loss": 0.4871, + "mean_token_accuracy": 0.8405184745788574, + "num_tokens": 17439431.0, + "step": 450 + }, + { + "epoch": 0.05737183564432006, + "ewc_loss": 0.00138092041015625, + "ewc_loss_diag": 1.341104507446289e-06, + "ewc_loss_parallel": 1.2442469596862793e-06, + "grad_norm": 3.8606088161468506, + "learning_rate": 1.907587961000424e-07, + "loss": 0.44, + "mean_token_accuracy": 0.8577737808227539, + "num_tokens": 17478395.0, + "step": 451 + }, + { + "epoch": 0.05749904592291057, + "ewc_loss": 0.0013885498046875, + "ewc_loss_diag": 1.341104507446289e-06, + "ewc_loss_parallel": 1.2516975402832031e-06, + "grad_norm": 3.802608013153076, + "learning_rate": 1.9118270453582024e-07, + "loss": 0.4792, + "mean_token_accuracy": 0.8445267081260681, + "num_tokens": 17521566.0, + "step": 452 + }, + { + "epoch": 0.05762625620150108, + "ewc_loss": 0.0013885498046875, + "ewc_loss_diag": 1.3560056686401367e-06, + "ewc_loss_parallel": 1.2516975402832031e-06, + "grad_norm": 3.851916551589966, + "learning_rate": 1.9160661297159814e-07, + "loss": 0.4992, + "mean_token_accuracy": 0.8408515453338623, + "num_tokens": 17560668.0, + "step": 453 + }, + { + "epoch": 0.057753466480091595, + "ewc_loss": 0.00138092041015625, + "ewc_loss_diag": 1.3485550880432129e-06, + "ewc_loss_parallel": 1.2442469596862793e-06, + "grad_norm": 4.091015338897705, + "learning_rate": 1.9203052140737599e-07, + "loss": 0.4972, + "mean_token_accuracy": 0.837082028388977, + "num_tokens": 17595346.0, + "step": 454 + }, + { + "epoch": 0.0578806767586821, + "ewc_loss": 0.00140380859375, + "ewc_loss_diag": 1.3485550880432129e-06, + "ewc_loss_parallel": 1.2665987014770508e-06, + "grad_norm": 3.858085870742798, + "learning_rate": 1.9245442984315389e-07, + "loss": 0.4894, + "mean_token_accuracy": 0.844639778137207, + "num_tokens": 17636559.0, + "step": 455 + }, + { + "epoch": 0.05800788703727261, + "ewc_loss": 0.00138092041015625, + "ewc_loss_diag": 1.3634562492370605e-06, + "ewc_loss_parallel": 1.2442469596862793e-06, + "grad_norm": 3.885249376296997, + "learning_rate": 1.9287833827893173e-07, + "loss": 0.4744, + "mean_token_accuracy": 0.8439666628837585, + "num_tokens": 17675094.0, + "step": 456 + }, + { + "epoch": 0.058135097315863124, + "ewc_loss": 0.00139617919921875, + "ewc_loss_diag": 1.3709068298339844e-06, + "ewc_loss_parallel": 1.259148120880127e-06, + "grad_norm": 4.009865760803223, + "learning_rate": 1.9330224671470963e-07, + "loss": 0.4817, + "mean_token_accuracy": 0.8431679606437683, + "num_tokens": 17710534.0, + "step": 457 + }, + { + "epoch": 0.05826230759445363, + "ewc_loss": 0.00141143798828125, + "ewc_loss_diag": 1.3783574104309082e-06, + "ewc_loss_parallel": 1.2740492820739746e-06, + "grad_norm": 4.194794178009033, + "learning_rate": 1.9372615515048748e-07, + "loss": 0.5196, + "mean_token_accuracy": 0.8365523219108582, + "num_tokens": 17744455.0, + "step": 458 + }, + { + "epoch": 0.05838951787304414, + "ewc_loss": 0.00141143798828125, + "ewc_loss_diag": 1.385807991027832e-06, + "ewc_loss_parallel": 1.2740492820739746e-06, + "grad_norm": 3.9928672313690186, + "learning_rate": 1.9415006358626535e-07, + "loss": 0.4724, + "mean_token_accuracy": 0.8472001552581787, + "num_tokens": 17780557.0, + "step": 459 + }, + { + "epoch": 0.05851672815163465, + "ewc_loss": 0.00140380859375, + "ewc_loss_diag": 1.3783574104309082e-06, + "ewc_loss_parallel": 1.2665987014770508e-06, + "grad_norm": 4.097584247589111, + "learning_rate": 1.9457397202204322e-07, + "loss": 0.4562, + "mean_token_accuracy": 0.8503340482711792, + "num_tokens": 17812776.0, + "step": 460 + }, + { + "epoch": 0.05864393843022516, + "ewc_loss": 0.0014190673828125, + "ewc_loss_diag": 1.385807991027832e-06, + "ewc_loss_parallel": 1.2814998626708984e-06, + "grad_norm": 3.9000349044799805, + "learning_rate": 1.949978804578211e-07, + "loss": 0.4217, + "mean_token_accuracy": 0.8643194437026978, + "num_tokens": 17847666.0, + "step": 461 + }, + { + "epoch": 0.058771148708815675, + "ewc_loss": 0.00141143798828125, + "ewc_loss_diag": 1.3932585716247559e-06, + "ewc_loss_parallel": 1.2740492820739746e-06, + "grad_norm": 3.905500888824463, + "learning_rate": 1.9542178889359897e-07, + "loss": 0.5113, + "mean_token_accuracy": 0.8331221342086792, + "num_tokens": 17887806.0, + "step": 462 + }, + { + "epoch": 0.05889835898740618, + "ewc_loss": 0.0014190673828125, + "ewc_loss_diag": 1.385807991027832e-06, + "ewc_loss_parallel": 1.2814998626708984e-06, + "grad_norm": 4.160917282104492, + "learning_rate": 1.9584569732937684e-07, + "loss": 0.4775, + "mean_token_accuracy": 0.8493009805679321, + "num_tokens": 17921364.0, + "step": 463 + }, + { + "epoch": 0.05902556926599669, + "ewc_loss": 0.0014495849609375, + "ewc_loss_diag": 1.3932585716247559e-06, + "ewc_loss_parallel": 1.3113021850585938e-06, + "grad_norm": 3.864290952682495, + "learning_rate": 1.962696057651547e-07, + "loss": 0.3958, + "mean_token_accuracy": 0.869142472743988, + "num_tokens": 17959121.0, + "step": 464 + }, + { + "epoch": 0.059152779544587204, + "ewc_loss": 0.0014190673828125, + "ewc_loss_diag": 1.3932585716247559e-06, + "ewc_loss_parallel": 1.2814998626708984e-06, + "grad_norm": 4.058414459228516, + "learning_rate": 1.9669351420093258e-07, + "loss": 0.486, + "mean_token_accuracy": 0.8438212275505066, + "num_tokens": 17994648.0, + "step": 465 + }, + { + "epoch": 0.05927998982317771, + "ewc_loss": 0.00144195556640625, + "ewc_loss_diag": 1.3932585716247559e-06, + "ewc_loss_parallel": 1.30385160446167e-06, + "grad_norm": 3.946721076965332, + "learning_rate": 1.9711742263671046e-07, + "loss": 0.4879, + "mean_token_accuracy": 0.8437789678573608, + "num_tokens": 18039489.0, + "step": 466 + }, + { + "epoch": 0.059407200101768226, + "ewc_loss": 0.001434326171875, + "ewc_loss_diag": 1.3932585716247559e-06, + "ewc_loss_parallel": 1.296401023864746e-06, + "grad_norm": 3.956051826477051, + "learning_rate": 1.9754133107248833e-07, + "loss": 0.4423, + "mean_token_accuracy": 0.8542479872703552, + "num_tokens": 18074448.0, + "step": 467 + }, + { + "epoch": 0.059534410380358734, + "ewc_loss": 0.001434326171875, + "ewc_loss_diag": 1.4007091522216797e-06, + "ewc_loss_parallel": 1.296401023864746e-06, + "grad_norm": 3.899142265319824, + "learning_rate": 1.979652395082662e-07, + "loss": 0.497, + "mean_token_accuracy": 0.8423391580581665, + "num_tokens": 18116657.0, + "step": 468 + }, + { + "epoch": 0.05966162065894924, + "ewc_loss": 0.001434326171875, + "ewc_loss_diag": 1.4081597328186035e-06, + "ewc_loss_parallel": 1.296401023864746e-06, + "grad_norm": 3.890122413635254, + "learning_rate": 1.9838914794404408e-07, + "loss": 0.4108, + "mean_token_accuracy": 0.8636353015899658, + "num_tokens": 18153530.0, + "step": 469 + }, + { + "epoch": 0.059788830937539755, + "ewc_loss": 0.0014495849609375, + "ewc_loss_diag": 1.4156103134155273e-06, + "ewc_loss_parallel": 1.30385160446167e-06, + "grad_norm": 3.7942068576812744, + "learning_rate": 1.9881305637982195e-07, + "loss": 0.4629, + "mean_token_accuracy": 0.8493771553039551, + "num_tokens": 18197220.0, + "step": 470 + }, + { + "epoch": 0.05991604121613026, + "ewc_loss": 0.001434326171875, + "ewc_loss_diag": 1.4156103134155273e-06, + "ewc_loss_parallel": 1.296401023864746e-06, + "grad_norm": 4.146648406982422, + "learning_rate": 1.9923696481559982e-07, + "loss": 0.4882, + "mean_token_accuracy": 0.8445352911949158, + "num_tokens": 18231527.0, + "step": 471 + }, + { + "epoch": 0.06004325149472077, + "ewc_loss": 0.00146484375, + "ewc_loss_diag": 1.4156103134155273e-06, + "ewc_loss_parallel": 1.3262033462524414e-06, + "grad_norm": 4.046437740325928, + "learning_rate": 1.996608732513777e-07, + "loss": 0.532, + "mean_token_accuracy": 0.8294687867164612, + "num_tokens": 18274746.0, + "step": 472 + }, + { + "epoch": 0.060170461773311285, + "ewc_loss": 0.00146484375, + "ewc_loss_diag": 1.4230608940124512e-06, + "ewc_loss_parallel": 1.3187527656555176e-06, + "grad_norm": 3.789064407348633, + "learning_rate": 2.0008478168715557e-07, + "loss": 0.412, + "mean_token_accuracy": 0.8631265163421631, + "num_tokens": 18315119.0, + "step": 473 + }, + { + "epoch": 0.06029767205190179, + "ewc_loss": 0.001434326171875, + "ewc_loss_diag": 1.4230608940124512e-06, + "ewc_loss_parallel": 1.2889504432678223e-06, + "grad_norm": 3.8778390884399414, + "learning_rate": 2.0050869012293344e-07, + "loss": 0.4109, + "mean_token_accuracy": 0.863065779209137, + "num_tokens": 18349939.0, + "step": 474 + }, + { + "epoch": 0.060424882330492306, + "ewc_loss": 0.00145721435546875, + "ewc_loss_diag": 1.430511474609375e-06, + "ewc_loss_parallel": 1.3113021850585938e-06, + "grad_norm": 3.8766491413116455, + "learning_rate": 2.009325985587113e-07, + "loss": 0.431, + "mean_token_accuracy": 0.8570324778556824, + "num_tokens": 18388432.0, + "step": 475 + }, + { + "epoch": 0.060552092609082814, + "ewc_loss": 0.00145721435546875, + "ewc_loss_diag": 1.4454126358032227e-06, + "ewc_loss_parallel": 1.3113021850585938e-06, + "grad_norm": 4.095795154571533, + "learning_rate": 2.0135650699448918e-07, + "loss": 0.536, + "mean_token_accuracy": 0.8276363611221313, + "num_tokens": 18427874.0, + "step": 476 + }, + { + "epoch": 0.06067930288767332, + "ewc_loss": 0.0014801025390625, + "ewc_loss_diag": 1.430511474609375e-06, + "ewc_loss_parallel": 1.3336539268493652e-06, + "grad_norm": 4.0352983474731445, + "learning_rate": 2.0178041543026706e-07, + "loss": 0.4988, + "mean_token_accuracy": 0.8385406136512756, + "num_tokens": 18466075.0, + "step": 477 + }, + { + "epoch": 0.060806513166263836, + "ewc_loss": 0.00147247314453125, + "ewc_loss_diag": 1.430511474609375e-06, + "ewc_loss_parallel": 1.3262033462524414e-06, + "grad_norm": 3.8504743576049805, + "learning_rate": 2.022043238660449e-07, + "loss": 0.4998, + "mean_token_accuracy": 0.8364512324333191, + "num_tokens": 18508717.0, + "step": 478 + }, + { + "epoch": 0.06093372344485434, + "ewc_loss": 0.00145721435546875, + "ewc_loss_diag": 1.4603137969970703e-06, + "ewc_loss_parallel": 1.3113021850585938e-06, + "grad_norm": 4.181152820587158, + "learning_rate": 2.026282323018228e-07, + "loss": 0.4903, + "mean_token_accuracy": 0.8408554196357727, + "num_tokens": 18545386.0, + "step": 479 + }, + { + "epoch": 0.06106093372344486, + "ewc_loss": 0.00148773193359375, + "ewc_loss_diag": 1.4603137969970703e-06, + "ewc_loss_parallel": 1.341104507446289e-06, + "grad_norm": 3.9789223670959473, + "learning_rate": 2.0305214073760065e-07, + "loss": 0.4931, + "mean_token_accuracy": 0.8413413763046265, + "num_tokens": 18586608.0, + "step": 480 + }, + { + "epoch": 0.061188144002035365, + "ewc_loss": 0.00147247314453125, + "ewc_loss_diag": 1.4528632164001465e-06, + "ewc_loss_parallel": 1.3262033462524414e-06, + "grad_norm": 4.2608513832092285, + "learning_rate": 2.0347604917337855e-07, + "loss": 0.4796, + "mean_token_accuracy": 0.8424350023269653, + "num_tokens": 18617935.0, + "step": 481 + }, + { + "epoch": 0.06131535428062587, + "ewc_loss": 0.001495361328125, + "ewc_loss_diag": 1.4603137969970703e-06, + "ewc_loss_parallel": 1.3485550880432129e-06, + "grad_norm": 3.9608960151672363, + "learning_rate": 2.038999576091564e-07, + "loss": 0.4402, + "mean_token_accuracy": 0.8553199172019958, + "num_tokens": 18655894.0, + "step": 482 + }, + { + "epoch": 0.06144256455921639, + "ewc_loss": 0.00147247314453125, + "ewc_loss_diag": 1.4677643775939941e-06, + "ewc_loss_parallel": 1.3262033462524414e-06, + "grad_norm": 3.935171604156494, + "learning_rate": 2.043238660449343e-07, + "loss": 0.4977, + "mean_token_accuracy": 0.8393880128860474, + "num_tokens": 18694996.0, + "step": 483 + }, + { + "epoch": 0.061569774837806894, + "ewc_loss": 0.0014801025390625, + "ewc_loss_diag": 1.4677643775939941e-06, + "ewc_loss_parallel": 1.3336539268493652e-06, + "grad_norm": 3.967348098754883, + "learning_rate": 2.0474777448071214e-07, + "loss": 0.4471, + "mean_token_accuracy": 0.8559409976005554, + "num_tokens": 18733038.0, + "step": 484 + }, + { + "epoch": 0.0616969851163974, + "ewc_loss": 0.0014801025390625, + "ewc_loss_diag": 1.4677643775939941e-06, + "ewc_loss_parallel": 1.3336539268493652e-06, + "grad_norm": 4.160225868225098, + "learning_rate": 2.0517168291649004e-07, + "loss": 0.4775, + "mean_token_accuracy": 0.8474780917167664, + "num_tokens": 18766283.0, + "step": 485 + }, + { + "epoch": 0.061824195394987916, + "ewc_loss": 0.00150299072265625, + "ewc_loss_diag": 1.4677643775939941e-06, + "ewc_loss_parallel": 1.3560056686401367e-06, + "grad_norm": 4.061379909515381, + "learning_rate": 2.0559559135226788e-07, + "loss": 0.486, + "mean_token_accuracy": 0.8393808603286743, + "num_tokens": 18800691.0, + "step": 486 + }, + { + "epoch": 0.06195140567357842, + "ewc_loss": 0.00148773193359375, + "ewc_loss_diag": 1.4677643775939941e-06, + "ewc_loss_parallel": 1.341104507446289e-06, + "grad_norm": 3.8990466594696045, + "learning_rate": 2.0601949978804578e-07, + "loss": 0.4485, + "mean_token_accuracy": 0.852867603302002, + "num_tokens": 18840310.0, + "step": 487 + }, + { + "epoch": 0.06207861595216894, + "ewc_loss": 0.0014801025390625, + "ewc_loss_diag": 1.4826655387878418e-06, + "ewc_loss_parallel": 1.3336539268493652e-06, + "grad_norm": 3.871713876724243, + "learning_rate": 2.0644340822382363e-07, + "loss": 0.3992, + "mean_token_accuracy": 0.8686286807060242, + "num_tokens": 18876932.0, + "step": 488 + }, + { + "epoch": 0.062205826230759445, + "ewc_loss": 0.00147247314453125, + "ewc_loss_diag": 1.4826655387878418e-06, + "ewc_loss_parallel": 1.3262033462524414e-06, + "grad_norm": 3.9931681156158447, + "learning_rate": 2.0686731665960153e-07, + "loss": 0.4632, + "mean_token_accuracy": 0.8512338399887085, + "num_tokens": 18913366.0, + "step": 489 + }, + { + "epoch": 0.06233303650934995, + "ewc_loss": 0.00152587890625, + "ewc_loss_diag": 1.4901161193847656e-06, + "ewc_loss_parallel": 1.3709068298339844e-06, + "grad_norm": 3.8855559825897217, + "learning_rate": 2.0729122509537937e-07, + "loss": 0.4245, + "mean_token_accuracy": 0.8616272211074829, + "num_tokens": 18952996.0, + "step": 490 + }, + { + "epoch": 0.06246024678794047, + "ewc_loss": 0.0015106201171875, + "ewc_loss_diag": 1.4901161193847656e-06, + "ewc_loss_parallel": 1.3560056686401367e-06, + "grad_norm": 4.031771659851074, + "learning_rate": 2.0771513353115727e-07, + "loss": 0.4661, + "mean_token_accuracy": 0.847821831703186, + "num_tokens": 18989725.0, + "step": 491 + }, + { + "epoch": 0.06258745706653097, + "ewc_loss": 0.0015411376953125, + "ewc_loss_diag": 1.4975666999816895e-06, + "ewc_loss_parallel": 1.385807991027832e-06, + "grad_norm": 3.9418914318084717, + "learning_rate": 2.0813904196693512e-07, + "loss": 0.4389, + "mean_token_accuracy": 0.8568496704101562, + "num_tokens": 19031282.0, + "step": 492 + }, + { + "epoch": 0.06271466734512149, + "ewc_loss": 0.00152587890625, + "ewc_loss_diag": 1.5124678611755371e-06, + "ewc_loss_parallel": 1.3709068298339844e-06, + "grad_norm": 4.120761394500732, + "learning_rate": 2.0856295040271302e-07, + "loss": 0.4951, + "mean_token_accuracy": 0.8389882445335388, + "num_tokens": 19067035.0, + "step": 493 + }, + { + "epoch": 0.06284187762371199, + "ewc_loss": 0.001556396484375, + "ewc_loss_diag": 1.5124678611755371e-06, + "ewc_loss_parallel": 1.4007091522216797e-06, + "grad_norm": 3.9813666343688965, + "learning_rate": 2.0898685883849086e-07, + "loss": 0.4726, + "mean_token_accuracy": 0.851721465587616, + "num_tokens": 19109318.0, + "step": 494 + }, + { + "epoch": 0.0629690879023025, + "ewc_loss": 0.00153350830078125, + "ewc_loss_diag": 1.5050172805786133e-06, + "ewc_loss_parallel": 1.3783574104309082e-06, + "grad_norm": 3.9399921894073486, + "learning_rate": 2.0941076727426874e-07, + "loss": 0.4438, + "mean_token_accuracy": 0.855963945388794, + "num_tokens": 19149053.0, + "step": 495 + }, + { + "epoch": 0.06309629818089302, + "ewc_loss": 0.00152587890625, + "ewc_loss_diag": 1.5050172805786133e-06, + "ewc_loss_parallel": 1.3709068298339844e-06, + "grad_norm": 3.7411739826202393, + "learning_rate": 2.098346757100466e-07, + "loss": 0.4343, + "mean_token_accuracy": 0.8573123216629028, + "num_tokens": 19195129.0, + "step": 496 + }, + { + "epoch": 0.06322350845948353, + "ewc_loss": 0.00150299072265625, + "ewc_loss_diag": 1.5050172805786133e-06, + "ewc_loss_parallel": 1.3485550880432129e-06, + "grad_norm": 4.097530364990234, + "learning_rate": 2.1025858414582448e-07, + "loss": 0.4779, + "mean_token_accuracy": 0.844820499420166, + "num_tokens": 19231700.0, + "step": 497 + }, + { + "epoch": 0.06335071873807403, + "ewc_loss": 0.00156402587890625, + "ewc_loss_diag": 1.519918441772461e-06, + "ewc_loss_parallel": 1.4081597328186035e-06, + "grad_norm": 4.059571743011475, + "learning_rate": 2.1068249258160238e-07, + "loss": 0.4487, + "mean_token_accuracy": 0.8476581573486328, + "num_tokens": 19266105.0, + "step": 498 + }, + { + "epoch": 0.06347792901666455, + "ewc_loss": 0.001556396484375, + "ewc_loss_diag": 1.519918441772461e-06, + "ewc_loss_parallel": 1.4007091522216797e-06, + "grad_norm": 4.041560649871826, + "learning_rate": 2.1110640101738023e-07, + "loss": 0.4406, + "mean_token_accuracy": 0.8574285507202148, + "num_tokens": 19302993.0, + "step": 499 + }, + { + "epoch": 0.06360513929525506, + "ewc_loss": 0.00154876708984375, + "ewc_loss_diag": 1.519918441772461e-06, + "ewc_loss_parallel": 1.3932585716247559e-06, + "grad_norm": 4.030236721038818, + "learning_rate": 2.1153030945315813e-07, + "loss": 0.4507, + "mean_token_accuracy": 0.8525518178939819, + "num_tokens": 19338427.0, + "step": 500 + }, + { + "epoch": 0.06373234957384556, + "ewc_loss": 0.00156402587890625, + "ewc_loss_diag": 1.519918441772461e-06, + "ewc_loss_parallel": 1.4081597328186035e-06, + "grad_norm": 4.0951008796691895, + "learning_rate": 2.1195421788893597e-07, + "loss": 0.4496, + "mean_token_accuracy": 0.8546662330627441, + "num_tokens": 19372535.0, + "step": 501 + }, + { + "epoch": 0.06385955985243608, + "ewc_loss": 0.00157928466796875, + "ewc_loss_diag": 1.519918441772461e-06, + "ewc_loss_parallel": 1.4230608940124512e-06, + "grad_norm": 3.9547605514526367, + "learning_rate": 2.1237812632471387e-07, + "loss": 0.4611, + "mean_token_accuracy": 0.8474135398864746, + "num_tokens": 19413930.0, + "step": 502 + }, + { + "epoch": 0.06398677013102659, + "ewc_loss": 0.001556396484375, + "ewc_loss_diag": 1.519918441772461e-06, + "ewc_loss_parallel": 1.4007091522216797e-06, + "grad_norm": 4.089630126953125, + "learning_rate": 2.1280203476049172e-07, + "loss": 0.5527, + "mean_token_accuracy": 0.8184943199157715, + "num_tokens": 19454250.0, + "step": 503 + }, + { + "epoch": 0.06411398040961709, + "ewc_loss": 0.0015869140625, + "ewc_loss_diag": 1.519918441772461e-06, + "ewc_loss_parallel": 1.430511474609375e-06, + "grad_norm": 4.00958251953125, + "learning_rate": 2.1322594319626962e-07, + "loss": 0.4627, + "mean_token_accuracy": 0.8511067032814026, + "num_tokens": 19495244.0, + "step": 504 + }, + { + "epoch": 0.0642411906882076, + "ewc_loss": 0.00156402587890625, + "ewc_loss_diag": 1.5273690223693848e-06, + "ewc_loss_parallel": 1.4081597328186035e-06, + "grad_norm": 4.288912296295166, + "learning_rate": 2.1364985163204746e-07, + "loss": 0.4803, + "mean_token_accuracy": 0.8434898853302002, + "num_tokens": 19526033.0, + "step": 505 + }, + { + "epoch": 0.06436840096679812, + "ewc_loss": 0.0015869140625, + "ewc_loss_diag": 1.519918441772461e-06, + "ewc_loss_parallel": 1.4379620552062988e-06, + "grad_norm": 3.9984946250915527, + "learning_rate": 2.1407376006782536e-07, + "loss": 0.4221, + "mean_token_accuracy": 0.8642009496688843, + "num_tokens": 19563626.0, + "step": 506 + }, + { + "epoch": 0.06449561124538863, + "ewc_loss": 0.0015716552734375, + "ewc_loss_diag": 1.519918441772461e-06, + "ewc_loss_parallel": 1.4156103134155273e-06, + "grad_norm": 4.019052028656006, + "learning_rate": 2.144976685036032e-07, + "loss": 0.4486, + "mean_token_accuracy": 0.8552937507629395, + "num_tokens": 19599889.0, + "step": 507 + }, + { + "epoch": 0.06462282152397913, + "ewc_loss": 0.00157928466796875, + "ewc_loss_diag": 1.519918441772461e-06, + "ewc_loss_parallel": 1.4230608940124512e-06, + "grad_norm": 4.061014652252197, + "learning_rate": 2.149215769393811e-07, + "loss": 0.4924, + "mean_token_accuracy": 0.839553952217102, + "num_tokens": 19639905.0, + "step": 508 + }, + { + "epoch": 0.06475003180256965, + "ewc_loss": 0.0015716552734375, + "ewc_loss_diag": 1.5422701835632324e-06, + "ewc_loss_parallel": 1.4156103134155273e-06, + "grad_norm": 4.095646381378174, + "learning_rate": 2.1534548537515895e-07, + "loss": 0.4528, + "mean_token_accuracy": 0.8517018556594849, + "num_tokens": 19676224.0, + "step": 509 + }, + { + "epoch": 0.06487724208116016, + "ewc_loss": 0.0015869140625, + "ewc_loss_diag": 1.5422701835632324e-06, + "ewc_loss_parallel": 1.430511474609375e-06, + "grad_norm": 3.8994028568267822, + "learning_rate": 2.1576939381093685e-07, + "loss": 0.4851, + "mean_token_accuracy": 0.8433685302734375, + "num_tokens": 19719350.0, + "step": 510 + }, + { + "epoch": 0.06500445235975066, + "ewc_loss": 0.00156402587890625, + "ewc_loss_diag": 1.5497207641601562e-06, + "ewc_loss_parallel": 1.4081597328186035e-06, + "grad_norm": 3.852342367172241, + "learning_rate": 2.161933022467147e-07, + "loss": 0.466, + "mean_token_accuracy": 0.846605658531189, + "num_tokens": 19763955.0, + "step": 511 + }, + { + "epoch": 0.06513166263834118, + "ewc_loss": 0.001556396484375, + "ewc_loss_diag": 1.564621925354004e-06, + "ewc_loss_parallel": 1.4007091522216797e-06, + "grad_norm": 3.9199254512786865, + "learning_rate": 2.166172106824926e-07, + "loss": 0.4562, + "mean_token_accuracy": 0.8535677194595337, + "num_tokens": 19809960.0, + "step": 512 + }, + { + "epoch": 0.06525887291693169, + "ewc_loss": 0.0015869140625, + "ewc_loss_diag": 1.564621925354004e-06, + "ewc_loss_parallel": 1.4230608940124512e-06, + "grad_norm": 4.104804992675781, + "learning_rate": 2.1704111911827044e-07, + "loss": 0.4538, + "mean_token_accuracy": 0.8504892587661743, + "num_tokens": 19847591.0, + "step": 513 + }, + { + "epoch": 0.0653860831955222, + "ewc_loss": 0.001617431640625, + "ewc_loss_diag": 1.55717134475708e-06, + "ewc_loss_parallel": 1.4677643775939941e-06, + "grad_norm": 4.02310037612915, + "learning_rate": 2.1746502755404831e-07, + "loss": 0.5091, + "mean_token_accuracy": 0.8384187817573547, + "num_tokens": 19889939.0, + "step": 514 + }, + { + "epoch": 0.06551329347411271, + "ewc_loss": 0.00159454345703125, + "ewc_loss_diag": 1.5795230865478516e-06, + "ewc_loss_parallel": 1.430511474609375e-06, + "grad_norm": 4.151050567626953, + "learning_rate": 2.178889359898262e-07, + "loss": 0.5024, + "mean_token_accuracy": 0.8383551836013794, + "num_tokens": 19926473.0, + "step": 515 + }, + { + "epoch": 0.06564050375270322, + "ewc_loss": 0.00159454345703125, + "ewc_loss_diag": 1.5795230865478516e-06, + "ewc_loss_parallel": 1.4379620552062988e-06, + "grad_norm": 4.02302360534668, + "learning_rate": 2.1831284442560406e-07, + "loss": 0.4991, + "mean_token_accuracy": 0.8399064540863037, + "num_tokens": 19970403.0, + "step": 516 + }, + { + "epoch": 0.06576771403129372, + "ewc_loss": 0.00159454345703125, + "ewc_loss_diag": 1.5795230865478516e-06, + "ewc_loss_parallel": 1.4379620552062988e-06, + "grad_norm": 4.004934310913086, + "learning_rate": 2.1873675286138193e-07, + "loss": 0.5125, + "mean_token_accuracy": 0.8324392437934875, + "num_tokens": 20012675.0, + "step": 517 + }, + { + "epoch": 0.06589492430988424, + "ewc_loss": 0.00159454345703125, + "ewc_loss_diag": 1.5795230865478516e-06, + "ewc_loss_parallel": 1.4379620552062988e-06, + "grad_norm": 3.991661548614502, + "learning_rate": 2.191606612971598e-07, + "loss": 0.5247, + "mean_token_accuracy": 0.829782247543335, + "num_tokens": 20053534.0, + "step": 518 + }, + { + "epoch": 0.06602213458847475, + "ewc_loss": 0.00159454345703125, + "ewc_loss_diag": 1.5869736671447754e-06, + "ewc_loss_parallel": 1.4379620552062988e-06, + "grad_norm": 4.054246425628662, + "learning_rate": 2.1958456973293768e-07, + "loss": 0.5045, + "mean_token_accuracy": 0.8334985971450806, + "num_tokens": 20093017.0, + "step": 519 + }, + { + "epoch": 0.06614934486706527, + "ewc_loss": 0.0016021728515625, + "ewc_loss_diag": 1.5869736671447754e-06, + "ewc_loss_parallel": 1.4454126358032227e-06, + "grad_norm": 3.948150873184204, + "learning_rate": 2.2000847816871555e-07, + "loss": 0.4041, + "mean_token_accuracy": 0.8647497892379761, + "num_tokens": 20133265.0, + "step": 520 + }, + { + "epoch": 0.06627655514565577, + "ewc_loss": 0.00160980224609375, + "ewc_loss_diag": 1.5795230865478516e-06, + "ewc_loss_parallel": 1.4528632164001465e-06, + "grad_norm": 3.973853826522827, + "learning_rate": 2.2043238660449342e-07, + "loss": 0.4504, + "mean_token_accuracy": 0.8539625406265259, + "num_tokens": 20173175.0, + "step": 521 + }, + { + "epoch": 0.06640376542424628, + "ewc_loss": 0.0016021728515625, + "ewc_loss_diag": 1.5795230865478516e-06, + "ewc_loss_parallel": 1.4454126358032227e-06, + "grad_norm": 3.9176290035247803, + "learning_rate": 2.208562950402713e-07, + "loss": 0.4328, + "mean_token_accuracy": 0.8583635091781616, + "num_tokens": 20211005.0, + "step": 522 + }, + { + "epoch": 0.0665309757028368, + "ewc_loss": 0.0016021728515625, + "ewc_loss_diag": 1.5795230865478516e-06, + "ewc_loss_parallel": 1.4454126358032227e-06, + "grad_norm": 4.048157215118408, + "learning_rate": 2.2128020347604917e-07, + "loss": 0.4762, + "mean_token_accuracy": 0.846017599105835, + "num_tokens": 20250837.0, + "step": 523 + }, + { + "epoch": 0.0666581859814273, + "ewc_loss": 0.00162506103515625, + "ewc_loss_diag": 1.5795230865478516e-06, + "ewc_loss_parallel": 1.4677643775939941e-06, + "grad_norm": 3.83699893951416, + "learning_rate": 2.2170411191182704e-07, + "loss": 0.4343, + "mean_token_accuracy": 0.8579227924346924, + "num_tokens": 20298782.0, + "step": 524 + }, + { + "epoch": 0.06678539626001781, + "ewc_loss": 0.0016021728515625, + "ewc_loss_diag": 1.5795230865478516e-06, + "ewc_loss_parallel": 1.4454126358032227e-06, + "grad_norm": 4.29062557220459, + "learning_rate": 2.221280203476049e-07, + "loss": 0.4809, + "mean_token_accuracy": 0.8430464267730713, + "num_tokens": 20332914.0, + "step": 525 + }, + { + "epoch": 0.06691260653860832, + "ewc_loss": 0.001678466796875, + "ewc_loss_diag": 1.5944242477416992e-06, + "ewc_loss_parallel": 1.519918441772461e-06, + "grad_norm": 4.263268947601318, + "learning_rate": 2.2255192878338279e-07, + "loss": 0.4904, + "mean_token_accuracy": 0.841921329498291, + "num_tokens": 20370362.0, + "step": 526 + }, + { + "epoch": 0.06703981681719882, + "ewc_loss": 0.00164794921875, + "ewc_loss_diag": 1.601874828338623e-06, + "ewc_loss_parallel": 1.4901161193847656e-06, + "grad_norm": 4.1609721183776855, + "learning_rate": 2.2297583721916066e-07, + "loss": 0.4764, + "mean_token_accuracy": 0.8452195525169373, + "num_tokens": 20405683.0, + "step": 527 + }, + { + "epoch": 0.06716702709578934, + "ewc_loss": 0.00162506103515625, + "ewc_loss_diag": 1.5944242477416992e-06, + "ewc_loss_parallel": 1.4677643775939941e-06, + "grad_norm": 3.9238877296447754, + "learning_rate": 2.2339974565493853e-07, + "loss": 0.4285, + "mean_token_accuracy": 0.8608116507530212, + "num_tokens": 20446338.0, + "step": 528 + }, + { + "epoch": 0.06729423737437985, + "ewc_loss": 0.001617431640625, + "ewc_loss_diag": 1.5944242477416992e-06, + "ewc_loss_parallel": 1.4603137969970703e-06, + "grad_norm": 4.001370906829834, + "learning_rate": 2.238236540907164e-07, + "loss": 0.48, + "mean_token_accuracy": 0.8456985950469971, + "num_tokens": 20486720.0, + "step": 529 + }, + { + "epoch": 0.06742144765297035, + "ewc_loss": 0.00162506103515625, + "ewc_loss_diag": 1.601874828338623e-06, + "ewc_loss_parallel": 1.4677643775939941e-06, + "grad_norm": 4.128762722015381, + "learning_rate": 2.2424756252649428e-07, + "loss": 0.4462, + "mean_token_accuracy": 0.8532615900039673, + "num_tokens": 20522510.0, + "step": 530 + }, + { + "epoch": 0.06754865793156087, + "ewc_loss": 0.0016632080078125, + "ewc_loss_diag": 1.601874828338623e-06, + "ewc_loss_parallel": 1.5050172805786133e-06, + "grad_norm": 4.1621599197387695, + "learning_rate": 2.2467147096227215e-07, + "loss": 0.4352, + "mean_token_accuracy": 0.8563224077224731, + "num_tokens": 20557842.0, + "step": 531 + }, + { + "epoch": 0.06767586821015138, + "ewc_loss": 0.00165557861328125, + "ewc_loss_diag": 1.5944242477416992e-06, + "ewc_loss_parallel": 1.4975666999816895e-06, + "grad_norm": 4.382727146148682, + "learning_rate": 2.2509537939805002e-07, + "loss": 0.443, + "mean_token_accuracy": 0.8547883629798889, + "num_tokens": 20589553.0, + "step": 532 + }, + { + "epoch": 0.0678030784887419, + "ewc_loss": 0.00168609619140625, + "ewc_loss_diag": 1.601874828338623e-06, + "ewc_loss_parallel": 1.5273690223693848e-06, + "grad_norm": 4.135645866394043, + "learning_rate": 2.2551928783382787e-07, + "loss": 0.4643, + "mean_token_accuracy": 0.8483564853668213, + "num_tokens": 20626319.0, + "step": 533 + }, + { + "epoch": 0.0679302887673324, + "ewc_loss": 0.00164794921875, + "ewc_loss_diag": 1.6093254089355469e-06, + "ewc_loss_parallel": 1.4901161193847656e-06, + "grad_norm": 4.075421333312988, + "learning_rate": 2.2594319626960577e-07, + "loss": 0.4719, + "mean_token_accuracy": 0.8501427173614502, + "num_tokens": 20665460.0, + "step": 534 + }, + { + "epoch": 0.06805749904592291, + "ewc_loss": 0.00164794921875, + "ewc_loss_diag": 1.6093254089355469e-06, + "ewc_loss_parallel": 1.4901161193847656e-06, + "grad_norm": 4.018291473388672, + "learning_rate": 2.263671047053836e-07, + "loss": 0.4, + "mean_token_accuracy": 0.8646053075790405, + "num_tokens": 20701953.0, + "step": 535 + }, + { + "epoch": 0.06818470932451343, + "ewc_loss": 0.00164794921875, + "ewc_loss_diag": 1.6242265701293945e-06, + "ewc_loss_parallel": 1.4901161193847656e-06, + "grad_norm": 3.9014649391174316, + "learning_rate": 2.267910131411615e-07, + "loss": 0.4478, + "mean_token_accuracy": 0.8554297685623169, + "num_tokens": 20744030.0, + "step": 536 + }, + { + "epoch": 0.06831191960310393, + "ewc_loss": 0.00164794921875, + "ewc_loss_diag": 1.6242265701293945e-06, + "ewc_loss_parallel": 1.4901161193847656e-06, + "grad_norm": 4.322457790374756, + "learning_rate": 2.2721492157693936e-07, + "loss": 0.4513, + "mean_token_accuracy": 0.8522049188613892, + "num_tokens": 20774501.0, + "step": 537 + }, + { + "epoch": 0.06843912988169444, + "ewc_loss": 0.00170135498046875, + "ewc_loss_diag": 1.6242265701293945e-06, + "ewc_loss_parallel": 1.5422701835632324e-06, + "grad_norm": 4.048244476318359, + "learning_rate": 2.2763883001271726e-07, + "loss": 0.4345, + "mean_token_accuracy": 0.8569178581237793, + "num_tokens": 20812245.0, + "step": 538 + }, + { + "epoch": 0.06856634016028496, + "ewc_loss": 0.001678466796875, + "ewc_loss_diag": 1.6167759895324707e-06, + "ewc_loss_parallel": 1.519918441772461e-06, + "grad_norm": 4.010420799255371, + "learning_rate": 2.280627384484951e-07, + "loss": 0.477, + "mean_token_accuracy": 0.8427124619483948, + "num_tokens": 20854412.0, + "step": 539 + }, + { + "epoch": 0.06869355043887546, + "ewc_loss": 0.0016632080078125, + "ewc_loss_diag": 1.6242265701293945e-06, + "ewc_loss_parallel": 1.5050172805786133e-06, + "grad_norm": 4.184619426727295, + "learning_rate": 2.28486646884273e-07, + "loss": 0.4511, + "mean_token_accuracy": 0.8504699468612671, + "num_tokens": 20892866.0, + "step": 540 + }, + { + "epoch": 0.06882076071746597, + "ewc_loss": 0.0016937255859375, + "ewc_loss_diag": 1.6242265701293945e-06, + "ewc_loss_parallel": 1.5348196029663086e-06, + "grad_norm": 4.445523262023926, + "learning_rate": 2.2891055532005085e-07, + "loss": 0.4933, + "mean_token_accuracy": 0.8426423072814941, + "num_tokens": 20926721.0, + "step": 541 + }, + { + "epoch": 0.06894797099605648, + "ewc_loss": 0.00171661376953125, + "ewc_loss_diag": 1.6316771507263184e-06, + "ewc_loss_parallel": 1.55717134475708e-06, + "grad_norm": 4.453059673309326, + "learning_rate": 2.2933446375582875e-07, + "loss": 0.473, + "mean_token_accuracy": 0.8490537405014038, + "num_tokens": 20959425.0, + "step": 542 + }, + { + "epoch": 0.06907518127464699, + "ewc_loss": 0.0017242431640625, + "ewc_loss_diag": 1.6316771507263184e-06, + "ewc_loss_parallel": 1.564621925354004e-06, + "grad_norm": 4.0118794441223145, + "learning_rate": 2.297583721916066e-07, + "loss": 0.4376, + "mean_token_accuracy": 0.8593034148216248, + "num_tokens": 21000619.0, + "step": 543 + }, + { + "epoch": 0.0692023915532375, + "ewc_loss": 0.001678466796875, + "ewc_loss_diag": 1.6391277313232422e-06, + "ewc_loss_parallel": 1.5124678611755371e-06, + "grad_norm": 3.862658977508545, + "learning_rate": 2.301822806273845e-07, + "loss": 0.4478, + "mean_token_accuracy": 0.8548648953437805, + "num_tokens": 21043609.0, + "step": 544 + }, + { + "epoch": 0.06932960183182801, + "ewc_loss": 0.0016632080078125, + "ewc_loss_diag": 1.6391277313232422e-06, + "ewc_loss_parallel": 1.5050172805786133e-06, + "grad_norm": 4.2329277992248535, + "learning_rate": 2.3060618906316234e-07, + "loss": 0.4841, + "mean_token_accuracy": 0.8437848091125488, + "num_tokens": 21079869.0, + "step": 545 + }, + { + "epoch": 0.06945681211041853, + "ewc_loss": 0.0016937255859375, + "ewc_loss_diag": 1.6391277313232422e-06, + "ewc_loss_parallel": 1.5348196029663086e-06, + "grad_norm": 4.214629650115967, + "learning_rate": 2.3103009749894024e-07, + "loss": 0.4677, + "mean_token_accuracy": 0.846368670463562, + "num_tokens": 21117083.0, + "step": 546 + }, + { + "epoch": 0.06958402238900903, + "ewc_loss": 0.001708984375, + "ewc_loss_diag": 1.646578311920166e-06, + "ewc_loss_parallel": 1.5422701835632324e-06, + "grad_norm": 3.9060397148132324, + "learning_rate": 2.3145400593471808e-07, + "loss": 0.4745, + "mean_token_accuracy": 0.844596803188324, + "num_tokens": 21162088.0, + "step": 547 + }, + { + "epoch": 0.06971123266759954, + "ewc_loss": 0.001678466796875, + "ewc_loss_diag": 1.6391277313232422e-06, + "ewc_loss_parallel": 1.5124678611755371e-06, + "grad_norm": 4.360280513763428, + "learning_rate": 2.3187791437049598e-07, + "loss": 0.5039, + "mean_token_accuracy": 0.8326468467712402, + "num_tokens": 21195055.0, + "step": 548 + }, + { + "epoch": 0.06983844294619006, + "ewc_loss": 0.001739501953125, + "ewc_loss_diag": 1.646578311920166e-06, + "ewc_loss_parallel": 1.5720725059509277e-06, + "grad_norm": 4.515969276428223, + "learning_rate": 2.3230182280627383e-07, + "loss": 0.4421, + "mean_token_accuracy": 0.8568742275238037, + "num_tokens": 21224945.0, + "step": 549 + }, + { + "epoch": 0.06996565322478056, + "ewc_loss": 0.0017547607421875, + "ewc_loss_diag": 1.646578311920166e-06, + "ewc_loss_parallel": 1.5869736671447754e-06, + "grad_norm": 4.102452754974365, + "learning_rate": 2.327257312420517e-07, + "loss": 0.418, + "mean_token_accuracy": 0.860891580581665, + "num_tokens": 21261630.0, + "step": 550 + }, + { + "epoch": 0.07009286350337107, + "ewc_loss": 0.00171661376953125, + "ewc_loss_diag": 1.646578311920166e-06, + "ewc_loss_parallel": 1.5497207641601562e-06, + "grad_norm": 4.039942741394043, + "learning_rate": 2.3314963967782957e-07, + "loss": 0.4495, + "mean_token_accuracy": 0.8515312671661377, + "num_tokens": 21298904.0, + "step": 551 + }, + { + "epoch": 0.07022007378196159, + "ewc_loss": 0.001708984375, + "ewc_loss_diag": 1.646578311920166e-06, + "ewc_loss_parallel": 1.5422701835632324e-06, + "grad_norm": 3.9062976837158203, + "learning_rate": 2.3357354811360745e-07, + "loss": 0.4707, + "mean_token_accuracy": 0.8486614227294922, + "num_tokens": 21343050.0, + "step": 552 + }, + { + "epoch": 0.07034728406055209, + "ewc_loss": 0.0016937255859375, + "ewc_loss_diag": 1.646578311920166e-06, + "ewc_loss_parallel": 1.5273690223693848e-06, + "grad_norm": 4.320732593536377, + "learning_rate": 2.3399745654938532e-07, + "loss": 0.4754, + "mean_token_accuracy": 0.8456447124481201, + "num_tokens": 21376089.0, + "step": 553 + }, + { + "epoch": 0.0704744943391426, + "ewc_loss": 0.001739501953125, + "ewc_loss_diag": 1.6540288925170898e-06, + "ewc_loss_parallel": 1.5720725059509277e-06, + "grad_norm": 4.234564304351807, + "learning_rate": 2.344213649851632e-07, + "loss": 0.469, + "mean_token_accuracy": 0.8488702774047852, + "num_tokens": 21413164.0, + "step": 554 + }, + { + "epoch": 0.07060170461773312, + "ewc_loss": 0.00173187255859375, + "ewc_loss_diag": 1.6540288925170898e-06, + "ewc_loss_parallel": 1.564621925354004e-06, + "grad_norm": 4.036349773406982, + "learning_rate": 2.3484527342094106e-07, + "loss": 0.543, + "mean_token_accuracy": 0.829637348651886, + "num_tokens": 21457259.0, + "step": 555 + }, + { + "epoch": 0.07072891489632362, + "ewc_loss": 0.001708984375, + "ewc_loss_diag": 1.6540288925170898e-06, + "ewc_loss_parallel": 1.5422701835632324e-06, + "grad_norm": 4.127403259277344, + "learning_rate": 2.3526918185671894e-07, + "loss": 0.5005, + "mean_token_accuracy": 0.8371353149414062, + "num_tokens": 21494636.0, + "step": 556 + }, + { + "epoch": 0.07085612517491413, + "ewc_loss": 0.001739501953125, + "ewc_loss_diag": 1.6540288925170898e-06, + "ewc_loss_parallel": 1.5720725059509277e-06, + "grad_norm": 4.021549224853516, + "learning_rate": 2.356930902924968e-07, + "loss": 0.4422, + "mean_token_accuracy": 0.8574621677398682, + "num_tokens": 21535347.0, + "step": 557 + }, + { + "epoch": 0.07098333545350465, + "ewc_loss": 0.00173187255859375, + "ewc_loss_diag": 1.6540288925170898e-06, + "ewc_loss_parallel": 1.564621925354004e-06, + "grad_norm": 3.9826738834381104, + "learning_rate": 2.3611699872827468e-07, + "loss": 0.5099, + "mean_token_accuracy": 0.834483802318573, + "num_tokens": 21580332.0, + "step": 558 + }, + { + "epoch": 0.07111054573209516, + "ewc_loss": 0.00171661376953125, + "ewc_loss_diag": 1.6540288925170898e-06, + "ewc_loss_parallel": 1.5497207641601562e-06, + "grad_norm": 4.162974834442139, + "learning_rate": 2.3654090716405255e-07, + "loss": 0.4638, + "mean_token_accuracy": 0.8488222360610962, + "num_tokens": 21618942.0, + "step": 559 + }, + { + "epoch": 0.07123775601068566, + "ewc_loss": 0.001739501953125, + "ewc_loss_diag": 1.6540288925170898e-06, + "ewc_loss_parallel": 1.5720725059509277e-06, + "grad_norm": 4.392390251159668, + "learning_rate": 2.3696481559983043e-07, + "loss": 0.4984, + "mean_token_accuracy": 0.8385518193244934, + "num_tokens": 21652180.0, + "step": 560 + }, + { + "epoch": 0.07136496628927617, + "ewc_loss": 0.0017852783203125, + "ewc_loss_diag": 1.6540288925170898e-06, + "ewc_loss_parallel": 1.6167759895324707e-06, + "grad_norm": 4.318714618682861, + "learning_rate": 2.373887240356083e-07, + "loss": 0.4945, + "mean_token_accuracy": 0.8387592434883118, + "num_tokens": 21689148.0, + "step": 561 + }, + { + "epoch": 0.07149217656786669, + "ewc_loss": 0.00177001953125, + "ewc_loss_diag": 1.6540288925170898e-06, + "ewc_loss_parallel": 1.601874828338623e-06, + "grad_norm": 4.181417942047119, + "learning_rate": 2.3781263247138617e-07, + "loss": 0.4561, + "mean_token_accuracy": 0.8512731790542603, + "num_tokens": 21726813.0, + "step": 562 + }, + { + "epoch": 0.07161938684645719, + "ewc_loss": 0.0017547607421875, + "ewc_loss_diag": 1.6614794731140137e-06, + "ewc_loss_parallel": 1.5869736671447754e-06, + "grad_norm": 4.249698638916016, + "learning_rate": 2.3823654090716404e-07, + "loss": 0.4667, + "mean_token_accuracy": 0.8508522510528564, + "num_tokens": 21762898.0, + "step": 563 + }, + { + "epoch": 0.0717465971250477, + "ewc_loss": 0.00176239013671875, + "ewc_loss_diag": 1.6614794731140137e-06, + "ewc_loss_parallel": 1.5944242477416992e-06, + "grad_norm": 4.049413681030273, + "learning_rate": 2.386604493429419e-07, + "loss": 0.4062, + "mean_token_accuracy": 0.8619323968887329, + "num_tokens": 21800263.0, + "step": 564 + }, + { + "epoch": 0.07187380740363822, + "ewc_loss": 0.001739501953125, + "ewc_loss_diag": 1.6540288925170898e-06, + "ewc_loss_parallel": 1.5720725059509277e-06, + "grad_norm": 4.174753189086914, + "learning_rate": 2.390843577787198e-07, + "loss": 0.4958, + "mean_token_accuracy": 0.8370048403739929, + "num_tokens": 21838972.0, + "step": 565 + }, + { + "epoch": 0.07200101768222872, + "ewc_loss": 0.00176239013671875, + "ewc_loss_diag": 1.6540288925170898e-06, + "ewc_loss_parallel": 1.5944242477416992e-06, + "grad_norm": 3.926851511001587, + "learning_rate": 2.3950826621449766e-07, + "loss": 0.3857, + "mean_token_accuracy": 0.8720824122428894, + "num_tokens": 21883733.0, + "step": 566 + }, + { + "epoch": 0.07212822796081923, + "ewc_loss": 0.001739501953125, + "ewc_loss_diag": 1.6763806343078613e-06, + "ewc_loss_parallel": 1.5720725059509277e-06, + "grad_norm": 4.050543308258057, + "learning_rate": 2.3993217465027556e-07, + "loss": 0.4642, + "mean_token_accuracy": 0.8505075573921204, + "num_tokens": 21925107.0, + "step": 567 + }, + { + "epoch": 0.07225543823940975, + "ewc_loss": 0.0017547607421875, + "ewc_loss_diag": 1.6763806343078613e-06, + "ewc_loss_parallel": 1.5869736671447754e-06, + "grad_norm": 4.23077392578125, + "learning_rate": 2.403560830860534e-07, + "loss": 0.4963, + "mean_token_accuracy": 0.8415454626083374, + "num_tokens": 21966026.0, + "step": 568 + }, + { + "epoch": 0.07238264851800025, + "ewc_loss": 0.00177764892578125, + "ewc_loss_diag": 1.6763806343078613e-06, + "ewc_loss_parallel": 1.6093254089355469e-06, + "grad_norm": 4.431772708892822, + "learning_rate": 2.4077999152183125e-07, + "loss": 0.5219, + "mean_token_accuracy": 0.8309879302978516, + "num_tokens": 22003881.0, + "step": 569 + }, + { + "epoch": 0.07250985879659076, + "ewc_loss": 0.00182342529296875, + "ewc_loss_diag": 1.6763806343078613e-06, + "ewc_loss_parallel": 1.6540288925170898e-06, + "grad_norm": 4.133216381072998, + "learning_rate": 2.4120389995760915e-07, + "loss": 0.4245, + "mean_token_accuracy": 0.862604022026062, + "num_tokens": 22042100.0, + "step": 570 + }, + { + "epoch": 0.07263706907518128, + "ewc_loss": 0.0017852783203125, + "ewc_loss_diag": 1.6763806343078613e-06, + "ewc_loss_parallel": 1.6167759895324707e-06, + "grad_norm": 3.951028823852539, + "learning_rate": 2.41627808393387e-07, + "loss": 0.4433, + "mean_token_accuracy": 0.8533610701560974, + "num_tokens": 22087335.0, + "step": 571 + }, + { + "epoch": 0.07276427935377179, + "ewc_loss": 0.0017547607421875, + "ewc_loss_diag": 1.6763806343078613e-06, + "ewc_loss_parallel": 1.5869736671447754e-06, + "grad_norm": 4.277191638946533, + "learning_rate": 2.420517168291649e-07, + "loss": 0.4579, + "mean_token_accuracy": 0.854519248008728, + "num_tokens": 22121377.0, + "step": 572 + }, + { + "epoch": 0.07289148963236229, + "ewc_loss": 0.0017852783203125, + "ewc_loss_diag": 1.6763806343078613e-06, + "ewc_loss_parallel": 1.6167759895324707e-06, + "grad_norm": 4.191690921783447, + "learning_rate": 2.4247562526494274e-07, + "loss": 0.4145, + "mean_token_accuracy": 0.86201012134552, + "num_tokens": 22155492.0, + "step": 573 + }, + { + "epoch": 0.0730186999109528, + "ewc_loss": 0.00179290771484375, + "ewc_loss_diag": 1.691281795501709e-06, + "ewc_loss_parallel": 1.6242265701293945e-06, + "grad_norm": 4.111926078796387, + "learning_rate": 2.4289953370072064e-07, + "loss": 0.478, + "mean_token_accuracy": 0.8437983989715576, + "num_tokens": 22195721.0, + "step": 574 + }, + { + "epoch": 0.07314591018954332, + "ewc_loss": 0.00177764892578125, + "ewc_loss_diag": 1.691281795501709e-06, + "ewc_loss_parallel": 1.6093254089355469e-06, + "grad_norm": 4.13964319229126, + "learning_rate": 2.433234421364985e-07, + "loss": 0.4475, + "mean_token_accuracy": 0.8570001721382141, + "num_tokens": 22233582.0, + "step": 575 + }, + { + "epoch": 0.07327312046813382, + "ewc_loss": 0.00179290771484375, + "ewc_loss_diag": 1.691281795501709e-06, + "ewc_loss_parallel": 1.6242265701293945e-06, + "grad_norm": 4.185166835784912, + "learning_rate": 2.437473505722764e-07, + "loss": 0.4133, + "mean_token_accuracy": 0.8624414205551147, + "num_tokens": 22271494.0, + "step": 576 + }, + { + "epoch": 0.07340033074672433, + "ewc_loss": 0.00179290771484375, + "ewc_loss_diag": 1.691281795501709e-06, + "ewc_loss_parallel": 1.6242265701293945e-06, + "grad_norm": 4.236614227294922, + "learning_rate": 2.4417125900805423e-07, + "loss": 0.4386, + "mean_token_accuracy": 0.851931095123291, + "num_tokens": 22308063.0, + "step": 577 + }, + { + "epoch": 0.07352754102531485, + "ewc_loss": 0.00180816650390625, + "ewc_loss_diag": 1.6987323760986328e-06, + "ewc_loss_parallel": 1.6391277313232422e-06, + "grad_norm": 4.272030830383301, + "learning_rate": 2.4459516744383213e-07, + "loss": 0.4809, + "mean_token_accuracy": 0.8440300822257996, + "num_tokens": 22343652.0, + "step": 578 + }, + { + "epoch": 0.07365475130390535, + "ewc_loss": 0.001800537109375, + "ewc_loss_diag": 1.7136335372924805e-06, + "ewc_loss_parallel": 1.6316771507263184e-06, + "grad_norm": 4.167291164398193, + "learning_rate": 2.4501907587961e-07, + "loss": 0.4358, + "mean_token_accuracy": 0.8565847873687744, + "num_tokens": 22379421.0, + "step": 579 + }, + { + "epoch": 0.07378196158249586, + "ewc_loss": 0.00182342529296875, + "ewc_loss_diag": 1.7285346984863281e-06, + "ewc_loss_parallel": 1.646578311920166e-06, + "grad_norm": 4.3421831130981445, + "learning_rate": 2.454429843153879e-07, + "loss": 0.548, + "mean_token_accuracy": 0.8246668577194214, + "num_tokens": 22419276.0, + "step": 580 + }, + { + "epoch": 0.07390917186108638, + "ewc_loss": 0.0018310546875, + "ewc_loss_diag": 1.7285346984863281e-06, + "ewc_loss_parallel": 1.6540288925170898e-06, + "grad_norm": 4.3821539878845215, + "learning_rate": 2.458668927511657e-07, + "loss": 0.4499, + "mean_token_accuracy": 0.8466086387634277, + "num_tokens": 22452337.0, + "step": 581 + }, + { + "epoch": 0.07403638213967688, + "ewc_loss": 0.00183868408203125, + "ewc_loss_diag": 1.7285346984863281e-06, + "ewc_loss_parallel": 1.6614794731140137e-06, + "grad_norm": 4.134225368499756, + "learning_rate": 2.462908011869436e-07, + "loss": 0.4226, + "mean_token_accuracy": 0.8601007461547852, + "num_tokens": 22491457.0, + "step": 582 + }, + { + "epoch": 0.0741635924182674, + "ewc_loss": 0.00180816650390625, + "ewc_loss_diag": 1.7285346984863281e-06, + "ewc_loss_parallel": 1.6316771507263184e-06, + "grad_norm": 4.072083473205566, + "learning_rate": 2.4671470962272147e-07, + "loss": 0.4381, + "mean_token_accuracy": 0.856719970703125, + "num_tokens": 22534006.0, + "step": 583 + }, + { + "epoch": 0.07429080269685791, + "ewc_loss": 0.001800537109375, + "ewc_loss_diag": 1.735985279083252e-06, + "ewc_loss_parallel": 1.6242265701293945e-06, + "grad_norm": 4.134252071380615, + "learning_rate": 2.4713861805849937e-07, + "loss": 0.4102, + "mean_token_accuracy": 0.8642500638961792, + "num_tokens": 22572499.0, + "step": 584 + }, + { + "epoch": 0.07441801297544842, + "ewc_loss": 0.00182342529296875, + "ewc_loss_diag": 1.7434358596801758e-06, + "ewc_loss_parallel": 1.646578311920166e-06, + "grad_norm": 4.108099460601807, + "learning_rate": 2.475625264942772e-07, + "loss": 0.431, + "mean_token_accuracy": 0.857848584651947, + "num_tokens": 22612788.0, + "step": 585 + }, + { + "epoch": 0.07454522325403892, + "ewc_loss": 0.00180816650390625, + "ewc_loss_diag": 1.7434358596801758e-06, + "ewc_loss_parallel": 1.6316771507263184e-06, + "grad_norm": 4.295751094818115, + "learning_rate": 2.479864349300551e-07, + "loss": 0.4582, + "mean_token_accuracy": 0.8546894788742065, + "num_tokens": 22649347.0, + "step": 586 + }, + { + "epoch": 0.07467243353262944, + "ewc_loss": 0.00183868408203125, + "ewc_loss_diag": 1.7434358596801758e-06, + "ewc_loss_parallel": 1.6614794731140137e-06, + "grad_norm": 4.479182243347168, + "learning_rate": 2.4841034336583296e-07, + "loss": 0.422, + "mean_token_accuracy": 0.8540608882904053, + "num_tokens": 22681365.0, + "step": 587 + }, + { + "epoch": 0.07479964381121995, + "ewc_loss": 0.001861572265625, + "ewc_loss_diag": 1.7508864402770996e-06, + "ewc_loss_parallel": 1.6838312149047852e-06, + "grad_norm": 4.223294258117676, + "learning_rate": 2.488342518016108e-07, + "loss": 0.4854, + "mean_token_accuracy": 0.8414573669433594, + "num_tokens": 22719803.0, + "step": 588 + }, + { + "epoch": 0.07492685408981045, + "ewc_loss": 0.0018310546875, + "ewc_loss_diag": 1.7508864402770996e-06, + "ewc_loss_parallel": 1.6540288925170898e-06, + "grad_norm": 4.150721073150635, + "learning_rate": 2.492581602373887e-07, + "loss": 0.4319, + "mean_token_accuracy": 0.8579552173614502, + "num_tokens": 22758871.0, + "step": 589 + }, + { + "epoch": 0.07505406436840097, + "ewc_loss": 0.0018310546875, + "ewc_loss_diag": 1.7508864402770996e-06, + "ewc_loss_parallel": 1.6540288925170898e-06, + "grad_norm": 4.0946478843688965, + "learning_rate": 2.4968206867316655e-07, + "loss": 0.4714, + "mean_token_accuracy": 0.8474628925323486, + "num_tokens": 22804357.0, + "step": 590 + }, + { + "epoch": 0.07518127464699148, + "ewc_loss": 0.00182342529296875, + "ewc_loss_diag": 1.7508864402770996e-06, + "ewc_loss_parallel": 1.646578311920166e-06, + "grad_norm": 4.2717108726501465, + "learning_rate": 2.5010597710894445e-07, + "loss": 0.3833, + "mean_token_accuracy": 0.8758425712585449, + "num_tokens": 22836330.0, + "step": 591 + }, + { + "epoch": 0.07530848492558198, + "ewc_loss": 0.0018463134765625, + "ewc_loss_diag": 1.7657876014709473e-06, + "ewc_loss_parallel": 1.6689300537109375e-06, + "grad_norm": 4.177026271820068, + "learning_rate": 2.505298855447223e-07, + "loss": 0.4493, + "mean_token_accuracy": 0.8535294532775879, + "num_tokens": 22878494.0, + "step": 592 + }, + { + "epoch": 0.0754356952041725, + "ewc_loss": 0.0018463134765625, + "ewc_loss_diag": 1.7657876014709473e-06, + "ewc_loss_parallel": 1.6689300537109375e-06, + "grad_norm": 4.1720499992370605, + "learning_rate": 2.509537939805002e-07, + "loss": 0.4543, + "mean_token_accuracy": 0.8515774607658386, + "num_tokens": 22920406.0, + "step": 593 + }, + { + "epoch": 0.07556290548276301, + "ewc_loss": 0.00183868408203125, + "ewc_loss_diag": 1.7657876014709473e-06, + "ewc_loss_parallel": 1.6614794731140137e-06, + "grad_norm": 4.443252086639404, + "learning_rate": 2.513777024162781e-07, + "loss": 0.4356, + "mean_token_accuracy": 0.8567330837249756, + "num_tokens": 22952318.0, + "step": 594 + }, + { + "epoch": 0.07569011576135352, + "ewc_loss": 0.00189971923828125, + "ewc_loss_diag": 1.7657876014709473e-06, + "ewc_loss_parallel": 1.7210841178894043e-06, + "grad_norm": 4.142069339752197, + "learning_rate": 2.5180161085205594e-07, + "loss": 0.4368, + "mean_token_accuracy": 0.857137143611908, + "num_tokens": 22990532.0, + "step": 595 + }, + { + "epoch": 0.07581732603994402, + "ewc_loss": 0.0018463134765625, + "ewc_loss_diag": 1.7657876014709473e-06, + "ewc_loss_parallel": 1.6689300537109375e-06, + "grad_norm": 4.106148719787598, + "learning_rate": 2.522255192878338e-07, + "loss": 0.4732, + "mean_token_accuracy": 0.8460907936096191, + "num_tokens": 23029157.0, + "step": 596 + }, + { + "epoch": 0.07594453631853454, + "ewc_loss": 0.0018463134765625, + "ewc_loss_diag": 1.7657876014709473e-06, + "ewc_loss_parallel": 1.6689300537109375e-06, + "grad_norm": 4.180429458618164, + "learning_rate": 2.526494277236117e-07, + "loss": 0.4563, + "mean_token_accuracy": 0.8535597920417786, + "num_tokens": 23065449.0, + "step": 597 + }, + { + "epoch": 0.07607174659712505, + "ewc_loss": 0.001861572265625, + "ewc_loss_diag": 1.7657876014709473e-06, + "ewc_loss_parallel": 1.6838312149047852e-06, + "grad_norm": 4.207340717315674, + "learning_rate": 2.530733361593896e-07, + "loss": 0.4386, + "mean_token_accuracy": 0.8550878763198853, + "num_tokens": 23103604.0, + "step": 598 + }, + { + "epoch": 0.07619895687571555, + "ewc_loss": 0.00186920166015625, + "ewc_loss_diag": 1.7657876014709473e-06, + "ewc_loss_parallel": 1.691281795501709e-06, + "grad_norm": 4.1495513916015625, + "learning_rate": 2.5349724459516743e-07, + "loss": 0.4528, + "mean_token_accuracy": 0.8517251014709473, + "num_tokens": 23142387.0, + "step": 599 + }, + { + "epoch": 0.07632616715430607, + "ewc_loss": 0.001861572265625, + "ewc_loss_diag": 1.773238182067871e-06, + "ewc_loss_parallel": 1.6838312149047852e-06, + "grad_norm": 4.293604373931885, + "learning_rate": 2.539211530309453e-07, + "loss": 0.4107, + "mean_token_accuracy": 0.8639712333679199, + "num_tokens": 23174471.0, + "step": 600 + }, + { + "epoch": 0.07645337743289658, + "ewc_loss": 0.00188446044921875, + "ewc_loss_diag": 1.773238182067871e-06, + "ewc_loss_parallel": 1.7061829566955566e-06, + "grad_norm": 4.508098125457764, + "learning_rate": 2.543450614667232e-07, + "loss": 0.4918, + "mean_token_accuracy": 0.840453028678894, + "num_tokens": 23207171.0, + "step": 601 + }, + { + "epoch": 0.07658058771148708, + "ewc_loss": 0.0019073486328125, + "ewc_loss_diag": 1.7881393432617188e-06, + "ewc_loss_parallel": 1.7285346984863281e-06, + "grad_norm": 4.340343475341797, + "learning_rate": 2.547689699025011e-07, + "loss": 0.5058, + "mean_token_accuracy": 0.8343265056610107, + "num_tokens": 23244739.0, + "step": 602 + }, + { + "epoch": 0.0767077979900776, + "ewc_loss": 0.001861572265625, + "ewc_loss_diag": 1.7881393432617188e-06, + "ewc_loss_parallel": 1.6838312149047852e-06, + "grad_norm": 4.052517414093018, + "learning_rate": 2.551928783382789e-07, + "loss": 0.4524, + "mean_token_accuracy": 0.8514273762702942, + "num_tokens": 23286710.0, + "step": 603 + }, + { + "epoch": 0.07683500826866811, + "ewc_loss": 0.0018463134765625, + "ewc_loss_diag": 1.780688762664795e-06, + "ewc_loss_parallel": 1.6689300537109375e-06, + "grad_norm": 4.486096382141113, + "learning_rate": 2.5561678677405677e-07, + "loss": 0.4255, + "mean_token_accuracy": 0.8582198619842529, + "num_tokens": 23316485.0, + "step": 604 + }, + { + "epoch": 0.07696221854725861, + "ewc_loss": 0.0019073486328125, + "ewc_loss_diag": 1.780688762664795e-06, + "ewc_loss_parallel": 1.7285346984863281e-06, + "grad_norm": 4.323619842529297, + "learning_rate": 2.5604069520983467e-07, + "loss": 0.4783, + "mean_token_accuracy": 0.8435865044593811, + "num_tokens": 23354893.0, + "step": 605 + }, + { + "epoch": 0.07708942882584913, + "ewc_loss": 0.00189208984375, + "ewc_loss_diag": 1.780688762664795e-06, + "ewc_loss_parallel": 1.7136335372924805e-06, + "grad_norm": 4.0138115882873535, + "learning_rate": 2.564646036456125e-07, + "loss": 0.4442, + "mean_token_accuracy": 0.8539012670516968, + "num_tokens": 23399987.0, + "step": 606 + }, + { + "epoch": 0.07721663910443964, + "ewc_loss": 0.00185394287109375, + "ewc_loss_diag": 1.780688762664795e-06, + "ewc_loss_parallel": 1.6763806343078613e-06, + "grad_norm": 4.2260613441467285, + "learning_rate": 2.568885120813904e-07, + "loss": 0.4425, + "mean_token_accuracy": 0.8530794978141785, + "num_tokens": 23438140.0, + "step": 607 + }, + { + "epoch": 0.07734384938303016, + "ewc_loss": 0.0019073486328125, + "ewc_loss_diag": 1.7955899238586426e-06, + "ewc_loss_parallel": 1.7285346984863281e-06, + "grad_norm": 4.206136226654053, + "learning_rate": 2.5731242051716826e-07, + "loss": 0.4467, + "mean_token_accuracy": 0.8563601970672607, + "num_tokens": 23476916.0, + "step": 608 + }, + { + "epoch": 0.07747105966162066, + "ewc_loss": 0.0019073486328125, + "ewc_loss_diag": 1.7955899238586426e-06, + "ewc_loss_parallel": 1.7210841178894043e-06, + "grad_norm": 4.2475762367248535, + "learning_rate": 2.5773632895294616e-07, + "loss": 0.4726, + "mean_token_accuracy": 0.8462111949920654, + "num_tokens": 23515260.0, + "step": 609 + }, + { + "epoch": 0.07759826994021117, + "ewc_loss": 0.0019073486328125, + "ewc_loss_diag": 1.7955899238586426e-06, + "ewc_loss_parallel": 1.7210841178894043e-06, + "grad_norm": 4.277976989746094, + "learning_rate": 2.58160237388724e-07, + "loss": 0.4007, + "mean_token_accuracy": 0.8654235601425171, + "num_tokens": 23546964.0, + "step": 610 + }, + { + "epoch": 0.07772548021880168, + "ewc_loss": 0.001922607421875, + "ewc_loss_diag": 1.7955899238586426e-06, + "ewc_loss_parallel": 1.7508864402770996e-06, + "grad_norm": 4.241230010986328, + "learning_rate": 2.585841458245019e-07, + "loss": 0.4397, + "mean_token_accuracy": 0.8547102212905884, + "num_tokens": 23583328.0, + "step": 611 + }, + { + "epoch": 0.07785269049739219, + "ewc_loss": 0.001922607421875, + "ewc_loss_diag": 1.8030405044555664e-06, + "ewc_loss_parallel": 1.735985279083252e-06, + "grad_norm": 4.282676696777344, + "learning_rate": 2.5900805426027975e-07, + "loss": 0.394, + "mean_token_accuracy": 0.8665364384651184, + "num_tokens": 23617174.0, + "step": 612 + }, + { + "epoch": 0.0779799007759827, + "ewc_loss": 0.00193023681640625, + "ewc_loss_diag": 1.8030405044555664e-06, + "ewc_loss_parallel": 1.7434358596801758e-06, + "grad_norm": 4.353269577026367, + "learning_rate": 2.5943196269605765e-07, + "loss": 0.4668, + "mean_token_accuracy": 0.8530626893043518, + "num_tokens": 23655990.0, + "step": 613 + }, + { + "epoch": 0.07810711105457321, + "ewc_loss": 0.0019378662109375, + "ewc_loss_diag": 1.8030405044555664e-06, + "ewc_loss_parallel": 1.7583370208740234e-06, + "grad_norm": 4.0377326011657715, + "learning_rate": 2.598558711318355e-07, + "loss": 0.4549, + "mean_token_accuracy": 0.8537740111351013, + "num_tokens": 23702932.0, + "step": 614 + }, + { + "epoch": 0.07823432133316371, + "ewc_loss": 0.00186920166015625, + "ewc_loss_diag": 1.8030405044555664e-06, + "ewc_loss_parallel": 1.6838312149047852e-06, + "grad_norm": 4.206052780151367, + "learning_rate": 2.602797795676134e-07, + "loss": 0.4759, + "mean_token_accuracy": 0.8454506397247314, + "num_tokens": 23741504.0, + "step": 615 + }, + { + "epoch": 0.07836153161175423, + "ewc_loss": 0.00193023681640625, + "ewc_loss_diag": 1.817941665649414e-06, + "ewc_loss_parallel": 1.7434358596801758e-06, + "grad_norm": 4.34000301361084, + "learning_rate": 2.6070368800339124e-07, + "loss": 0.4675, + "mean_token_accuracy": 0.8440254330635071, + "num_tokens": 23776014.0, + "step": 616 + }, + { + "epoch": 0.07848874189034474, + "ewc_loss": 0.0019378662109375, + "ewc_loss_diag": 1.817941665649414e-06, + "ewc_loss_parallel": 1.7583370208740234e-06, + "grad_norm": 4.418918132781982, + "learning_rate": 2.6112759643916914e-07, + "loss": 0.4286, + "mean_token_accuracy": 0.8591986298561096, + "num_tokens": 23807412.0, + "step": 617 + }, + { + "epoch": 0.07861595216893524, + "ewc_loss": 0.001953125, + "ewc_loss_diag": 1.8104910850524902e-06, + "ewc_loss_parallel": 1.773238182067871e-06, + "grad_norm": 4.421381950378418, + "learning_rate": 2.61551504874947e-07, + "loss": 0.4371, + "mean_token_accuracy": 0.8552342057228088, + "num_tokens": 23840842.0, + "step": 618 + }, + { + "epoch": 0.07874316244752576, + "ewc_loss": 0.0019683837890625, + "ewc_loss_diag": 1.817941665649414e-06, + "ewc_loss_parallel": 1.7881393432617188e-06, + "grad_norm": 4.144287586212158, + "learning_rate": 2.619754133107249e-07, + "loss": 0.4321, + "mean_token_accuracy": 0.8563964366912842, + "num_tokens": 23882832.0, + "step": 619 + }, + { + "epoch": 0.07887037272611627, + "ewc_loss": 0.0019378662109375, + "ewc_loss_diag": 1.8104910850524902e-06, + "ewc_loss_parallel": 1.7583370208740234e-06, + "grad_norm": 4.325422286987305, + "learning_rate": 2.623993217465028e-07, + "loss": 0.4729, + "mean_token_accuracy": 0.842217206954956, + "num_tokens": 23917558.0, + "step": 620 + }, + { + "epoch": 0.07899758300470679, + "ewc_loss": 0.00194549560546875, + "ewc_loss_diag": 1.817941665649414e-06, + "ewc_loss_parallel": 1.7657876014709473e-06, + "grad_norm": 4.1219658851623535, + "learning_rate": 2.6282323018228063e-07, + "loss": 0.4389, + "mean_token_accuracy": 0.8569743633270264, + "num_tokens": 23959217.0, + "step": 621 + }, + { + "epoch": 0.07912479328329729, + "ewc_loss": 0.00193023681640625, + "ewc_loss_diag": 1.817941665649414e-06, + "ewc_loss_parallel": 1.7434358596801758e-06, + "grad_norm": 4.176352500915527, + "learning_rate": 2.632471386180585e-07, + "loss": 0.4843, + "mean_token_accuracy": 0.8418596982955933, + "num_tokens": 23999238.0, + "step": 622 + }, + { + "epoch": 0.0792520035618878, + "ewc_loss": 0.0019378662109375, + "ewc_loss_diag": 1.8253922462463379e-06, + "ewc_loss_parallel": 1.7583370208740234e-06, + "grad_norm": 4.498849868774414, + "learning_rate": 2.6367104705383637e-07, + "loss": 0.4965, + "mean_token_accuracy": 0.8368397355079651, + "num_tokens": 24034030.0, + "step": 623 + }, + { + "epoch": 0.07937921384047832, + "ewc_loss": 0.001983642578125, + "ewc_loss_diag": 1.8253922462463379e-06, + "ewc_loss_parallel": 1.8030405044555664e-06, + "grad_norm": 4.205366134643555, + "learning_rate": 2.6409495548961427e-07, + "loss": 0.4828, + "mean_token_accuracy": 0.8451204299926758, + "num_tokens": 24076940.0, + "step": 624 + }, + { + "epoch": 0.07950642411906882, + "ewc_loss": 0.0019683837890625, + "ewc_loss_diag": 1.817941665649414e-06, + "ewc_loss_parallel": 1.7881393432617188e-06, + "grad_norm": 4.519134998321533, + "learning_rate": 2.6451886392539206e-07, + "loss": 0.4671, + "mean_token_accuracy": 0.8488202095031738, + "num_tokens": 24108095.0, + "step": 625 + }, + { + "epoch": 0.07963363439765933, + "ewc_loss": 0.001983642578125, + "ewc_loss_diag": 1.8253922462463379e-06, + "ewc_loss_parallel": 1.8104910850524902e-06, + "grad_norm": 4.229473114013672, + "learning_rate": 2.6494277236116996e-07, + "loss": 0.4308, + "mean_token_accuracy": 0.8557287454605103, + "num_tokens": 24149401.0, + "step": 626 + }, + { + "epoch": 0.07976084467624985, + "ewc_loss": 0.001953125, + "ewc_loss_diag": 1.8253922462463379e-06, + "ewc_loss_parallel": 1.773238182067871e-06, + "grad_norm": 4.27863073348999, + "learning_rate": 2.6536668079694786e-07, + "loss": 0.5141, + "mean_token_accuracy": 0.8329358100891113, + "num_tokens": 24188380.0, + "step": 627 + }, + { + "epoch": 0.07988805495484035, + "ewc_loss": 0.0019683837890625, + "ewc_loss_diag": 1.8253922462463379e-06, + "ewc_loss_parallel": 1.7881393432617188e-06, + "grad_norm": 4.27744722366333, + "learning_rate": 2.6579058923272576e-07, + "loss": 0.3907, + "mean_token_accuracy": 0.8718739748001099, + "num_tokens": 24222220.0, + "step": 628 + }, + { + "epoch": 0.08001526523343086, + "ewc_loss": 0.001953125, + "ewc_loss_diag": 1.8253922462463379e-06, + "ewc_loss_parallel": 1.780688762664795e-06, + "grad_norm": 4.371853828430176, + "learning_rate": 2.6621449766850356e-07, + "loss": 0.4729, + "mean_token_accuracy": 0.8451552391052246, + "num_tokens": 24258739.0, + "step": 629 + }, + { + "epoch": 0.08014247551202137, + "ewc_loss": 0.0019989013671875, + "ewc_loss_diag": 1.8328428268432617e-06, + "ewc_loss_parallel": 1.817941665649414e-06, + "grad_norm": 4.276233196258545, + "learning_rate": 2.6663840610428145e-07, + "loss": 0.4199, + "mean_token_accuracy": 0.8622139096260071, + "num_tokens": 24293732.0, + "step": 630 + }, + { + "epoch": 0.08026968579061187, + "ewc_loss": 0.001983642578125, + "ewc_loss_diag": 1.8253922462463379e-06, + "ewc_loss_parallel": 1.8104910850524902e-06, + "grad_norm": 4.14270544052124, + "learning_rate": 2.6706231454005935e-07, + "loss": 0.4224, + "mean_token_accuracy": 0.8620599508285522, + "num_tokens": 24341299.0, + "step": 631 + }, + { + "epoch": 0.08039689606920239, + "ewc_loss": 0.001953125, + "ewc_loss_diag": 1.8328428268432617e-06, + "ewc_loss_parallel": 1.773238182067871e-06, + "grad_norm": 4.117955207824707, + "learning_rate": 2.6748622297583725e-07, + "loss": 0.4409, + "mean_token_accuracy": 0.8547608256340027, + "num_tokens": 24384314.0, + "step": 632 + }, + { + "epoch": 0.0805241063477929, + "ewc_loss": 0.001953125, + "ewc_loss_diag": 1.8253922462463379e-06, + "ewc_loss_parallel": 1.780688762664795e-06, + "grad_norm": 4.347110748291016, + "learning_rate": 2.6791013141161505e-07, + "loss": 0.4501, + "mean_token_accuracy": 0.8546258807182312, + "num_tokens": 24420071.0, + "step": 633 + }, + { + "epoch": 0.08065131662638342, + "ewc_loss": 0.0019989013671875, + "ewc_loss_diag": 1.8253922462463379e-06, + "ewc_loss_parallel": 1.817941665649414e-06, + "grad_norm": 4.427215576171875, + "learning_rate": 2.6833403984739294e-07, + "loss": 0.4398, + "mean_token_accuracy": 0.8544711470603943, + "num_tokens": 24456783.0, + "step": 634 + }, + { + "epoch": 0.08077852690497392, + "ewc_loss": 0.00201416015625, + "ewc_loss_diag": 1.8253922462463379e-06, + "ewc_loss_parallel": 1.8402934074401855e-06, + "grad_norm": 4.07306432723999, + "learning_rate": 2.6875794828317084e-07, + "loss": 0.4278, + "mean_token_accuracy": 0.8628152012825012, + "num_tokens": 24502933.0, + "step": 635 + }, + { + "epoch": 0.08090573718356443, + "ewc_loss": 0.001983642578125, + "ewc_loss_diag": 1.8402934074401855e-06, + "ewc_loss_parallel": 1.7955899238586426e-06, + "grad_norm": 4.388856887817383, + "learning_rate": 2.6918185671894874e-07, + "loss": 0.4794, + "mean_token_accuracy": 0.846930980682373, + "num_tokens": 24538058.0, + "step": 636 + }, + { + "epoch": 0.08103294746215495, + "ewc_loss": 0.00201416015625, + "ewc_loss_diag": 1.8402934074401855e-06, + "ewc_loss_parallel": 1.8328428268432617e-06, + "grad_norm": 4.323983669281006, + "learning_rate": 2.6960576515472654e-07, + "loss": 0.5348, + "mean_token_accuracy": 0.8286483287811279, + "num_tokens": 24580020.0, + "step": 637 + }, + { + "epoch": 0.08116015774074545, + "ewc_loss": 0.00201416015625, + "ewc_loss_diag": 1.8402934074401855e-06, + "ewc_loss_parallel": 1.8253922462463379e-06, + "grad_norm": 4.1172990798950195, + "learning_rate": 2.7002967359050443e-07, + "loss": 0.419, + "mean_token_accuracy": 0.8620327711105347, + "num_tokens": 24622641.0, + "step": 638 + }, + { + "epoch": 0.08128736801933596, + "ewc_loss": 0.001983642578125, + "ewc_loss_diag": 1.8477439880371094e-06, + "ewc_loss_parallel": 1.7955899238586426e-06, + "grad_norm": 4.096683979034424, + "learning_rate": 2.7045358202628233e-07, + "loss": 0.4577, + "mean_token_accuracy": 0.8465542793273926, + "num_tokens": 24669853.0, + "step": 639 + }, + { + "epoch": 0.08141457829792648, + "ewc_loss": 0.001983642578125, + "ewc_loss_diag": 1.8477439880371094e-06, + "ewc_loss_parallel": 1.8030405044555664e-06, + "grad_norm": 4.476733684539795, + "learning_rate": 2.7087749046206023e-07, + "loss": 0.419, + "mean_token_accuracy": 0.8621079325675964, + "num_tokens": 24703499.0, + "step": 640 + }, + { + "epoch": 0.08154178857651698, + "ewc_loss": 0.0020294189453125, + "ewc_loss_diag": 1.8700957298278809e-06, + "ewc_loss_parallel": 1.8477439880371094e-06, + "grad_norm": 4.40878438949585, + "learning_rate": 2.71301398897838e-07, + "loss": 0.4743, + "mean_token_accuracy": 0.8433186411857605, + "num_tokens": 24741940.0, + "step": 641 + }, + { + "epoch": 0.08166899885510749, + "ewc_loss": 0.0020294189453125, + "ewc_loss_diag": 1.862645149230957e-06, + "ewc_loss_parallel": 1.8477439880371094e-06, + "grad_norm": 4.0892558097839355, + "learning_rate": 2.717253073336159e-07, + "loss": 0.3872, + "mean_token_accuracy": 0.8708286881446838, + "num_tokens": 24785109.0, + "step": 642 + }, + { + "epoch": 0.081796209133698, + "ewc_loss": 0.001983642578125, + "ewc_loss_diag": 1.8775463104248047e-06, + "ewc_loss_parallel": 1.8030405044555664e-06, + "grad_norm": 4.120628356933594, + "learning_rate": 2.721492157693938e-07, + "loss": 0.3874, + "mean_token_accuracy": 0.8706983327865601, + "num_tokens": 24827901.0, + "step": 643 + }, + { + "epoch": 0.0819234194122885, + "ewc_loss": 0.001983642578125, + "ewc_loss_diag": 1.8849968910217285e-06, + "ewc_loss_parallel": 1.7955899238586426e-06, + "grad_norm": 4.213315010070801, + "learning_rate": 2.7257312420517167e-07, + "loss": 0.4272, + "mean_token_accuracy": 0.8605637550354004, + "num_tokens": 24867500.0, + "step": 644 + }, + { + "epoch": 0.08205062969087902, + "ewc_loss": 0.0019989013671875, + "ewc_loss_diag": 1.8849968910217285e-06, + "ewc_loss_parallel": 1.8104910850524902e-06, + "grad_norm": 4.206606864929199, + "learning_rate": 2.729970326409495e-07, + "loss": 0.4576, + "mean_token_accuracy": 0.8522768020629883, + "num_tokens": 24912183.0, + "step": 645 + }, + { + "epoch": 0.08217783996946953, + "ewc_loss": 0.0019989013671875, + "ewc_loss_diag": 1.8998980522155762e-06, + "ewc_loss_parallel": 1.817941665649414e-06, + "grad_norm": 4.371002197265625, + "learning_rate": 2.734209410767274e-07, + "loss": 0.4313, + "mean_token_accuracy": 0.8560465574264526, + "num_tokens": 24946987.0, + "step": 646 + }, + { + "epoch": 0.08230505024806005, + "ewc_loss": 0.0020294189453125, + "ewc_loss_diag": 1.8998980522155762e-06, + "ewc_loss_parallel": 1.8402934074401855e-06, + "grad_norm": 4.291432857513428, + "learning_rate": 2.738448495125053e-07, + "loss": 0.4472, + "mean_token_accuracy": 0.85433030128479, + "num_tokens": 24989238.0, + "step": 647 + }, + { + "epoch": 0.08243226052665055, + "ewc_loss": 0.0020294189453125, + "ewc_loss_diag": 1.9222497940063477e-06, + "ewc_loss_parallel": 1.8402934074401855e-06, + "grad_norm": 4.567536354064941, + "learning_rate": 2.7426875794828316e-07, + "loss": 0.4618, + "mean_token_accuracy": 0.8457241058349609, + "num_tokens": 25024489.0, + "step": 648 + }, + { + "epoch": 0.08255947080524106, + "ewc_loss": 0.002044677734375, + "ewc_loss_diag": 1.9073486328125e-06, + "ewc_loss_parallel": 1.8477439880371094e-06, + "grad_norm": 4.205100059509277, + "learning_rate": 2.74692666384061e-07, + "loss": 0.4074, + "mean_token_accuracy": 0.8646705150604248, + "num_tokens": 25064228.0, + "step": 649 + }, + { + "epoch": 0.08268668108383158, + "ewc_loss": 0.0019989013671875, + "ewc_loss_diag": 1.9073486328125e-06, + "ewc_loss_parallel": 1.8104910850524902e-06, + "grad_norm": 4.209630012512207, + "learning_rate": 2.751165748198389e-07, + "loss": 0.4462, + "mean_token_accuracy": 0.8500805497169495, + "num_tokens": 25104911.0, + "step": 650 + }, + { + "epoch": 0.08281389136242208, + "ewc_loss": 0.00201416015625, + "ewc_loss_diag": 1.9073486328125e-06, + "ewc_loss_parallel": 1.8253922462463379e-06, + "grad_norm": 4.354191303253174, + "learning_rate": 2.755404832556168e-07, + "loss": 0.4506, + "mean_token_accuracy": 0.8537148237228394, + "num_tokens": 25143187.0, + "step": 651 + }, + { + "epoch": 0.0829411016410126, + "ewc_loss": 0.002044677734375, + "ewc_loss_diag": 1.9073486328125e-06, + "ewc_loss_parallel": 1.8477439880371094e-06, + "grad_norm": 4.361428260803223, + "learning_rate": 2.7596439169139465e-07, + "loss": 0.4377, + "mean_token_accuracy": 0.8580126166343689, + "num_tokens": 25180494.0, + "step": 652 + }, + { + "epoch": 0.08306831191960311, + "ewc_loss": 0.002044677734375, + "ewc_loss_diag": 1.9222497940063477e-06, + "ewc_loss_parallel": 1.8477439880371094e-06, + "grad_norm": 4.498020172119141, + "learning_rate": 2.763883001271725e-07, + "loss": 0.4884, + "mean_token_accuracy": 0.8407297134399414, + "num_tokens": 25218143.0, + "step": 653 + }, + { + "epoch": 0.08319552219819361, + "ewc_loss": 0.0020599365234375, + "ewc_loss_diag": 1.9222497940063477e-06, + "ewc_loss_parallel": 1.862645149230957e-06, + "grad_norm": 4.413679122924805, + "learning_rate": 2.768122085629504e-07, + "loss": 0.4653, + "mean_token_accuracy": 0.8446899652481079, + "num_tokens": 25254813.0, + "step": 654 + }, + { + "epoch": 0.08332273247678412, + "ewc_loss": 0.002044677734375, + "ewc_loss_diag": 1.9222497940063477e-06, + "ewc_loss_parallel": 1.8551945686340332e-06, + "grad_norm": 4.3295207023620605, + "learning_rate": 2.772361169987283e-07, + "loss": 0.4316, + "mean_token_accuracy": 0.8605014085769653, + "num_tokens": 25291138.0, + "step": 655 + }, + { + "epoch": 0.08344994275537464, + "ewc_loss": 0.0020294189453125, + "ewc_loss_diag": 1.9222497940063477e-06, + "ewc_loss_parallel": 1.8402934074401855e-06, + "grad_norm": 4.203985214233398, + "learning_rate": 2.7766002543450614e-07, + "loss": 0.4406, + "mean_token_accuracy": 0.8529159426689148, + "num_tokens": 25334648.0, + "step": 656 + }, + { + "epoch": 0.08357715303396514, + "ewc_loss": 0.0020294189453125, + "ewc_loss_diag": 1.9222497940063477e-06, + "ewc_loss_parallel": 1.8402934074401855e-06, + "grad_norm": 4.259815692901611, + "learning_rate": 2.78083933870284e-07, + "loss": 0.4978, + "mean_token_accuracy": 0.8371514081954956, + "num_tokens": 25377370.0, + "step": 657 + }, + { + "epoch": 0.08370436331255565, + "ewc_loss": 0.0020294189453125, + "ewc_loss_diag": 1.9222497940063477e-06, + "ewc_loss_parallel": 1.8402934074401855e-06, + "grad_norm": 4.466520309448242, + "learning_rate": 2.785078423060619e-07, + "loss": 0.4497, + "mean_token_accuracy": 0.8503884077072144, + "num_tokens": 25413123.0, + "step": 658 + }, + { + "epoch": 0.08383157359114617, + "ewc_loss": 0.0020751953125, + "ewc_loss_diag": 1.9222497940063477e-06, + "ewc_loss_parallel": 1.8775463104248047e-06, + "grad_norm": 4.280397415161133, + "learning_rate": 2.789317507418398e-07, + "loss": 0.4454, + "mean_token_accuracy": 0.852203369140625, + "num_tokens": 25452947.0, + "step": 659 + }, + { + "epoch": 0.08395878386973668, + "ewc_loss": 0.0020294189453125, + "ewc_loss_diag": 1.9222497940063477e-06, + "ewc_loss_parallel": 1.8402934074401855e-06, + "grad_norm": 4.317070960998535, + "learning_rate": 2.7935565917761763e-07, + "loss": 0.4241, + "mean_token_accuracy": 0.8580552339553833, + "num_tokens": 25490161.0, + "step": 660 + }, + { + "epoch": 0.08408599414832718, + "ewc_loss": 0.002044677734375, + "ewc_loss_diag": 1.9371509552001953e-06, + "ewc_loss_parallel": 1.8477439880371094e-06, + "grad_norm": 4.404637336730957, + "learning_rate": 2.797795676133955e-07, + "loss": 0.4779, + "mean_token_accuracy": 0.8439526557922363, + "num_tokens": 25528817.0, + "step": 661 + }, + { + "epoch": 0.0842132044269177, + "ewc_loss": 0.0020751953125, + "ewc_loss_diag": 1.9222497940063477e-06, + "ewc_loss_parallel": 1.8849968910217285e-06, + "grad_norm": 4.367966175079346, + "learning_rate": 2.802034760491734e-07, + "loss": 0.3766, + "mean_token_accuracy": 0.8759845495223999, + "num_tokens": 25565209.0, + "step": 662 + }, + { + "epoch": 0.08434041470550821, + "ewc_loss": 0.0020751953125, + "ewc_loss_diag": 1.9222497940063477e-06, + "ewc_loss_parallel": 1.8849968910217285e-06, + "grad_norm": 4.286657333374023, + "learning_rate": 2.806273844849512e-07, + "loss": 0.403, + "mean_token_accuracy": 0.8669461011886597, + "num_tokens": 25605600.0, + "step": 663 + }, + { + "epoch": 0.08446762498409871, + "ewc_loss": 0.0020599365234375, + "ewc_loss_diag": 1.9222497940063477e-06, + "ewc_loss_parallel": 1.8700957298278809e-06, + "grad_norm": 4.494673728942871, + "learning_rate": 2.810512929207291e-07, + "loss": 0.458, + "mean_token_accuracy": 0.8460433483123779, + "num_tokens": 25645371.0, + "step": 664 + }, + { + "epoch": 0.08459483526268922, + "ewc_loss": 0.002105712890625, + "ewc_loss_diag": 1.9371509552001953e-06, + "ewc_loss_parallel": 1.9073486328125e-06, + "grad_norm": 4.325600624084473, + "learning_rate": 2.8147520135650697e-07, + "loss": 0.4213, + "mean_token_accuracy": 0.8595209121704102, + "num_tokens": 25681763.0, + "step": 665 + }, + { + "epoch": 0.08472204554127974, + "ewc_loss": 0.0020599365234375, + "ewc_loss_diag": 1.9371509552001953e-06, + "ewc_loss_parallel": 1.862645149230957e-06, + "grad_norm": 4.339459419250488, + "learning_rate": 2.8189910979228487e-07, + "loss": 0.4573, + "mean_token_accuracy": 0.8532427549362183, + "num_tokens": 25720709.0, + "step": 666 + }, + { + "epoch": 0.08484925581987024, + "ewc_loss": 0.0020904541015625, + "ewc_loss_diag": 1.9371509552001953e-06, + "ewc_loss_parallel": 1.8924474716186523e-06, + "grad_norm": 4.557458400726318, + "learning_rate": 2.823230182280627e-07, + "loss": 0.4146, + "mean_token_accuracy": 0.8655015230178833, + "num_tokens": 25753360.0, + "step": 667 + }, + { + "epoch": 0.08497646609846075, + "ewc_loss": 0.0021209716796875, + "ewc_loss_diag": 1.952052116394043e-06, + "ewc_loss_parallel": 1.9222497940063477e-06, + "grad_norm": 4.2667436599731445, + "learning_rate": 2.827469266638406e-07, + "loss": 0.4341, + "mean_token_accuracy": 0.8561649322509766, + "num_tokens": 25796050.0, + "step": 668 + }, + { + "epoch": 0.08510367637705127, + "ewc_loss": 0.0020751953125, + "ewc_loss_diag": 1.952052116394043e-06, + "ewc_loss_parallel": 1.8775463104248047e-06, + "grad_norm": 4.248732566833496, + "learning_rate": 2.8317083509961846e-07, + "loss": 0.4506, + "mean_token_accuracy": 0.8522270321846008, + "num_tokens": 25840853.0, + "step": 669 + }, + { + "epoch": 0.08523088665564178, + "ewc_loss": 0.0020599365234375, + "ewc_loss_diag": 1.952052116394043e-06, + "ewc_loss_parallel": 1.8700957298278809e-06, + "grad_norm": 4.260396957397461, + "learning_rate": 2.8359474353539636e-07, + "loss": 0.422, + "mean_token_accuracy": 0.8606446385383606, + "num_tokens": 25882265.0, + "step": 670 + }, + { + "epoch": 0.08535809693423228, + "ewc_loss": 0.0020904541015625, + "ewc_loss_diag": 1.952052116394043e-06, + "ewc_loss_parallel": 1.8924474716186523e-06, + "grad_norm": 4.336796760559082, + "learning_rate": 2.840186519711742e-07, + "loss": 0.4714, + "mean_token_accuracy": 0.8512734770774841, + "num_tokens": 25923664.0, + "step": 671 + }, + { + "epoch": 0.0854853072128228, + "ewc_loss": 0.002105712890625, + "ewc_loss_diag": 1.952052116394043e-06, + "ewc_loss_parallel": 1.9073486328125e-06, + "grad_norm": 4.4839558601379395, + "learning_rate": 2.844425604069521e-07, + "loss": 0.4658, + "mean_token_accuracy": 0.8493942022323608, + "num_tokens": 25959370.0, + "step": 672 + }, + { + "epoch": 0.08561251749141331, + "ewc_loss": 0.0021209716796875, + "ewc_loss_diag": 1.952052116394043e-06, + "ewc_loss_parallel": 1.9222497940063477e-06, + "grad_norm": 4.35671329498291, + "learning_rate": 2.8486646884272995e-07, + "loss": 0.4425, + "mean_token_accuracy": 0.8541832566261292, + "num_tokens": 25998398.0, + "step": 673 + }, + { + "epoch": 0.08573972777000381, + "ewc_loss": 0.002105712890625, + "ewc_loss_diag": 1.9669532775878906e-06, + "ewc_loss_parallel": 1.9073486328125e-06, + "grad_norm": 4.517917633056641, + "learning_rate": 2.8529037727850785e-07, + "loss": 0.4047, + "mean_token_accuracy": 0.8633705377578735, + "num_tokens": 26032499.0, + "step": 674 + }, + { + "epoch": 0.08586693804859433, + "ewc_loss": 0.00213623046875, + "ewc_loss_diag": 1.9669532775878906e-06, + "ewc_loss_parallel": 1.9371509552001953e-06, + "grad_norm": 4.322088241577148, + "learning_rate": 2.857142857142857e-07, + "loss": 0.4061, + "mean_token_accuracy": 0.8680689334869385, + "num_tokens": 26072080.0, + "step": 675 + }, + { + "epoch": 0.08599414832718484, + "ewc_loss": 0.0020904541015625, + "ewc_loss_diag": 1.9669532775878906e-06, + "ewc_loss_parallel": 1.8998980522155762e-06, + "grad_norm": 4.222485065460205, + "learning_rate": 2.861381941500636e-07, + "loss": 0.4374, + "mean_token_accuracy": 0.8568308353424072, + "num_tokens": 26118393.0, + "step": 676 + }, + { + "epoch": 0.08612135860577534, + "ewc_loss": 0.002105712890625, + "ewc_loss_diag": 1.9818544387817383e-06, + "ewc_loss_parallel": 1.9073486328125e-06, + "grad_norm": 4.399379730224609, + "learning_rate": 2.8656210258584144e-07, + "loss": 0.4909, + "mean_token_accuracy": 0.8400934934616089, + "num_tokens": 26158414.0, + "step": 677 + }, + { + "epoch": 0.08624856888436586, + "ewc_loss": 0.00213623046875, + "ewc_loss_diag": 1.9818544387817383e-06, + "ewc_loss_parallel": 1.9371509552001953e-06, + "grad_norm": 4.656198978424072, + "learning_rate": 2.869860110216193e-07, + "loss": 0.4952, + "mean_token_accuracy": 0.8366773724555969, + "num_tokens": 26191618.0, + "step": 678 + }, + { + "epoch": 0.08637577916295637, + "ewc_loss": 0.002166748046875, + "ewc_loss_diag": 1.9669532775878906e-06, + "ewc_loss_parallel": 1.9669532775878906e-06, + "grad_norm": 4.148981094360352, + "learning_rate": 2.874099194573972e-07, + "loss": 0.4214, + "mean_token_accuracy": 0.8592063188552856, + "num_tokens": 26236764.0, + "step": 679 + }, + { + "epoch": 0.08650298944154687, + "ewc_loss": 0.0020751953125, + "ewc_loss_diag": 1.9669532775878906e-06, + "ewc_loss_parallel": 1.8775463104248047e-06, + "grad_norm": 4.403696537017822, + "learning_rate": 2.878338278931751e-07, + "loss": 0.4629, + "mean_token_accuracy": 0.8505814075469971, + "num_tokens": 26271609.0, + "step": 680 + }, + { + "epoch": 0.08663019972013739, + "ewc_loss": 0.00213623046875, + "ewc_loss_diag": 1.9818544387817383e-06, + "ewc_loss_parallel": 1.9371509552001953e-06, + "grad_norm": 4.473111629486084, + "learning_rate": 2.8825773632895293e-07, + "loss": 0.4213, + "mean_token_accuracy": 0.8648535013198853, + "num_tokens": 26307811.0, + "step": 681 + }, + { + "epoch": 0.0867574099987279, + "ewc_loss": 0.002166748046875, + "ewc_loss_diag": 1.9669532775878906e-06, + "ewc_loss_parallel": 1.9669532775878906e-06, + "grad_norm": 4.203890800476074, + "learning_rate": 2.886816447647308e-07, + "loss": 0.3953, + "mean_token_accuracy": 0.8675832152366638, + "num_tokens": 26349691.0, + "step": 682 + }, + { + "epoch": 0.08688462027731841, + "ewc_loss": 0.002105712890625, + "ewc_loss_diag": 1.9818544387817383e-06, + "ewc_loss_parallel": 1.9073486328125e-06, + "grad_norm": 4.214193344116211, + "learning_rate": 2.891055532005087e-07, + "loss": 0.4373, + "mean_token_accuracy": 0.8561104536056519, + "num_tokens": 26392268.0, + "step": 683 + }, + { + "epoch": 0.08701183055590891, + "ewc_loss": 0.002105712890625, + "ewc_loss_diag": 1.9669532775878906e-06, + "ewc_loss_parallel": 1.9073486328125e-06, + "grad_norm": 4.56649923324585, + "learning_rate": 2.8952946163628657e-07, + "loss": 0.4595, + "mean_token_accuracy": 0.8466790914535522, + "num_tokens": 26428695.0, + "step": 684 + }, + { + "epoch": 0.08713904083449943, + "ewc_loss": 0.002166748046875, + "ewc_loss_diag": 1.9669532775878906e-06, + "ewc_loss_parallel": 1.9669532775878906e-06, + "grad_norm": 4.414578437805176, + "learning_rate": 2.899533700720644e-07, + "loss": 0.4318, + "mean_token_accuracy": 0.8551557064056396, + "num_tokens": 26467014.0, + "step": 685 + }, + { + "epoch": 0.08726625111308994, + "ewc_loss": 0.0021514892578125, + "ewc_loss_diag": 1.9818544387817383e-06, + "ewc_loss_parallel": 1.952052116394043e-06, + "grad_norm": 4.323354244232178, + "learning_rate": 2.9037727850784227e-07, + "loss": 0.4593, + "mean_token_accuracy": 0.8474814891815186, + "num_tokens": 26504581.0, + "step": 686 + }, + { + "epoch": 0.08739346139168044, + "ewc_loss": 0.00213623046875, + "ewc_loss_diag": 1.9669532775878906e-06, + "ewc_loss_parallel": 1.9371509552001953e-06, + "grad_norm": 4.313834190368652, + "learning_rate": 2.9080118694362016e-07, + "loss": 0.4199, + "mean_token_accuracy": 0.8625014424324036, + "num_tokens": 26545659.0, + "step": 687 + }, + { + "epoch": 0.08752067167027096, + "ewc_loss": 0.00213623046875, + "ewc_loss_diag": 1.9818544387817383e-06, + "ewc_loss_parallel": 1.9371509552001953e-06, + "grad_norm": 4.2269606590271, + "learning_rate": 2.9122509537939806e-07, + "loss": 0.4317, + "mean_token_accuracy": 0.8581088185310364, + "num_tokens": 26589477.0, + "step": 688 + }, + { + "epoch": 0.08764788194886147, + "ewc_loss": 0.00213623046875, + "ewc_loss_diag": 1.9818544387817383e-06, + "ewc_loss_parallel": 1.9371509552001953e-06, + "grad_norm": 4.408390522003174, + "learning_rate": 2.916490038151759e-07, + "loss": 0.4864, + "mean_token_accuracy": 0.8380724191665649, + "num_tokens": 26628880.0, + "step": 689 + }, + { + "epoch": 0.08777509222745197, + "ewc_loss": 0.002166748046875, + "ewc_loss_diag": 2.0116567611694336e-06, + "ewc_loss_parallel": 1.9669532775878906e-06, + "grad_norm": 4.246224880218506, + "learning_rate": 2.9207291225095376e-07, + "loss": 0.4616, + "mean_token_accuracy": 0.8481272459030151, + "num_tokens": 26677159.0, + "step": 690 + }, + { + "epoch": 0.08790230250604249, + "ewc_loss": 0.0021514892578125, + "ewc_loss_diag": 2.0116567611694336e-06, + "ewc_loss_parallel": 1.952052116394043e-06, + "grad_norm": 4.462778568267822, + "learning_rate": 2.9249682068673166e-07, + "loss": 0.4419, + "mean_token_accuracy": 0.8524442315101624, + "num_tokens": 26715667.0, + "step": 691 + }, + { + "epoch": 0.088029512784633, + "ewc_loss": 0.002166748046875, + "ewc_loss_diag": 2.0116567611694336e-06, + "ewc_loss_parallel": 1.9669532775878906e-06, + "grad_norm": 4.341683387756348, + "learning_rate": 2.9292072912250955e-07, + "loss": 0.4176, + "mean_token_accuracy": 0.8655830025672913, + "num_tokens": 26756570.0, + "step": 692 + }, + { + "epoch": 0.0881567230632235, + "ewc_loss": 0.002166748046875, + "ewc_loss_diag": 2.0265579223632812e-06, + "ewc_loss_parallel": 1.9669532775878906e-06, + "grad_norm": 4.415509223937988, + "learning_rate": 2.933446375582874e-07, + "loss": 0.4803, + "mean_token_accuracy": 0.841132402420044, + "num_tokens": 26795856.0, + "step": 693 + }, + { + "epoch": 0.08828393334181402, + "ewc_loss": 0.002166748046875, + "ewc_loss_diag": 2.041459083557129e-06, + "ewc_loss_parallel": 1.9669532775878906e-06, + "grad_norm": 4.473684310913086, + "learning_rate": 2.9376854599406525e-07, + "loss": 0.4323, + "mean_token_accuracy": 0.8559545278549194, + "num_tokens": 26830254.0, + "step": 694 + }, + { + "epoch": 0.08841114362040453, + "ewc_loss": 0.002166748046875, + "ewc_loss_diag": 2.041459083557129e-06, + "ewc_loss_parallel": 1.9669532775878906e-06, + "grad_norm": 4.55148458480835, + "learning_rate": 2.9419245442984315e-07, + "loss": 0.4731, + "mean_token_accuracy": 0.8488110303878784, + "num_tokens": 26864105.0, + "step": 695 + }, + { + "epoch": 0.08853835389899505, + "ewc_loss": 0.0021820068359375, + "ewc_loss_diag": 2.041459083557129e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.514918804168701, + "learning_rate": 2.9461636286562104e-07, + "loss": 0.4379, + "mean_token_accuracy": 0.85334712266922, + "num_tokens": 26897945.0, + "step": 696 + }, + { + "epoch": 0.08866556417758555, + "ewc_loss": 0.0021820068359375, + "ewc_loss_diag": 2.041459083557129e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.179494380950928, + "learning_rate": 2.9504027130139884e-07, + "loss": 0.4234, + "mean_token_accuracy": 0.8603556156158447, + "num_tokens": 26942744.0, + "step": 697 + }, + { + "epoch": 0.08879277445617606, + "ewc_loss": 0.00213623046875, + "ewc_loss_diag": 2.041459083557129e-06, + "ewc_loss_parallel": 1.9371509552001953e-06, + "grad_norm": 4.300869941711426, + "learning_rate": 2.9546417973717674e-07, + "loss": 0.4538, + "mean_token_accuracy": 0.8502785563468933, + "num_tokens": 26983866.0, + "step": 698 + }, + { + "epoch": 0.08891998473476657, + "ewc_loss": 0.002166748046875, + "ewc_loss_diag": 2.041459083557129e-06, + "ewc_loss_parallel": 1.9669532775878906e-06, + "grad_norm": 4.3599677085876465, + "learning_rate": 2.9588808817295464e-07, + "loss": 0.395, + "mean_token_accuracy": 0.8691436052322388, + "num_tokens": 27022873.0, + "step": 699 + }, + { + "epoch": 0.08904719501335707, + "ewc_loss": 0.002197265625, + "ewc_loss_diag": 2.0563602447509766e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.4990949630737305, + "learning_rate": 2.9631199660873253e-07, + "loss": 0.4684, + "mean_token_accuracy": 0.8483444452285767, + "num_tokens": 27059178.0, + "step": 700 + }, + { + "epoch": 0.08917440529194759, + "ewc_loss": 0.002197265625, + "ewc_loss_diag": 2.0563602447509766e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.394871711730957, + "learning_rate": 2.9673590504451033e-07, + "loss": 0.3937, + "mean_token_accuracy": 0.8706023693084717, + "num_tokens": 27095660.0, + "step": 701 + }, + { + "epoch": 0.0893016155705381, + "ewc_loss": 0.002197265625, + "ewc_loss_diag": 2.0563602447509766e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.320557117462158, + "learning_rate": 2.9715981348028823e-07, + "loss": 0.4149, + "mean_token_accuracy": 0.8630788326263428, + "num_tokens": 27135513.0, + "step": 702 + }, + { + "epoch": 0.0894288258491286, + "ewc_loss": 0.002197265625, + "ewc_loss_diag": 2.0712614059448242e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.468873977661133, + "learning_rate": 2.975837219160661e-07, + "loss": 0.4724, + "mean_token_accuracy": 0.8436055183410645, + "num_tokens": 27175512.0, + "step": 703 + }, + { + "epoch": 0.08955603612771912, + "ewc_loss": 0.002197265625, + "ewc_loss_diag": 2.0712614059448242e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.384293079376221, + "learning_rate": 2.98007630351844e-07, + "loss": 0.453, + "mean_token_accuracy": 0.8494729399681091, + "num_tokens": 27217830.0, + "step": 704 + }, + { + "epoch": 0.08968324640630963, + "ewc_loss": 0.002197265625, + "ewc_loss_diag": 2.0712614059448242e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.458218574523926, + "learning_rate": 2.984315387876218e-07, + "loss": 0.4504, + "mean_token_accuracy": 0.8508626222610474, + "num_tokens": 27253675.0, + "step": 705 + }, + { + "epoch": 0.08981045668490013, + "ewc_loss": 0.002197265625, + "ewc_loss_diag": 2.0712614059448242e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.572619438171387, + "learning_rate": 2.988554472233997e-07, + "loss": 0.4647, + "mean_token_accuracy": 0.8480578660964966, + "num_tokens": 27294031.0, + "step": 706 + }, + { + "epoch": 0.08993766696349065, + "ewc_loss": 0.0022125244140625, + "ewc_loss_diag": 2.0712614059448242e-06, + "ewc_loss_parallel": 1.996755599975586e-06, + "grad_norm": 4.391758441925049, + "learning_rate": 2.992793556591776e-07, + "loss": 0.4486, + "mean_token_accuracy": 0.851860761642456, + "num_tokens": 27334836.0, + "step": 707 + }, + { + "epoch": 0.09006487724208116, + "ewc_loss": 0.002197265625, + "ewc_loss_diag": 2.0712614059448242e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.407960891723633, + "learning_rate": 2.997032640949555e-07, + "loss": 0.4548, + "mean_token_accuracy": 0.8495573997497559, + "num_tokens": 27375836.0, + "step": 708 + }, + { + "epoch": 0.09019208752067168, + "ewc_loss": 0.002197265625, + "ewc_loss_diag": 2.0712614059448242e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.301726341247559, + "learning_rate": 3.001271725307333e-07, + "loss": 0.3886, + "mean_token_accuracy": 0.8696674108505249, + "num_tokens": 27413113.0, + "step": 709 + }, + { + "epoch": 0.09031929779926218, + "ewc_loss": 0.002197265625, + "ewc_loss_diag": 2.0712614059448242e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.540442943572998, + "learning_rate": 3.005510809665112e-07, + "loss": 0.4482, + "mean_token_accuracy": 0.8541300296783447, + "num_tokens": 27447566.0, + "step": 710 + }, + { + "epoch": 0.09044650807785269, + "ewc_loss": 0.0022125244140625, + "ewc_loss_diag": 2.086162567138672e-06, + "ewc_loss_parallel": 1.996755599975586e-06, + "grad_norm": 4.38201379776001, + "learning_rate": 3.009749894022891e-07, + "loss": 0.397, + "mean_token_accuracy": 0.8688907623291016, + "num_tokens": 27487044.0, + "step": 711 + }, + { + "epoch": 0.0905737183564432, + "ewc_loss": 0.002197265625, + "ewc_loss_diag": 2.086162567138672e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.42263650894165, + "learning_rate": 3.01398897838067e-07, + "loss": 0.4245, + "mean_token_accuracy": 0.8621422648429871, + "num_tokens": 27529316.0, + "step": 712 + }, + { + "epoch": 0.0907009286350337, + "ewc_loss": 0.002197265625, + "ewc_loss_diag": 2.086162567138672e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.555224895477295, + "learning_rate": 3.018228062738448e-07, + "loss": 0.4673, + "mean_token_accuracy": 0.8478794693946838, + "num_tokens": 27566245.0, + "step": 713 + }, + { + "epoch": 0.09082813891362422, + "ewc_loss": 0.0022125244140625, + "ewc_loss_diag": 2.086162567138672e-06, + "ewc_loss_parallel": 1.996755599975586e-06, + "grad_norm": 4.395656108856201, + "learning_rate": 3.022467147096227e-07, + "loss": 0.4139, + "mean_token_accuracy": 0.8627766966819763, + "num_tokens": 27603820.0, + "step": 714 + }, + { + "epoch": 0.09095534919221474, + "ewc_loss": 0.0022125244140625, + "ewc_loss_diag": 2.086162567138672e-06, + "ewc_loss_parallel": 1.996755599975586e-06, + "grad_norm": 4.234278678894043, + "learning_rate": 3.026706231454006e-07, + "loss": 0.4272, + "mean_token_accuracy": 0.8564444780349731, + "num_tokens": 27649616.0, + "step": 715 + }, + { + "epoch": 0.09108255947080524, + "ewc_loss": 0.002197265625, + "ewc_loss_diag": 2.086162567138672e-06, + "ewc_loss_parallel": 1.9818544387817383e-06, + "grad_norm": 4.416505336761475, + "learning_rate": 3.0309453158117844e-07, + "loss": 0.4233, + "mean_token_accuracy": 0.8589411973953247, + "num_tokens": 27687116.0, + "step": 716 + }, + { + "epoch": 0.09120976974939575, + "ewc_loss": 0.0022125244140625, + "ewc_loss_diag": 2.086162567138672e-06, + "ewc_loss_parallel": 1.996755599975586e-06, + "grad_norm": 4.626882076263428, + "learning_rate": 3.035184400169563e-07, + "loss": 0.4955, + "mean_token_accuracy": 0.8387105464935303, + "num_tokens": 27725687.0, + "step": 717 + }, + { + "epoch": 0.09133698002798626, + "ewc_loss": 0.0022430419921875, + "ewc_loss_diag": 2.1010637283325195e-06, + "ewc_loss_parallel": 2.0265579223632812e-06, + "grad_norm": 4.491110801696777, + "learning_rate": 3.039423484527342e-07, + "loss": 0.4228, + "mean_token_accuracy": 0.8588642477989197, + "num_tokens": 27764801.0, + "step": 718 + }, + { + "epoch": 0.09146419030657676, + "ewc_loss": 0.0022125244140625, + "ewc_loss_diag": 2.1010637283325195e-06, + "ewc_loss_parallel": 1.996755599975586e-06, + "grad_norm": 4.520818710327148, + "learning_rate": 3.043662568885121e-07, + "loss": 0.4408, + "mean_token_accuracy": 0.8515306711196899, + "num_tokens": 27800977.0, + "step": 719 + }, + { + "epoch": 0.09159140058516728, + "ewc_loss": 0.0022125244140625, + "ewc_loss_diag": 2.1010637283325195e-06, + "ewc_loss_parallel": 1.996755599975586e-06, + "grad_norm": 4.5953497886657715, + "learning_rate": 3.0479016532428993e-07, + "loss": 0.4769, + "mean_token_accuracy": 0.8432960510253906, + "num_tokens": 27836320.0, + "step": 720 + }, + { + "epoch": 0.0917186108637578, + "ewc_loss": 0.0022735595703125, + "ewc_loss_diag": 2.1010637283325195e-06, + "ewc_loss_parallel": 2.0563602447509766e-06, + "grad_norm": 4.4675397872924805, + "learning_rate": 3.052140737600678e-07, + "loss": 0.4135, + "mean_token_accuracy": 0.8633217215538025, + "num_tokens": 27871231.0, + "step": 721 + }, + { + "epoch": 0.09184582114234831, + "ewc_loss": 0.002227783203125, + "ewc_loss_diag": 2.1010637283325195e-06, + "ewc_loss_parallel": 2.0116567611694336e-06, + "grad_norm": 4.414269924163818, + "learning_rate": 3.056379821958457e-07, + "loss": 0.4316, + "mean_token_accuracy": 0.8582567572593689, + "num_tokens": 27908522.0, + "step": 722 + }, + { + "epoch": 0.09197303142093881, + "ewc_loss": 0.0022125244140625, + "ewc_loss_diag": 2.115964889526367e-06, + "ewc_loss_parallel": 1.996755599975586e-06, + "grad_norm": 4.522061347961426, + "learning_rate": 3.060618906316236e-07, + "loss": 0.4261, + "mean_token_accuracy": 0.8602964878082275, + "num_tokens": 27944651.0, + "step": 723 + }, + { + "epoch": 0.09210024169952932, + "ewc_loss": 0.0022125244140625, + "ewc_loss_diag": 2.115964889526367e-06, + "ewc_loss_parallel": 1.996755599975586e-06, + "grad_norm": 4.460301399230957, + "learning_rate": 3.064857990674014e-07, + "loss": 0.4137, + "mean_token_accuracy": 0.862521231174469, + "num_tokens": 27981491.0, + "step": 724 + }, + { + "epoch": 0.09222745197811984, + "ewc_loss": 0.002227783203125, + "ewc_loss_diag": 2.130866050720215e-06, + "ewc_loss_parallel": 2.0116567611694336e-06, + "grad_norm": 4.409663200378418, + "learning_rate": 3.0690970750317927e-07, + "loss": 0.4402, + "mean_token_accuracy": 0.8545078039169312, + "num_tokens": 28022448.0, + "step": 725 + }, + { + "epoch": 0.09235466225671034, + "ewc_loss": 0.0022125244140625, + "ewc_loss_diag": 2.130866050720215e-06, + "ewc_loss_parallel": 1.996755599975586e-06, + "grad_norm": 4.5572357177734375, + "learning_rate": 3.0733361593895717e-07, + "loss": 0.4174, + "mean_token_accuracy": 0.8627495169639587, + "num_tokens": 28058241.0, + "step": 726 + }, + { + "epoch": 0.09248187253530085, + "ewc_loss": 0.0022735595703125, + "ewc_loss_diag": 2.130866050720215e-06, + "ewc_loss_parallel": 2.0563602447509766e-06, + "grad_norm": 4.35283088684082, + "learning_rate": 3.0775752437473507e-07, + "loss": 0.4419, + "mean_token_accuracy": 0.8517745733261108, + "num_tokens": 28099959.0, + "step": 727 + }, + { + "epoch": 0.09260908281389137, + "ewc_loss": 0.0022125244140625, + "ewc_loss_diag": 2.130866050720215e-06, + "ewc_loss_parallel": 1.996755599975586e-06, + "grad_norm": 4.262871265411377, + "learning_rate": 3.081814328105129e-07, + "loss": 0.4541, + "mean_token_accuracy": 0.8492827415466309, + "num_tokens": 28147277.0, + "step": 728 + }, + { + "epoch": 0.09273629309248187, + "ewc_loss": 0.0022125244140625, + "ewc_loss_diag": 2.130866050720215e-06, + "ewc_loss_parallel": 1.996755599975586e-06, + "grad_norm": 4.3961286544799805, + "learning_rate": 3.0860534124629076e-07, + "loss": 0.4792, + "mean_token_accuracy": 0.8429089784622192, + "num_tokens": 28191619.0, + "step": 729 + }, + { + "epoch": 0.09286350337107238, + "ewc_loss": 0.0022735595703125, + "ewc_loss_diag": 2.130866050720215e-06, + "ewc_loss_parallel": 2.0563602447509766e-06, + "grad_norm": 4.535382270812988, + "learning_rate": 3.0902924968206866e-07, + "loss": 0.4355, + "mean_token_accuracy": 0.8585734963417053, + "num_tokens": 28227916.0, + "step": 730 + }, + { + "epoch": 0.0929907136496629, + "ewc_loss": 0.0023040771484375, + "ewc_loss_diag": 2.1457672119140625e-06, + "ewc_loss_parallel": 2.086162567138672e-06, + "grad_norm": 4.725759029388428, + "learning_rate": 3.0945315811784656e-07, + "loss": 0.4138, + "mean_token_accuracy": 0.8596022725105286, + "num_tokens": 28262852.0, + "step": 731 + }, + { + "epoch": 0.0931179239282534, + "ewc_loss": 0.002349853515625, + "ewc_loss_diag": 2.1457672119140625e-06, + "ewc_loss_parallel": 2.130866050720215e-06, + "grad_norm": 4.492702484130859, + "learning_rate": 3.098770665536244e-07, + "loss": 0.5568, + "mean_token_accuracy": 0.8189314603805542, + "num_tokens": 28306681.0, + "step": 732 + }, + { + "epoch": 0.09324513420684391, + "ewc_loss": 0.0023193359375, + "ewc_loss_diag": 2.175569534301758e-06, + "ewc_loss_parallel": 2.1010637283325195e-06, + "grad_norm": 4.45076322555542, + "learning_rate": 3.1030097498940225e-07, + "loss": 0.4099, + "mean_token_accuracy": 0.8636717796325684, + "num_tokens": 28347125.0, + "step": 733 + }, + { + "epoch": 0.09337234448543442, + "ewc_loss": 0.0022735595703125, + "ewc_loss_diag": 2.175569534301758e-06, + "ewc_loss_parallel": 2.0563602447509766e-06, + "grad_norm": 4.530194282531738, + "learning_rate": 3.1072488342518015e-07, + "loss": 0.4333, + "mean_token_accuracy": 0.8516634106636047, + "num_tokens": 28383846.0, + "step": 734 + }, + { + "epoch": 0.09349955476402494, + "ewc_loss": 0.002288818359375, + "ewc_loss_diag": 2.16066837310791e-06, + "ewc_loss_parallel": 2.0712614059448242e-06, + "grad_norm": 4.440769672393799, + "learning_rate": 3.11148791860958e-07, + "loss": 0.4314, + "mean_token_accuracy": 0.8559668660163879, + "num_tokens": 28421166.0, + "step": 735 + }, + { + "epoch": 0.09362676504261544, + "ewc_loss": 0.002288818359375, + "ewc_loss_diag": 2.16066837310791e-06, + "ewc_loss_parallel": 2.0712614059448242e-06, + "grad_norm": 4.747913837432861, + "learning_rate": 3.115727002967359e-07, + "loss": 0.4596, + "mean_token_accuracy": 0.8451461791992188, + "num_tokens": 28451092.0, + "step": 736 + }, + { + "epoch": 0.09375397532120595, + "ewc_loss": 0.0023651123046875, + "ewc_loss_diag": 2.16066837310791e-06, + "ewc_loss_parallel": 2.1457672119140625e-06, + "grad_norm": 4.340391159057617, + "learning_rate": 3.1199660873251374e-07, + "loss": 0.4337, + "mean_token_accuracy": 0.857040524482727, + "num_tokens": 28491797.0, + "step": 737 + }, + { + "epoch": 0.09388118559979647, + "ewc_loss": 0.00225830078125, + "ewc_loss_diag": 2.16066837310791e-06, + "ewc_loss_parallel": 2.041459083557129e-06, + "grad_norm": 4.307288646697998, + "learning_rate": 3.1242051716829164e-07, + "loss": 0.4641, + "mean_token_accuracy": 0.8452820777893066, + "num_tokens": 28538703.0, + "step": 738 + }, + { + "epoch": 0.09400839587838697, + "ewc_loss": 0.0023193359375, + "ewc_loss_diag": 2.175569534301758e-06, + "ewc_loss_parallel": 2.1010637283325195e-06, + "grad_norm": 4.498228549957275, + "learning_rate": 3.128444256040695e-07, + "loss": 0.4589, + "mean_token_accuracy": 0.8489775061607361, + "num_tokens": 28579393.0, + "step": 739 + }, + { + "epoch": 0.09413560615697748, + "ewc_loss": 0.0023651123046875, + "ewc_loss_diag": 2.175569534301758e-06, + "ewc_loss_parallel": 2.1457672119140625e-06, + "grad_norm": 4.371544361114502, + "learning_rate": 3.132683340398474e-07, + "loss": 0.3948, + "mean_token_accuracy": 0.8671221733093262, + "num_tokens": 28618706.0, + "step": 740 + }, + { + "epoch": 0.094262816435568, + "ewc_loss": 0.0023345947265625, + "ewc_loss_diag": 2.175569534301758e-06, + "ewc_loss_parallel": 2.115964889526367e-06, + "grad_norm": 4.395749568939209, + "learning_rate": 3.1369224247562523e-07, + "loss": 0.3952, + "mean_token_accuracy": 0.869446337223053, + "num_tokens": 28655123.0, + "step": 741 + }, + { + "epoch": 0.0943900267141585, + "ewc_loss": 0.0023193359375, + "ewc_loss_diag": 2.16066837310791e-06, + "ewc_loss_parallel": 2.1010637283325195e-06, + "grad_norm": 4.181960105895996, + "learning_rate": 3.1411615091140313e-07, + "loss": 0.3777, + "mean_token_accuracy": 0.8755345940589905, + "num_tokens": 28699965.0, + "step": 742 + }, + { + "epoch": 0.09451723699274901, + "ewc_loss": 0.002288818359375, + "ewc_loss_diag": 2.16066837310791e-06, + "ewc_loss_parallel": 2.0712614059448242e-06, + "grad_norm": 4.394887924194336, + "learning_rate": 3.14540059347181e-07, + "loss": 0.4104, + "mean_token_accuracy": 0.8626682758331299, + "num_tokens": 28738167.0, + "step": 743 + }, + { + "epoch": 0.09464444727133953, + "ewc_loss": 0.002349853515625, + "ewc_loss_diag": 2.16066837310791e-06, + "ewc_loss_parallel": 2.130866050720215e-06, + "grad_norm": 4.449260234832764, + "learning_rate": 3.149639677829589e-07, + "loss": 0.3842, + "mean_token_accuracy": 0.8730118274688721, + "num_tokens": 28776693.0, + "step": 744 + }, + { + "epoch": 0.09477165754993004, + "ewc_loss": 0.00238037109375, + "ewc_loss_diag": 2.16066837310791e-06, + "ewc_loss_parallel": 2.16066837310791e-06, + "grad_norm": 4.590622425079346, + "learning_rate": 3.153878762187368e-07, + "loss": 0.4802, + "mean_token_accuracy": 0.845291793346405, + "num_tokens": 28814718.0, + "step": 745 + }, + { + "epoch": 0.09489886782852054, + "ewc_loss": 0.0023956298828125, + "ewc_loss_diag": 2.1904706954956055e-06, + "ewc_loss_parallel": 2.175569534301758e-06, + "grad_norm": 4.551178932189941, + "learning_rate": 3.158117846545146e-07, + "loss": 0.4934, + "mean_token_accuracy": 0.8397955894470215, + "num_tokens": 28858240.0, + "step": 746 + }, + { + "epoch": 0.09502607810711106, + "ewc_loss": 0.0023956298828125, + "ewc_loss_diag": 2.175569534301758e-06, + "ewc_loss_parallel": 2.175569534301758e-06, + "grad_norm": 4.6920881271362305, + "learning_rate": 3.1623569309029247e-07, + "loss": 0.4322, + "mean_token_accuracy": 0.8563714027404785, + "num_tokens": 28889079.0, + "step": 747 + }, + { + "epoch": 0.09515328838570157, + "ewc_loss": 0.0023956298828125, + "ewc_loss_diag": 2.1904706954956055e-06, + "ewc_loss_parallel": 2.175569534301758e-06, + "grad_norm": 4.359278678894043, + "learning_rate": 3.1665960152607037e-07, + "loss": 0.3939, + "mean_token_accuracy": 0.8678805232048035, + "num_tokens": 28930824.0, + "step": 748 + }, + { + "epoch": 0.09528049866429207, + "ewc_loss": 0.0023345947265625, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.115964889526367e-06, + "grad_norm": 4.44692850112915, + "learning_rate": 3.1708350996184826e-07, + "loss": 0.4327, + "mean_token_accuracy": 0.8555715084075928, + "num_tokens": 28971197.0, + "step": 749 + }, + { + "epoch": 0.09540770894288259, + "ewc_loss": 0.00238037109375, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.16066837310791e-06, + "grad_norm": 4.4165754318237305, + "learning_rate": 3.175074183976261e-07, + "loss": 0.4055, + "mean_token_accuracy": 0.8651984930038452, + "num_tokens": 29010249.0, + "step": 750 + }, + { + "epoch": 0.0955349192214731, + "ewc_loss": 0.00238037109375, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.16066837310791e-06, + "grad_norm": 4.568894863128662, + "learning_rate": 3.1793132683340396e-07, + "loss": 0.4262, + "mean_token_accuracy": 0.8605321645736694, + "num_tokens": 29046266.0, + "step": 751 + }, + { + "epoch": 0.0956621295000636, + "ewc_loss": 0.0023956298828125, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.175569534301758e-06, + "grad_norm": 4.551288604736328, + "learning_rate": 3.1835523526918186e-07, + "loss": 0.3865, + "mean_token_accuracy": 0.8709444403648376, + "num_tokens": 29083625.0, + "step": 752 + }, + { + "epoch": 0.09578933977865411, + "ewc_loss": 0.0023956298828125, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.175569534301758e-06, + "grad_norm": 4.515514850616455, + "learning_rate": 3.1877914370495975e-07, + "loss": 0.4538, + "mean_token_accuracy": 0.8502762317657471, + "num_tokens": 29124650.0, + "step": 753 + }, + { + "epoch": 0.09591655005724463, + "ewc_loss": 0.0023956298828125, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.175569534301758e-06, + "grad_norm": 4.506228923797607, + "learning_rate": 3.1920305214073755e-07, + "loss": 0.4785, + "mean_token_accuracy": 0.8415970802307129, + "num_tokens": 29165719.0, + "step": 754 + }, + { + "epoch": 0.09604376033583513, + "ewc_loss": 0.0023956298828125, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.175569534301758e-06, + "grad_norm": 4.5603156089782715, + "learning_rate": 3.1962696057651545e-07, + "loss": 0.4155, + "mean_token_accuracy": 0.8606787919998169, + "num_tokens": 29206375.0, + "step": 755 + }, + { + "epoch": 0.09617097061442564, + "ewc_loss": 0.002410888671875, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.1904706954956055e-06, + "grad_norm": 4.674526214599609, + "learning_rate": 3.2005086901229335e-07, + "loss": 0.4572, + "mean_token_accuracy": 0.847220242023468, + "num_tokens": 29241409.0, + "step": 756 + }, + { + "epoch": 0.09629818089301616, + "ewc_loss": 0.002410888671875, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.1904706954956055e-06, + "grad_norm": 4.487938404083252, + "learning_rate": 3.2047477744807125e-07, + "loss": 0.451, + "mean_token_accuracy": 0.8517212867736816, + "num_tokens": 29280308.0, + "step": 757 + }, + { + "epoch": 0.09642539117160667, + "ewc_loss": 0.0023956298828125, + "ewc_loss_diag": 2.1904706954956055e-06, + "ewc_loss_parallel": 2.175569534301758e-06, + "grad_norm": 4.553189277648926, + "learning_rate": 3.2089868588384904e-07, + "loss": 0.4721, + "mean_token_accuracy": 0.8420071601867676, + "num_tokens": 29320404.0, + "step": 758 + }, + { + "epoch": 0.09655260145019717, + "ewc_loss": 0.0023956298828125, + "ewc_loss_diag": 2.1904706954956055e-06, + "ewc_loss_parallel": 2.175569534301758e-06, + "grad_norm": 4.48773193359375, + "learning_rate": 3.2132259431962694e-07, + "loss": 0.4031, + "mean_token_accuracy": 0.8689550161361694, + "num_tokens": 29359294.0, + "step": 759 + }, + { + "epoch": 0.09667981172878769, + "ewc_loss": 0.0023956298828125, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.175569534301758e-06, + "grad_norm": 4.45759391784668, + "learning_rate": 3.2174650275540484e-07, + "loss": 0.4456, + "mean_token_accuracy": 0.857010543346405, + "num_tokens": 29398994.0, + "step": 760 + }, + { + "epoch": 0.0968070220073782, + "ewc_loss": 0.0023956298828125, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.175569534301758e-06, + "grad_norm": 4.369357109069824, + "learning_rate": 3.2217041119118274e-07, + "loss": 0.409, + "mean_token_accuracy": 0.8641924858093262, + "num_tokens": 29441092.0, + "step": 761 + }, + { + "epoch": 0.0969342322859687, + "ewc_loss": 0.0023956298828125, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.175569534301758e-06, + "grad_norm": 4.5957465171813965, + "learning_rate": 3.2259431962696053e-07, + "loss": 0.44, + "mean_token_accuracy": 0.8549743294715881, + "num_tokens": 29474964.0, + "step": 762 + }, + { + "epoch": 0.09706144256455922, + "ewc_loss": 0.002410888671875, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.1904706954956055e-06, + "grad_norm": 4.617809772491455, + "learning_rate": 3.2301822806273843e-07, + "loss": 0.4282, + "mean_token_accuracy": 0.8542261123657227, + "num_tokens": 29510991.0, + "step": 763 + }, + { + "epoch": 0.09718865284314973, + "ewc_loss": 0.002410888671875, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.1904706954956055e-06, + "grad_norm": 4.5429205894470215, + "learning_rate": 3.2344213649851633e-07, + "loss": 0.4559, + "mean_token_accuracy": 0.8487820625305176, + "num_tokens": 29555596.0, + "step": 764 + }, + { + "epoch": 0.09731586312174023, + "ewc_loss": 0.002410888671875, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.1904706954956055e-06, + "grad_norm": 4.4555983543396, + "learning_rate": 3.238660449342942e-07, + "loss": 0.4199, + "mean_token_accuracy": 0.8588057160377502, + "num_tokens": 29597536.0, + "step": 765 + }, + { + "epoch": 0.09744307340033075, + "ewc_loss": 0.002410888671875, + "ewc_loss_diag": 2.1904706954956055e-06, + "ewc_loss_parallel": 2.1904706954956055e-06, + "grad_norm": 4.694119453430176, + "learning_rate": 3.24289953370072e-07, + "loss": 0.4, + "mean_token_accuracy": 0.860832929611206, + "num_tokens": 29630840.0, + "step": 766 + }, + { + "epoch": 0.09757028367892126, + "ewc_loss": 0.0024261474609375, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.205371856689453e-06, + "grad_norm": 4.382583141326904, + "learning_rate": 3.247138618058499e-07, + "loss": 0.4437, + "mean_token_accuracy": 0.8489798307418823, + "num_tokens": 29676140.0, + "step": 767 + }, + { + "epoch": 0.09769749395751176, + "ewc_loss": 0.0023956298828125, + "ewc_loss_diag": 2.1904706954956055e-06, + "ewc_loss_parallel": 2.175569534301758e-06, + "grad_norm": 4.467190742492676, + "learning_rate": 3.251377702416278e-07, + "loss": 0.472, + "mean_token_accuracy": 0.8454699516296387, + "num_tokens": 29718950.0, + "step": 768 + }, + { + "epoch": 0.09782470423610228, + "ewc_loss": 0.0024261474609375, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.205371856689453e-06, + "grad_norm": 4.357626438140869, + "learning_rate": 3.255616786774057e-07, + "loss": 0.4083, + "mean_token_accuracy": 0.8640967607498169, + "num_tokens": 29764145.0, + "step": 769 + }, + { + "epoch": 0.09795191451469279, + "ewc_loss": 0.002410888671875, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.1904706954956055e-06, + "grad_norm": 4.277886390686035, + "learning_rate": 3.259855871131835e-07, + "loss": 0.3665, + "mean_token_accuracy": 0.877937912940979, + "num_tokens": 29807155.0, + "step": 770 + }, + { + "epoch": 0.0980791247932833, + "ewc_loss": 0.002410888671875, + "ewc_loss_diag": 2.1904706954956055e-06, + "ewc_loss_parallel": 2.1904706954956055e-06, + "grad_norm": 4.499448776245117, + "learning_rate": 3.264094955489614e-07, + "loss": 0.4181, + "mean_token_accuracy": 0.8625502586364746, + "num_tokens": 29848463.0, + "step": 771 + }, + { + "epoch": 0.0982063350718738, + "ewc_loss": 0.0024261474609375, + "ewc_loss_diag": 2.1904706954956055e-06, + "ewc_loss_parallel": 2.205371856689453e-06, + "grad_norm": 4.538425445556641, + "learning_rate": 3.268334039847393e-07, + "loss": 0.4392, + "mean_token_accuracy": 0.8556604385375977, + "num_tokens": 29887956.0, + "step": 772 + }, + { + "epoch": 0.09833354535046432, + "ewc_loss": 0.0024261474609375, + "ewc_loss_diag": 2.1904706954956055e-06, + "ewc_loss_parallel": 2.205371856689453e-06, + "grad_norm": 4.479448318481445, + "learning_rate": 3.2725731242051715e-07, + "loss": 0.4274, + "mean_token_accuracy": 0.860374927520752, + "num_tokens": 29930395.0, + "step": 773 + }, + { + "epoch": 0.09846075562905483, + "ewc_loss": 0.0024261474609375, + "ewc_loss_diag": 2.1904706954956055e-06, + "ewc_loss_parallel": 2.205371856689453e-06, + "grad_norm": 4.5086822509765625, + "learning_rate": 3.27681220856295e-07, + "loss": 0.4184, + "mean_token_accuracy": 0.8601645231246948, + "num_tokens": 29972911.0, + "step": 774 + }, + { + "epoch": 0.09858796590764533, + "ewc_loss": 0.002410888671875, + "ewc_loss_diag": 2.1904706954956055e-06, + "ewc_loss_parallel": 2.1904706954956055e-06, + "grad_norm": 4.638214111328125, + "learning_rate": 3.281051292920729e-07, + "loss": 0.4127, + "mean_token_accuracy": 0.8594253063201904, + "num_tokens": 30007473.0, + "step": 775 + }, + { + "epoch": 0.09871517618623585, + "ewc_loss": 0.00244140625, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.2202730178833008e-06, + "grad_norm": 4.532679080963135, + "learning_rate": 3.285290377278508e-07, + "loss": 0.4599, + "mean_token_accuracy": 0.8481907844543457, + "num_tokens": 30047325.0, + "step": 776 + }, + { + "epoch": 0.09884238646482636, + "ewc_loss": 0.002410888671875, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.1904706954956055e-06, + "grad_norm": 4.525417327880859, + "learning_rate": 3.2895294616362864e-07, + "loss": 0.409, + "mean_token_accuracy": 0.8655389547348022, + "num_tokens": 30084104.0, + "step": 777 + }, + { + "epoch": 0.09896959674341686, + "ewc_loss": 0.00244140625, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.205371856689453e-06, + "grad_norm": 4.56219482421875, + "learning_rate": 3.293768545994065e-07, + "loss": 0.4248, + "mean_token_accuracy": 0.857632040977478, + "num_tokens": 30122134.0, + "step": 778 + }, + { + "epoch": 0.09909680702200738, + "ewc_loss": 0.00244140625, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.205371856689453e-06, + "grad_norm": 4.60991907119751, + "learning_rate": 3.298007630351844e-07, + "loss": 0.4786, + "mean_token_accuracy": 0.8467869162559509, + "num_tokens": 30160337.0, + "step": 779 + }, + { + "epoch": 0.09922401730059789, + "ewc_loss": 0.0024261474609375, + "ewc_loss_diag": 2.205371856689453e-06, + "ewc_loss_parallel": 2.205371856689453e-06, + "grad_norm": 4.386682510375977, + "learning_rate": 3.302246714709623e-07, + "loss": 0.406, + "mean_token_accuracy": 0.8623753786087036, + "num_tokens": 30204192.0, + "step": 780 + }, + { + "epoch": 0.09935122757918839, + "ewc_loss": 0.0024261474609375, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.1904706954956055e-06, + "grad_norm": 4.535546779632568, + "learning_rate": 3.3064857990674013e-07, + "loss": 0.4332, + "mean_token_accuracy": 0.8566017746925354, + "num_tokens": 30242892.0, + "step": 781 + }, + { + "epoch": 0.0994784378577789, + "ewc_loss": 0.00244140625, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.205371856689453e-06, + "grad_norm": 4.7248148918151855, + "learning_rate": 3.31072488342518e-07, + "loss": 0.4022, + "mean_token_accuracy": 0.8641558885574341, + "num_tokens": 30275359.0, + "step": 782 + }, + { + "epoch": 0.09960564813636942, + "ewc_loss": 0.0024566650390625, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.2351741790771484e-06, + "grad_norm": 4.410841941833496, + "learning_rate": 3.314963967782959e-07, + "loss": 0.399, + "mean_token_accuracy": 0.8669601678848267, + "num_tokens": 30319464.0, + "step": 783 + }, + { + "epoch": 0.09973285841495994, + "ewc_loss": 0.00244140625, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.205371856689453e-06, + "grad_norm": 4.453790664672852, + "learning_rate": 3.319203052140738e-07, + "loss": 0.4349, + "mean_token_accuracy": 0.8583430051803589, + "num_tokens": 30361415.0, + "step": 784 + }, + { + "epoch": 0.09986006869355044, + "ewc_loss": 0.0024566650390625, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.2202730178833008e-06, + "grad_norm": 4.903705596923828, + "learning_rate": 3.323442136498516e-07, + "loss": 0.4136, + "mean_token_accuracy": 0.861691415309906, + "num_tokens": 30392330.0, + "step": 785 + }, + { + "epoch": 0.09998727897214095, + "ewc_loss": 0.0024871826171875, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.2649765014648438e-06, + "grad_norm": 4.753676414489746, + "learning_rate": 3.3276812208562947e-07, + "loss": 0.4168, + "mean_token_accuracy": 0.8592026829719543, + "num_tokens": 30428010.0, + "step": 786 + }, + { + "epoch": 0.10011448925073146, + "ewc_loss": 0.0024566650390625, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.2351741790771484e-06, + "grad_norm": 4.663175106048584, + "learning_rate": 3.3319203052140737e-07, + "loss": 0.4713, + "mean_token_accuracy": 0.8516366481781006, + "num_tokens": 30465989.0, + "step": 787 + }, + { + "epoch": 0.10024169952932196, + "ewc_loss": 0.0024566650390625, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.2351741790771484e-06, + "grad_norm": 4.567439556121826, + "learning_rate": 3.336159389571852e-07, + "loss": 0.4157, + "mean_token_accuracy": 0.8654674291610718, + "num_tokens": 30502113.0, + "step": 788 + }, + { + "epoch": 0.10036890980791248, + "ewc_loss": 0.0024566650390625, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.2351741790771484e-06, + "grad_norm": 4.591548919677734, + "learning_rate": 3.340398473929631e-07, + "loss": 0.4778, + "mean_token_accuracy": 0.8418172001838684, + "num_tokens": 30545413.0, + "step": 789 + }, + { + "epoch": 0.100496120086503, + "ewc_loss": 0.0024871826171875, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.2649765014648438e-06, + "grad_norm": 4.654505729675293, + "learning_rate": 3.3446375582874096e-07, + "loss": 0.4252, + "mean_token_accuracy": 0.8600649237632751, + "num_tokens": 30583845.0, + "step": 790 + }, + { + "epoch": 0.1006233303650935, + "ewc_loss": 0.002471923828125, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.250075340270996e-06, + "grad_norm": 4.681400775909424, + "learning_rate": 3.3488766426451886e-07, + "loss": 0.473, + "mean_token_accuracy": 0.8464956283569336, + "num_tokens": 30622219.0, + "step": 791 + }, + { + "epoch": 0.10075054064368401, + "ewc_loss": 0.0024871826171875, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.2649765014648438e-06, + "grad_norm": 4.539174556732178, + "learning_rate": 3.353115727002967e-07, + "loss": 0.4603, + "mean_token_accuracy": 0.8460835218429565, + "num_tokens": 30663659.0, + "step": 792 + }, + { + "epoch": 0.10087775092227452, + "ewc_loss": 0.0024566650390625, + "ewc_loss_diag": 2.2202730178833008e-06, + "ewc_loss_parallel": 2.2351741790771484e-06, + "grad_norm": 4.470034122467041, + "learning_rate": 3.357354811360746e-07, + "loss": 0.4496, + "mean_token_accuracy": 0.849274754524231, + "num_tokens": 30706611.0, + "step": 793 + }, + { + "epoch": 0.10100496120086502, + "ewc_loss": 0.002471923828125, + "ewc_loss_diag": 2.2351741790771484e-06, + "ewc_loss_parallel": 2.250075340270996e-06, + "grad_norm": 4.6532979011535645, + "learning_rate": 3.3615938957185245e-07, + "loss": 0.4322, + "mean_token_accuracy": 0.8542823195457458, + "num_tokens": 30741965.0, + "step": 794 + }, + { + "epoch": 0.10113217147945554, + "ewc_loss": 0.00250244140625, + "ewc_loss_diag": 2.250075340270996e-06, + "ewc_loss_parallel": 2.2798776626586914e-06, + "grad_norm": 4.823664665222168, + "learning_rate": 3.3658329800763035e-07, + "loss": 0.4805, + "mean_token_accuracy": 0.841437578201294, + "num_tokens": 30774389.0, + "step": 795 + }, + { + "epoch": 0.10125938175804605, + "ewc_loss": 0.0025177001953125, + "ewc_loss_diag": 2.250075340270996e-06, + "ewc_loss_parallel": 2.294778823852539e-06, + "grad_norm": 4.629420757293701, + "learning_rate": 3.370072064434082e-07, + "loss": 0.44, + "mean_token_accuracy": 0.8478363752365112, + "num_tokens": 30810957.0, + "step": 796 + }, + { + "epoch": 0.10138659203663657, + "ewc_loss": 0.002471923828125, + "ewc_loss_diag": 2.250075340270996e-06, + "ewc_loss_parallel": 2.250075340270996e-06, + "grad_norm": 4.469390392303467, + "learning_rate": 3.374311148791861e-07, + "loss": 0.4178, + "mean_token_accuracy": 0.8620976209640503, + "num_tokens": 30850448.0, + "step": 797 + }, + { + "epoch": 0.10151380231522707, + "ewc_loss": 0.0024871826171875, + "ewc_loss_diag": 2.250075340270996e-06, + "ewc_loss_parallel": 2.2649765014648438e-06, + "grad_norm": 4.827708721160889, + "learning_rate": 3.3785502331496394e-07, + "loss": 0.4279, + "mean_token_accuracy": 0.8543643951416016, + "num_tokens": 30881215.0, + "step": 798 + }, + { + "epoch": 0.10164101259381758, + "ewc_loss": 0.002532958984375, + "ewc_loss_diag": 2.250075340270996e-06, + "ewc_loss_parallel": 2.3096799850463867e-06, + "grad_norm": 4.85662317276001, + "learning_rate": 3.3827893175074184e-07, + "loss": 0.4464, + "mean_token_accuracy": 0.8515278100967407, + "num_tokens": 30912589.0, + "step": 799 + }, + { + "epoch": 0.1017682228724081, + "ewc_loss": 0.002532958984375, + "ewc_loss_diag": 2.250075340270996e-06, + "ewc_loss_parallel": 2.3096799850463867e-06, + "grad_norm": 4.636258602142334, + "learning_rate": 3.387028401865197e-07, + "loss": 0.4469, + "mean_token_accuracy": 0.8558505773544312, + "num_tokens": 30950766.0, + "step": 800 + }, + { + "epoch": 0.1018954331509986, + "ewc_loss": 0.002471923828125, + "ewc_loss_diag": 2.2649765014648438e-06, + "ewc_loss_parallel": 2.250075340270996e-06, + "grad_norm": 4.514335632324219, + "learning_rate": 3.391267486222976e-07, + "loss": 0.4385, + "mean_token_accuracy": 0.8517608642578125, + "num_tokens": 30989198.0, + "step": 801 + }, + { + "epoch": 0.10202264342958911, + "ewc_loss": 0.00250244140625, + "ewc_loss_diag": 2.2649765014648438e-06, + "ewc_loss_parallel": 2.2798776626586914e-06, + "grad_norm": 4.903257369995117, + "learning_rate": 3.3955065705807543e-07, + "loss": 0.4714, + "mean_token_accuracy": 0.8439974784851074, + "num_tokens": 31020969.0, + "step": 802 + }, + { + "epoch": 0.10214985370817962, + "ewc_loss": 0.0025634765625, + "ewc_loss_diag": 2.2649765014648438e-06, + "ewc_loss_parallel": 2.339482307434082e-06, + "grad_norm": 4.74793815612793, + "learning_rate": 3.3997456549385333e-07, + "loss": 0.4082, + "mean_token_accuracy": 0.8632534742355347, + "num_tokens": 31056712.0, + "step": 803 + }, + { + "epoch": 0.10227706398677013, + "ewc_loss": 0.0025177001953125, + "ewc_loss_diag": 2.2649765014648438e-06, + "ewc_loss_parallel": 2.294778823852539e-06, + "grad_norm": 4.961784362792969, + "learning_rate": 3.403984739296312e-07, + "loss": 0.4718, + "mean_token_accuracy": 0.8435531854629517, + "num_tokens": 31090884.0, + "step": 804 + }, + { + "epoch": 0.10240427426536064, + "ewc_loss": 0.002532958984375, + "ewc_loss_diag": 2.2649765014648438e-06, + "ewc_loss_parallel": 2.3096799850463867e-06, + "grad_norm": 4.647226333618164, + "learning_rate": 3.408223823654091e-07, + "loss": 0.4054, + "mean_token_accuracy": 0.8645779490470886, + "num_tokens": 31128203.0, + "step": 805 + }, + { + "epoch": 0.10253148454395115, + "ewc_loss": 0.00250244140625, + "ewc_loss_diag": 2.2649765014648438e-06, + "ewc_loss_parallel": 2.2798776626586914e-06, + "grad_norm": 4.571200847625732, + "learning_rate": 3.412462908011869e-07, + "loss": 0.4496, + "mean_token_accuracy": 0.8561752438545227, + "num_tokens": 31169041.0, + "step": 806 + }, + { + "epoch": 0.10265869482254165, + "ewc_loss": 0.00250244140625, + "ewc_loss_diag": 2.2649765014648438e-06, + "ewc_loss_parallel": 2.2798776626586914e-06, + "grad_norm": 4.5484137535095215, + "learning_rate": 3.4167019923696477e-07, + "loss": 0.4412, + "mean_token_accuracy": 0.8565827012062073, + "num_tokens": 31216707.0, + "step": 807 + }, + { + "epoch": 0.10278590510113217, + "ewc_loss": 0.002532958984375, + "ewc_loss_diag": 2.2798776626586914e-06, + "ewc_loss_parallel": 2.3096799850463867e-06, + "grad_norm": 4.5416717529296875, + "learning_rate": 3.4209410767274267e-07, + "loss": 0.4025, + "mean_token_accuracy": 0.8664133548736572, + "num_tokens": 31256996.0, + "step": 808 + }, + { + "epoch": 0.10291311537972268, + "ewc_loss": 0.00250244140625, + "ewc_loss_diag": 2.2798776626586914e-06, + "ewc_loss_parallel": 2.2798776626586914e-06, + "grad_norm": 4.617299556732178, + "learning_rate": 3.4251801610852057e-07, + "loss": 0.4655, + "mean_token_accuracy": 0.8430564999580383, + "num_tokens": 31299212.0, + "step": 809 + }, + { + "epoch": 0.1030403256583132, + "ewc_loss": 0.0025177001953125, + "ewc_loss_diag": 2.2798776626586914e-06, + "ewc_loss_parallel": 2.294778823852539e-06, + "grad_norm": 4.71851110458374, + "learning_rate": 3.429419245442984e-07, + "loss": 0.4257, + "mean_token_accuracy": 0.8556865453720093, + "num_tokens": 31336054.0, + "step": 810 + }, + { + "epoch": 0.1031675359369037, + "ewc_loss": 0.002532958984375, + "ewc_loss_diag": 2.2798776626586914e-06, + "ewc_loss_parallel": 2.3096799850463867e-06, + "grad_norm": 4.620654582977295, + "learning_rate": 3.4336583298007626e-07, + "loss": 0.4146, + "mean_token_accuracy": 0.8632978200912476, + "num_tokens": 31372743.0, + "step": 811 + }, + { + "epoch": 0.10329474621549421, + "ewc_loss": 0.002532958984375, + "ewc_loss_diag": 2.2798776626586914e-06, + "ewc_loss_parallel": 2.3096799850463867e-06, + "grad_norm": 4.558590888977051, + "learning_rate": 3.4378974141585416e-07, + "loss": 0.4465, + "mean_token_accuracy": 0.8524379134178162, + "num_tokens": 31412704.0, + "step": 812 + }, + { + "epoch": 0.10342195649408473, + "ewc_loss": 0.0025177001953125, + "ewc_loss_diag": 2.294778823852539e-06, + "ewc_loss_parallel": 2.294778823852539e-06, + "grad_norm": 4.41620397567749, + "learning_rate": 3.4421364985163206e-07, + "loss": 0.3756, + "mean_token_accuracy": 0.8731765151023865, + "num_tokens": 31456751.0, + "step": 813 + }, + { + "epoch": 0.10354916677267523, + "ewc_loss": 0.00250244140625, + "ewc_loss_diag": 2.3096799850463867e-06, + "ewc_loss_parallel": 2.2798776626586914e-06, + "grad_norm": 4.589330673217773, + "learning_rate": 3.446375582874099e-07, + "loss": 0.4392, + "mean_token_accuracy": 0.8563827872276306, + "num_tokens": 31501291.0, + "step": 814 + }, + { + "epoch": 0.10367637705126574, + "ewc_loss": 0.0025482177734375, + "ewc_loss_diag": 2.3096799850463867e-06, + "ewc_loss_parallel": 2.3245811462402344e-06, + "grad_norm": 4.798825740814209, + "learning_rate": 3.4506146672318775e-07, + "loss": 0.4081, + "mean_token_accuracy": 0.8621509075164795, + "num_tokens": 31536420.0, + "step": 815 + }, + { + "epoch": 0.10380358732985626, + "ewc_loss": 0.0025482177734375, + "ewc_loss_diag": 2.3096799850463867e-06, + "ewc_loss_parallel": 2.3245811462402344e-06, + "grad_norm": 4.5060014724731445, + "learning_rate": 3.4548537515896565e-07, + "loss": 0.4277, + "mean_token_accuracy": 0.8582004904747009, + "num_tokens": 31578252.0, + "step": 816 + }, + { + "epoch": 0.10393079760844676, + "ewc_loss": 0.0025177001953125, + "ewc_loss_diag": 2.3245811462402344e-06, + "ewc_loss_parallel": 2.294778823852539e-06, + "grad_norm": 4.647758960723877, + "learning_rate": 3.4590928359474355e-07, + "loss": 0.4349, + "mean_token_accuracy": 0.8572646379470825, + "num_tokens": 31618656.0, + "step": 817 + }, + { + "epoch": 0.10405800788703727, + "ewc_loss": 0.002532958984375, + "ewc_loss_diag": 2.3245811462402344e-06, + "ewc_loss_parallel": 2.3096799850463867e-06, + "grad_norm": 4.656642913818359, + "learning_rate": 3.463331920305214e-07, + "loss": 0.4446, + "mean_token_accuracy": 0.8529086112976074, + "num_tokens": 31660080.0, + "step": 818 + }, + { + "epoch": 0.10418521816562779, + "ewc_loss": 0.002532958984375, + "ewc_loss_diag": 2.3245811462402344e-06, + "ewc_loss_parallel": 2.3096799850463867e-06, + "grad_norm": 4.726684093475342, + "learning_rate": 3.4675710046629924e-07, + "loss": 0.4036, + "mean_token_accuracy": 0.8649632930755615, + "num_tokens": 31694682.0, + "step": 819 + }, + { + "epoch": 0.1043124284442183, + "ewc_loss": 0.0025482177734375, + "ewc_loss_diag": 2.3245811462402344e-06, + "ewc_loss_parallel": 2.3245811462402344e-06, + "grad_norm": 4.583160877227783, + "learning_rate": 3.4718100890207714e-07, + "loss": 0.4456, + "mean_token_accuracy": 0.8539668321609497, + "num_tokens": 31732650.0, + "step": 820 + }, + { + "epoch": 0.1044396387228088, + "ewc_loss": 0.0025177001953125, + "ewc_loss_diag": 2.339482307434082e-06, + "ewc_loss_parallel": 2.294778823852539e-06, + "grad_norm": 4.682031154632568, + "learning_rate": 3.4760491733785504e-07, + "loss": 0.4416, + "mean_token_accuracy": 0.8512126207351685, + "num_tokens": 31768073.0, + "step": 821 + }, + { + "epoch": 0.10456684900139931, + "ewc_loss": 0.0025482177734375, + "ewc_loss_diag": 2.339482307434082e-06, + "ewc_loss_parallel": 2.3245811462402344e-06, + "grad_norm": 4.73130989074707, + "learning_rate": 3.480288257736329e-07, + "loss": 0.4473, + "mean_token_accuracy": 0.8499759435653687, + "num_tokens": 31803876.0, + "step": 822 + }, + { + "epoch": 0.10469405927998983, + "ewc_loss": 0.0025482177734375, + "ewc_loss_diag": 2.339482307434082e-06, + "ewc_loss_parallel": 2.3245811462402344e-06, + "grad_norm": 4.586796283721924, + "learning_rate": 3.4845273420941073e-07, + "loss": 0.4146, + "mean_token_accuracy": 0.8618370294570923, + "num_tokens": 31843355.0, + "step": 823 + }, + { + "epoch": 0.10482126955858033, + "ewc_loss": 0.002532958984375, + "ewc_loss_diag": 2.339482307434082e-06, + "ewc_loss_parallel": 2.3096799850463867e-06, + "grad_norm": 4.65579080581665, + "learning_rate": 3.4887664264518863e-07, + "loss": 0.429, + "mean_token_accuracy": 0.8532091975212097, + "num_tokens": 31878949.0, + "step": 824 + }, + { + "epoch": 0.10494847983717084, + "ewc_loss": 0.002532958984375, + "ewc_loss_diag": 2.3543834686279297e-06, + "ewc_loss_parallel": 2.3096799850463867e-06, + "grad_norm": 4.572150230407715, + "learning_rate": 3.4930055108096653e-07, + "loss": 0.444, + "mean_token_accuracy": 0.8541386127471924, + "num_tokens": 31921801.0, + "step": 825 + }, + { + "epoch": 0.10507569011576136, + "ewc_loss": 0.002532958984375, + "ewc_loss_diag": 2.3543834686279297e-06, + "ewc_loss_parallel": 2.3096799850463867e-06, + "grad_norm": 4.503349304199219, + "learning_rate": 3.497244595167443e-07, + "loss": 0.4087, + "mean_token_accuracy": 0.8632067441940308, + "num_tokens": 31965260.0, + "step": 826 + }, + { + "epoch": 0.10520290039435186, + "ewc_loss": 0.0025482177734375, + "ewc_loss_diag": 2.3543834686279297e-06, + "ewc_loss_parallel": 2.3245811462402344e-06, + "grad_norm": 4.7491326332092285, + "learning_rate": 3.501483679525222e-07, + "loss": 0.4642, + "mean_token_accuracy": 0.8499563932418823, + "num_tokens": 32002312.0, + "step": 827 + }, + { + "epoch": 0.10533011067294237, + "ewc_loss": 0.0025634765625, + "ewc_loss_diag": 2.3543834686279297e-06, + "ewc_loss_parallel": 2.339482307434082e-06, + "grad_norm": 4.6376495361328125, + "learning_rate": 3.505722763883001e-07, + "loss": 0.4078, + "mean_token_accuracy": 0.8613065481185913, + "num_tokens": 32042919.0, + "step": 828 + }, + { + "epoch": 0.10545732095153289, + "ewc_loss": 0.0025482177734375, + "ewc_loss_diag": 2.3543834686279297e-06, + "ewc_loss_parallel": 2.3245811462402344e-06, + "grad_norm": 4.626224517822266, + "learning_rate": 3.50996184824078e-07, + "loss": 0.4501, + "mean_token_accuracy": 0.8497263193130493, + "num_tokens": 32079052.0, + "step": 829 + }, + { + "epoch": 0.10558453123012339, + "ewc_loss": 0.0025634765625, + "ewc_loss_diag": 2.3543834686279297e-06, + "ewc_loss_parallel": 2.339482307434082e-06, + "grad_norm": 4.620519161224365, + "learning_rate": 3.514200932598558e-07, + "loss": 0.48, + "mean_token_accuracy": 0.8418017625808716, + "num_tokens": 32124302.0, + "step": 830 + }, + { + "epoch": 0.1057117415087139, + "ewc_loss": 0.0025634765625, + "ewc_loss_diag": 2.3692846298217773e-06, + "ewc_loss_parallel": 2.3245811462402344e-06, + "grad_norm": 4.6609721183776855, + "learning_rate": 3.518440016956337e-07, + "loss": 0.4415, + "mean_token_accuracy": 0.8554748892784119, + "num_tokens": 32165339.0, + "step": 831 + }, + { + "epoch": 0.10583895178730442, + "ewc_loss": 0.0025634765625, + "ewc_loss_diag": 2.384185791015625e-06, + "ewc_loss_parallel": 2.3245811462402344e-06, + "grad_norm": 4.551398277282715, + "learning_rate": 3.522679101314116e-07, + "loss": 0.4343, + "mean_token_accuracy": 0.8586567044258118, + "num_tokens": 32209600.0, + "step": 832 + }, + { + "epoch": 0.10596616206589493, + "ewc_loss": 0.0025482177734375, + "ewc_loss_diag": 2.384185791015625e-06, + "ewc_loss_parallel": 2.3096799850463867e-06, + "grad_norm": 4.6454854011535645, + "learning_rate": 3.526918185671895e-07, + "loss": 0.4607, + "mean_token_accuracy": 0.8474937677383423, + "num_tokens": 32252643.0, + "step": 833 + }, + { + "epoch": 0.10609337234448543, + "ewc_loss": 0.0026092529296875, + "ewc_loss_diag": 2.384185791015625e-06, + "ewc_loss_parallel": 2.3692846298217773e-06, + "grad_norm": 4.812746524810791, + "learning_rate": 3.531157270029673e-07, + "loss": 0.392, + "mean_token_accuracy": 0.8697580099105835, + "num_tokens": 32287610.0, + "step": 834 + }, + { + "epoch": 0.10622058262307595, + "ewc_loss": 0.00262451171875, + "ewc_loss_diag": 2.384185791015625e-06, + "ewc_loss_parallel": 2.384185791015625e-06, + "grad_norm": 4.602745532989502, + "learning_rate": 3.535396354387452e-07, + "loss": 0.4082, + "mean_token_accuracy": 0.8636078834533691, + "num_tokens": 32327936.0, + "step": 835 + }, + { + "epoch": 0.10634779290166646, + "ewc_loss": 0.0025482177734375, + "ewc_loss_diag": 2.384185791015625e-06, + "ewc_loss_parallel": 2.3096799850463867e-06, + "grad_norm": 4.790728569030762, + "learning_rate": 3.539635438745231e-07, + "loss": 0.442, + "mean_token_accuracy": 0.852357029914856, + "num_tokens": 32366762.0, + "step": 836 + }, + { + "epoch": 0.10647500318025696, + "ewc_loss": 0.002655029296875, + "ewc_loss_diag": 2.3990869522094727e-06, + "ewc_loss_parallel": 2.4139881134033203e-06, + "grad_norm": 4.547393321990967, + "learning_rate": 3.54387452310301e-07, + "loss": 0.4248, + "mean_token_accuracy": 0.8578025102615356, + "num_tokens": 32411798.0, + "step": 837 + }, + { + "epoch": 0.10660221345884748, + "ewc_loss": 0.0025634765625, + "ewc_loss_diag": 2.3990869522094727e-06, + "ewc_loss_parallel": 2.3245811462402344e-06, + "grad_norm": 4.63274621963501, + "learning_rate": 3.548113607460788e-07, + "loss": 0.3976, + "mean_token_accuracy": 0.8685393333435059, + "num_tokens": 32448022.0, + "step": 838 + }, + { + "epoch": 0.10672942373743799, + "ewc_loss": 0.0026092529296875, + "ewc_loss_diag": 2.3990869522094727e-06, + "ewc_loss_parallel": 2.3692846298217773e-06, + "grad_norm": 4.646221160888672, + "learning_rate": 3.552352691818567e-07, + "loss": 0.4167, + "mean_token_accuracy": 0.862809419631958, + "num_tokens": 32486608.0, + "step": 839 + }, + { + "epoch": 0.10685663401602849, + "ewc_loss": 0.0026092529296875, + "ewc_loss_diag": 2.3990869522094727e-06, + "ewc_loss_parallel": 2.3692846298217773e-06, + "grad_norm": 4.597223281860352, + "learning_rate": 3.556591776176346e-07, + "loss": 0.3999, + "mean_token_accuracy": 0.8659024834632874, + "num_tokens": 32529956.0, + "step": 840 + }, + { + "epoch": 0.106983844294619, + "ewc_loss": 0.0025787353515625, + "ewc_loss_diag": 2.3990869522094727e-06, + "ewc_loss_parallel": 2.339482307434082e-06, + "grad_norm": 4.960271835327148, + "learning_rate": 3.560830860534125e-07, + "loss": 0.4338, + "mean_token_accuracy": 0.8571613430976868, + "num_tokens": 32562632.0, + "step": 841 + }, + { + "epoch": 0.10711105457320952, + "ewc_loss": 0.002716064453125, + "ewc_loss_diag": 2.3990869522094727e-06, + "ewc_loss_parallel": 2.473592758178711e-06, + "grad_norm": 4.954997539520264, + "learning_rate": 3.565069944891903e-07, + "loss": 0.467, + "mean_token_accuracy": 0.8450220823287964, + "num_tokens": 32596222.0, + "step": 842 + }, + { + "epoch": 0.10723826485180002, + "ewc_loss": 0.0026702880859375, + "ewc_loss_diag": 2.3990869522094727e-06, + "ewc_loss_parallel": 2.428889274597168e-06, + "grad_norm": 4.82945442199707, + "learning_rate": 3.569309029249682e-07, + "loss": 0.4458, + "mean_token_accuracy": 0.8525278568267822, + "num_tokens": 32631036.0, + "step": 843 + }, + { + "epoch": 0.10736547513039053, + "ewc_loss": 0.00262451171875, + "ewc_loss_diag": 2.3990869522094727e-06, + "ewc_loss_parallel": 2.384185791015625e-06, + "grad_norm": 4.63046932220459, + "learning_rate": 3.573548113607461e-07, + "loss": 0.4314, + "mean_token_accuracy": 0.8567546606063843, + "num_tokens": 32669476.0, + "step": 844 + }, + { + "epoch": 0.10749268540898105, + "ewc_loss": 0.002593994140625, + "ewc_loss_diag": 2.3990869522094727e-06, + "ewc_loss_parallel": 2.3543834686279297e-06, + "grad_norm": 4.860330104827881, + "learning_rate": 3.577787197965239e-07, + "loss": 0.4725, + "mean_token_accuracy": 0.8446064591407776, + "num_tokens": 32704478.0, + "step": 845 + }, + { + "epoch": 0.10761989568757156, + "ewc_loss": 0.002685546875, + "ewc_loss_diag": 2.3990869522094727e-06, + "ewc_loss_parallel": 2.4437904357910156e-06, + "grad_norm": 4.568073272705078, + "learning_rate": 3.5820262823230177e-07, + "loss": 0.4046, + "mean_token_accuracy": 0.8686227202415466, + "num_tokens": 32745363.0, + "step": 846 + }, + { + "epoch": 0.10774710596616206, + "ewc_loss": 0.0025787353515625, + "ewc_loss_diag": 2.3990869522094727e-06, + "ewc_loss_parallel": 2.339482307434082e-06, + "grad_norm": 4.825430870056152, + "learning_rate": 3.5862653666807967e-07, + "loss": 0.4861, + "mean_token_accuracy": 0.8399807214736938, + "num_tokens": 32781515.0, + "step": 847 + }, + { + "epoch": 0.10787431624475258, + "ewc_loss": 0.002716064453125, + "ewc_loss_diag": 2.428889274597168e-06, + "ewc_loss_parallel": 2.473592758178711e-06, + "grad_norm": 4.7038655281066895, + "learning_rate": 3.5905044510385757e-07, + "loss": 0.4491, + "mean_token_accuracy": 0.8520631790161133, + "num_tokens": 32821791.0, + "step": 848 + }, + { + "epoch": 0.10800152652334309, + "ewc_loss": 0.0026397705078125, + "ewc_loss_diag": 2.428889274597168e-06, + "ewc_loss_parallel": 2.3990869522094727e-06, + "grad_norm": 4.703080654144287, + "learning_rate": 3.594743535396354e-07, + "loss": 0.436, + "mean_token_accuracy": 0.8555315136909485, + "num_tokens": 32859169.0, + "step": 849 + }, + { + "epoch": 0.10812873680193359, + "ewc_loss": 0.00262451171875, + "ewc_loss_diag": 2.428889274597168e-06, + "ewc_loss_parallel": 2.384185791015625e-06, + "grad_norm": 4.63823938369751, + "learning_rate": 3.5989826197541326e-07, + "loss": 0.4609, + "mean_token_accuracy": 0.8484243154525757, + "num_tokens": 32901493.0, + "step": 850 + }, + { + "epoch": 0.1082559470805241, + "ewc_loss": 0.002655029296875, + "ewc_loss_diag": 2.428889274597168e-06, + "ewc_loss_parallel": 2.4139881134033203e-06, + "grad_norm": 4.551346778869629, + "learning_rate": 3.6032217041119116e-07, + "loss": 0.4118, + "mean_token_accuracy": 0.8593107461929321, + "num_tokens": 32945203.0, + "step": 851 + }, + { + "epoch": 0.10838315735911462, + "ewc_loss": 0.0026397705078125, + "ewc_loss_diag": 2.428889274597168e-06, + "ewc_loss_parallel": 2.3990869522094727e-06, + "grad_norm": 4.588437557220459, + "learning_rate": 3.6074607884696906e-07, + "loss": 0.41, + "mean_token_accuracy": 0.8682833909988403, + "num_tokens": 32989835.0, + "step": 852 + }, + { + "epoch": 0.10851036763770512, + "ewc_loss": 0.0026702880859375, + "ewc_loss_diag": 2.4586915969848633e-06, + "ewc_loss_parallel": 2.428889274597168e-06, + "grad_norm": 4.5422749519348145, + "learning_rate": 3.611699872827469e-07, + "loss": 0.3819, + "mean_token_accuracy": 0.8700084090232849, + "num_tokens": 33030206.0, + "step": 853 + }, + { + "epoch": 0.10863757791629564, + "ewc_loss": 0.002655029296875, + "ewc_loss_diag": 2.4437904357910156e-06, + "ewc_loss_parallel": 2.4139881134033203e-06, + "grad_norm": 4.801825046539307, + "learning_rate": 3.6159389571852475e-07, + "loss": 0.4801, + "mean_token_accuracy": 0.8437409400939941, + "num_tokens": 33068064.0, + "step": 854 + }, + { + "epoch": 0.10876478819488615, + "ewc_loss": 0.002716064453125, + "ewc_loss_diag": 2.4586915969848633e-06, + "ewc_loss_parallel": 2.473592758178711e-06, + "grad_norm": 4.676944255828857, + "learning_rate": 3.6201780415430265e-07, + "loss": 0.4392, + "mean_token_accuracy": 0.8557655811309814, + "num_tokens": 33112473.0, + "step": 855 + }, + { + "epoch": 0.10889199847347665, + "ewc_loss": 0.0026702880859375, + "ewc_loss_diag": 2.4586915969848633e-06, + "ewc_loss_parallel": 2.428889274597168e-06, + "grad_norm": 4.659951686859131, + "learning_rate": 3.6244171259008055e-07, + "loss": 0.4524, + "mean_token_accuracy": 0.8517106771469116, + "num_tokens": 33151368.0, + "step": 856 + }, + { + "epoch": 0.10901920875206716, + "ewc_loss": 0.0026702880859375, + "ewc_loss_diag": 2.4586915969848633e-06, + "ewc_loss_parallel": 2.428889274597168e-06, + "grad_norm": 4.749464511871338, + "learning_rate": 3.628656210258584e-07, + "loss": 0.4586, + "mean_token_accuracy": 0.8474034070968628, + "num_tokens": 33190114.0, + "step": 857 + }, + { + "epoch": 0.10914641903065768, + "ewc_loss": 0.002716064453125, + "ewc_loss_diag": 2.4884939193725586e-06, + "ewc_loss_parallel": 2.473592758178711e-06, + "grad_norm": 4.676156997680664, + "learning_rate": 3.6328952946163624e-07, + "loss": 0.4123, + "mean_token_accuracy": 0.8617921471595764, + "num_tokens": 33227444.0, + "step": 858 + }, + { + "epoch": 0.1092736293092482, + "ewc_loss": 0.0026702880859375, + "ewc_loss_diag": 2.4884939193725586e-06, + "ewc_loss_parallel": 2.428889274597168e-06, + "grad_norm": 4.6009087562561035, + "learning_rate": 3.6371343789741414e-07, + "loss": 0.3906, + "mean_token_accuracy": 0.869072437286377, + "num_tokens": 33268807.0, + "step": 859 + }, + { + "epoch": 0.1094008395878387, + "ewc_loss": 0.002655029296875, + "ewc_loss_diag": 2.4884939193725586e-06, + "ewc_loss_parallel": 2.4139881134033203e-06, + "grad_norm": 4.757453918457031, + "learning_rate": 3.6413734633319204e-07, + "loss": 0.4543, + "mean_token_accuracy": 0.8522282838821411, + "num_tokens": 33308190.0, + "step": 860 + }, + { + "epoch": 0.10952804986642921, + "ewc_loss": 0.0027313232421875, + "ewc_loss_diag": 2.5033950805664062e-06, + "ewc_loss_parallel": 2.4884939193725586e-06, + "grad_norm": 4.714212894439697, + "learning_rate": 3.645612547689699e-07, + "loss": 0.4074, + "mean_token_accuracy": 0.8606286644935608, + "num_tokens": 33345602.0, + "step": 861 + }, + { + "epoch": 0.10965526014501972, + "ewc_loss": 0.002685546875, + "ewc_loss_diag": 2.5033950805664062e-06, + "ewc_loss_parallel": 2.4437904357910156e-06, + "grad_norm": 4.524340629577637, + "learning_rate": 3.6498516320474773e-07, + "loss": 0.3974, + "mean_token_accuracy": 0.8699963092803955, + "num_tokens": 33389557.0, + "step": 862 + }, + { + "epoch": 0.10978247042361022, + "ewc_loss": 0.002655029296875, + "ewc_loss_diag": 2.5033950805664062e-06, + "ewc_loss_parallel": 2.4139881134033203e-06, + "grad_norm": 4.622642517089844, + "learning_rate": 3.6540907164052563e-07, + "loss": 0.3779, + "mean_token_accuracy": 0.8744686841964722, + "num_tokens": 33427568.0, + "step": 863 + }, + { + "epoch": 0.10990968070220074, + "ewc_loss": 0.0027008056640625, + "ewc_loss_diag": 2.4884939193725586e-06, + "ewc_loss_parallel": 2.4586915969848633e-06, + "grad_norm": 4.815435409545898, + "learning_rate": 3.658329800763035e-07, + "loss": 0.4694, + "mean_token_accuracy": 0.8431452512741089, + "num_tokens": 33466892.0, + "step": 864 + }, + { + "epoch": 0.11003689098079125, + "ewc_loss": 0.00274658203125, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.4884939193725586e-06, + "grad_norm": 4.824426174163818, + "learning_rate": 3.662568885120814e-07, + "loss": 0.4254, + "mean_token_accuracy": 0.8584744334220886, + "num_tokens": 33502260.0, + "step": 865 + }, + { + "epoch": 0.11016410125938175, + "ewc_loss": 0.002716064453125, + "ewc_loss_diag": 2.5033950805664062e-06, + "ewc_loss_parallel": 2.473592758178711e-06, + "grad_norm": 4.758352756500244, + "learning_rate": 3.666807969478592e-07, + "loss": 0.4674, + "mean_token_accuracy": 0.8438858985900879, + "num_tokens": 33541584.0, + "step": 866 + }, + { + "epoch": 0.11029131153797227, + "ewc_loss": 0.0027313232421875, + "ewc_loss_diag": 2.5033950805664062e-06, + "ewc_loss_parallel": 2.4884939193725586e-06, + "grad_norm": 4.7184672355651855, + "learning_rate": 3.671047053836371e-07, + "loss": 0.443, + "mean_token_accuracy": 0.8493238687515259, + "num_tokens": 33579324.0, + "step": 867 + }, + { + "epoch": 0.11041852181656278, + "ewc_loss": 0.002716064453125, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.4586915969848633e-06, + "grad_norm": 4.738546848297119, + "learning_rate": 3.6752861381941497e-07, + "loss": 0.419, + "mean_token_accuracy": 0.8602910041809082, + "num_tokens": 33616760.0, + "step": 868 + }, + { + "epoch": 0.11054573209515328, + "ewc_loss": 0.002685546875, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.4437904357910156e-06, + "grad_norm": 4.841792106628418, + "learning_rate": 3.6795252225519287e-07, + "loss": 0.4684, + "mean_token_accuracy": 0.8480024337768555, + "num_tokens": 33655140.0, + "step": 869 + }, + { + "epoch": 0.1106729423737438, + "ewc_loss": 0.0027618408203125, + "ewc_loss_diag": 2.5033950805664062e-06, + "ewc_loss_parallel": 2.518296241760254e-06, + "grad_norm": 4.650030136108398, + "learning_rate": 3.6837643069097077e-07, + "loss": 0.3808, + "mean_token_accuracy": 0.8726588487625122, + "num_tokens": 33693966.0, + "step": 870 + }, + { + "epoch": 0.11080015265233431, + "ewc_loss": 0.002716064453125, + "ewc_loss_diag": 2.5033950805664062e-06, + "ewc_loss_parallel": 2.473592758178711e-06, + "grad_norm": 4.545680046081543, + "learning_rate": 3.688003391267486e-07, + "loss": 0.426, + "mean_token_accuracy": 0.8616846203804016, + "num_tokens": 33739942.0, + "step": 871 + }, + { + "epoch": 0.11092736293092482, + "ewc_loss": 0.002685546875, + "ewc_loss_diag": 2.5033950805664062e-06, + "ewc_loss_parallel": 2.4437904357910156e-06, + "grad_norm": 4.453439712524414, + "learning_rate": 3.6922424756252646e-07, + "loss": 0.4006, + "mean_token_accuracy": 0.8664568066596985, + "num_tokens": 33789980.0, + "step": 872 + }, + { + "epoch": 0.11105457320951533, + "ewc_loss": 0.0027008056640625, + "ewc_loss_diag": 2.5033950805664062e-06, + "ewc_loss_parallel": 2.4586915969848633e-06, + "grad_norm": 4.95708703994751, + "learning_rate": 3.6964815599830436e-07, + "loss": 0.4549, + "mean_token_accuracy": 0.8487563133239746, + "num_tokens": 33824226.0, + "step": 873 + }, + { + "epoch": 0.11118178348810584, + "ewc_loss": 0.0028228759765625, + "ewc_loss_diag": 2.5033950805664062e-06, + "ewc_loss_parallel": 2.5779008865356445e-06, + "grad_norm": 4.828579902648926, + "learning_rate": 3.7007206443408226e-07, + "loss": 0.3858, + "mean_token_accuracy": 0.8718733787536621, + "num_tokens": 33861725.0, + "step": 874 + }, + { + "epoch": 0.11130899376669635, + "ewc_loss": 0.002777099609375, + "ewc_loss_diag": 2.5033950805664062e-06, + "ewc_loss_parallel": 2.5331974029541016e-06, + "grad_norm": 4.828266620635986, + "learning_rate": 3.704959728698601e-07, + "loss": 0.4938, + "mean_token_accuracy": 0.8385657072067261, + "num_tokens": 33900754.0, + "step": 875 + }, + { + "epoch": 0.11143620404528685, + "ewc_loss": 0.002777099609375, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.518296241760254e-06, + "grad_norm": 4.734104156494141, + "learning_rate": 3.7091988130563795e-07, + "loss": 0.4337, + "mean_token_accuracy": 0.8579641580581665, + "num_tokens": 33941514.0, + "step": 876 + }, + { + "epoch": 0.11156341432387737, + "ewc_loss": 0.002716064453125, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.4586915969848633e-06, + "grad_norm": 4.570194244384766, + "learning_rate": 3.7134378974141585e-07, + "loss": 0.4345, + "mean_token_accuracy": 0.8562740087509155, + "num_tokens": 33989959.0, + "step": 877 + }, + { + "epoch": 0.11169062460246788, + "ewc_loss": 0.002685546875, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.4437904357910156e-06, + "grad_norm": 4.799293041229248, + "learning_rate": 3.7176769817719375e-07, + "loss": 0.461, + "mean_token_accuracy": 0.849136471748352, + "num_tokens": 34026696.0, + "step": 878 + }, + { + "epoch": 0.11181783488105838, + "ewc_loss": 0.0028076171875, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.562999725341797e-06, + "grad_norm": 4.6921162605285645, + "learning_rate": 3.7219160661297154e-07, + "loss": 0.3691, + "mean_token_accuracy": 0.8779081106185913, + "num_tokens": 34064724.0, + "step": 879 + }, + { + "epoch": 0.1119450451596489, + "ewc_loss": 0.002777099609375, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.518296241760254e-06, + "grad_norm": 4.856264591217041, + "learning_rate": 3.7261551504874944e-07, + "loss": 0.4375, + "mean_token_accuracy": 0.8571827411651611, + "num_tokens": 34099198.0, + "step": 880 + }, + { + "epoch": 0.11207225543823941, + "ewc_loss": 0.002777099609375, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.5331974029541016e-06, + "grad_norm": 4.718142032623291, + "learning_rate": 3.7303942348452734e-07, + "loss": 0.3827, + "mean_token_accuracy": 0.870418906211853, + "num_tokens": 34134041.0, + "step": 881 + }, + { + "epoch": 0.11219946571682991, + "ewc_loss": 0.002777099609375, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.518296241760254e-06, + "grad_norm": 4.663057804107666, + "learning_rate": 3.7346333192030524e-07, + "loss": 0.3883, + "mean_token_accuracy": 0.8729684352874756, + "num_tokens": 34174981.0, + "step": 882 + }, + { + "epoch": 0.11232667599542043, + "ewc_loss": 0.00274658203125, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.5033950805664062e-06, + "grad_norm": 4.619054317474365, + "learning_rate": 3.7388724035608303e-07, + "loss": 0.4083, + "mean_token_accuracy": 0.8659716844558716, + "num_tokens": 34219616.0, + "step": 883 + }, + { + "epoch": 0.11245388627401094, + "ewc_loss": 0.002777099609375, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.5331974029541016e-06, + "grad_norm": 4.860881328582764, + "learning_rate": 3.7431114879186093e-07, + "loss": 0.4544, + "mean_token_accuracy": 0.8503944873809814, + "num_tokens": 34255280.0, + "step": 884 + }, + { + "epoch": 0.11258109655260146, + "ewc_loss": 0.002838134765625, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.5779008865356445e-06, + "grad_norm": 4.874965667724609, + "learning_rate": 3.7473505722763883e-07, + "loss": 0.4701, + "mean_token_accuracy": 0.84566730260849, + "num_tokens": 34290741.0, + "step": 885 + }, + { + "epoch": 0.11270830683119196, + "ewc_loss": 0.002838134765625, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.5779008865356445e-06, + "grad_norm": 4.79910945892334, + "learning_rate": 3.7515896566341673e-07, + "loss": 0.4239, + "mean_token_accuracy": 0.8553659915924072, + "num_tokens": 34327090.0, + "step": 886 + }, + { + "epoch": 0.11283551710978247, + "ewc_loss": 0.002838134765625, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.5779008865356445e-06, + "grad_norm": 4.73891019821167, + "learning_rate": 3.755828740991945e-07, + "loss": 0.4404, + "mean_token_accuracy": 0.8544859886169434, + "num_tokens": 34370487.0, + "step": 887 + }, + { + "epoch": 0.11296272738837299, + "ewc_loss": 0.0028076171875, + "ewc_loss_diag": 2.518296241760254e-06, + "ewc_loss_parallel": 2.5480985641479492e-06, + "grad_norm": 4.766743183135986, + "learning_rate": 3.760067825349724e-07, + "loss": 0.4121, + "mean_token_accuracy": 0.8634305596351624, + "num_tokens": 34407204.0, + "step": 888 + }, + { + "epoch": 0.11308993766696349, + "ewc_loss": 0.0028228759765625, + "ewc_loss_diag": 2.5480985641479492e-06, + "ewc_loss_parallel": 2.562999725341797e-06, + "grad_norm": 4.559676170349121, + "learning_rate": 3.764306909707503e-07, + "loss": 0.3783, + "mean_token_accuracy": 0.8720571398735046, + "num_tokens": 34450561.0, + "step": 889 + }, + { + "epoch": 0.113217147945554, + "ewc_loss": 0.002777099609375, + "ewc_loss_diag": 2.5480985641479492e-06, + "ewc_loss_parallel": 2.518296241760254e-06, + "grad_norm": 4.919986724853516, + "learning_rate": 3.768545994065282e-07, + "loss": 0.4289, + "mean_token_accuracy": 0.8577309846878052, + "num_tokens": 34481865.0, + "step": 890 + }, + { + "epoch": 0.11334435822414451, + "ewc_loss": 0.00286865234375, + "ewc_loss_diag": 2.5480985641479492e-06, + "ewc_loss_parallel": 2.60770320892334e-06, + "grad_norm": 4.773432731628418, + "learning_rate": 3.77278507842306e-07, + "loss": 0.4353, + "mean_token_accuracy": 0.8552359938621521, + "num_tokens": 34523716.0, + "step": 891 + }, + { + "epoch": 0.11347156850273502, + "ewc_loss": 0.0028228759765625, + "ewc_loss_diag": 2.5480985641479492e-06, + "ewc_loss_parallel": 2.562999725341797e-06, + "grad_norm": 4.729046821594238, + "learning_rate": 3.777024162780839e-07, + "loss": 0.4135, + "mean_token_accuracy": 0.8615119457244873, + "num_tokens": 34562600.0, + "step": 892 + }, + { + "epoch": 0.11359877878132553, + "ewc_loss": 0.0028228759765625, + "ewc_loss_diag": 2.562999725341797e-06, + "ewc_loss_parallel": 2.562999725341797e-06, + "grad_norm": 4.712028980255127, + "learning_rate": 3.781263247138618e-07, + "loss": 0.3961, + "mean_token_accuracy": 0.8647950291633606, + "num_tokens": 34601774.0, + "step": 893 + }, + { + "epoch": 0.11372598905991604, + "ewc_loss": 0.0027923583984375, + "ewc_loss_diag": 2.562999725341797e-06, + "ewc_loss_parallel": 2.5331974029541016e-06, + "grad_norm": 4.847597122192383, + "learning_rate": 3.785502331496397e-07, + "loss": 0.4273, + "mean_token_accuracy": 0.8529124855995178, + "num_tokens": 34637753.0, + "step": 894 + }, + { + "epoch": 0.11385319933850654, + "ewc_loss": 0.0028533935546875, + "ewc_loss_diag": 2.562999725341797e-06, + "ewc_loss_parallel": 2.592802047729492e-06, + "grad_norm": 4.767055034637451, + "learning_rate": 3.789741415854175e-07, + "loss": 0.4501, + "mean_token_accuracy": 0.8510909676551819, + "num_tokens": 34679151.0, + "step": 895 + }, + { + "epoch": 0.11398040961709706, + "ewc_loss": 0.0028228759765625, + "ewc_loss_diag": 2.562999725341797e-06, + "ewc_loss_parallel": 2.562999725341797e-06, + "grad_norm": 4.705942153930664, + "learning_rate": 3.793980500211954e-07, + "loss": 0.4319, + "mean_token_accuracy": 0.8580830693244934, + "num_tokens": 34718784.0, + "step": 896 + }, + { + "epoch": 0.11410761989568757, + "ewc_loss": 0.002838134765625, + "ewc_loss_diag": 2.562999725341797e-06, + "ewc_loss_parallel": 2.5779008865356445e-06, + "grad_norm": 4.698460102081299, + "learning_rate": 3.798219584569733e-07, + "loss": 0.3736, + "mean_token_accuracy": 0.874287486076355, + "num_tokens": 34755185.0, + "step": 897 + }, + { + "epoch": 0.11423483017427809, + "ewc_loss": 0.0028076171875, + "ewc_loss_diag": 2.562999725341797e-06, + "ewc_loss_parallel": 2.5480985641479492e-06, + "grad_norm": 4.887856483459473, + "learning_rate": 3.8024586689275115e-07, + "loss": 0.4851, + "mean_token_accuracy": 0.8445873856544495, + "num_tokens": 34794513.0, + "step": 898 + }, + { + "epoch": 0.11436204045286859, + "ewc_loss": 0.00286865234375, + "ewc_loss_diag": 2.5779008865356445e-06, + "ewc_loss_parallel": 2.60770320892334e-06, + "grad_norm": 4.85175085067749, + "learning_rate": 3.80669775328529e-07, + "loss": 0.4176, + "mean_token_accuracy": 0.8596330881118774, + "num_tokens": 34828937.0, + "step": 899 + }, + { + "epoch": 0.1144892507314591, + "ewc_loss": 0.00286865234375, + "ewc_loss_diag": 2.5779008865356445e-06, + "ewc_loss_parallel": 2.60770320892334e-06, + "grad_norm": 4.796310901641846, + "learning_rate": 3.810936837643069e-07, + "loss": 0.4004, + "mean_token_accuracy": 0.8653970956802368, + "num_tokens": 34866358.0, + "step": 900 + }, + { + "epoch": 0.11461646101004962, + "ewc_loss": 0.0028839111328125, + "ewc_loss_diag": 2.5779008865356445e-06, + "ewc_loss_parallel": 2.6226043701171875e-06, + "grad_norm": 4.760780334472656, + "learning_rate": 3.815175922000848e-07, + "loss": 0.3832, + "mean_token_accuracy": 0.8695778250694275, + "num_tokens": 34901026.0, + "step": 901 + }, + { + "epoch": 0.11474367128864012, + "ewc_loss": 0.00286865234375, + "ewc_loss_diag": 2.5779008865356445e-06, + "ewc_loss_parallel": 2.60770320892334e-06, + "grad_norm": 4.674431324005127, + "learning_rate": 3.8194150063586264e-07, + "loss": 0.451, + "mean_token_accuracy": 0.8515045046806335, + "num_tokens": 34941529.0, + "step": 902 + }, + { + "epoch": 0.11487088156723063, + "ewc_loss": 0.0028533935546875, + "ewc_loss_diag": 2.5779008865356445e-06, + "ewc_loss_parallel": 2.592802047729492e-06, + "grad_norm": 4.6782355308532715, + "learning_rate": 3.823654090716405e-07, + "loss": 0.3981, + "mean_token_accuracy": 0.86704421043396, + "num_tokens": 34983432.0, + "step": 903 + }, + { + "epoch": 0.11499809184582115, + "ewc_loss": 0.0028533935546875, + "ewc_loss_diag": 2.60770320892334e-06, + "ewc_loss_parallel": 2.592802047729492e-06, + "grad_norm": 4.727399826049805, + "learning_rate": 3.827893175074184e-07, + "loss": 0.4206, + "mean_token_accuracy": 0.8654200434684753, + "num_tokens": 35025268.0, + "step": 904 + }, + { + "epoch": 0.11512530212441165, + "ewc_loss": 0.0028839111328125, + "ewc_loss_diag": 2.60770320892334e-06, + "ewc_loss_parallel": 2.6226043701171875e-06, + "grad_norm": 4.943138122558594, + "learning_rate": 3.832132259431963e-07, + "loss": 0.3849, + "mean_token_accuracy": 0.8717172145843506, + "num_tokens": 35058915.0, + "step": 905 + }, + { + "epoch": 0.11525251240300216, + "ewc_loss": 0.002899169921875, + "ewc_loss_diag": 2.60770320892334e-06, + "ewc_loss_parallel": 2.637505531311035e-06, + "grad_norm": 4.81934928894043, + "learning_rate": 3.8363713437897413e-07, + "loss": 0.4576, + "mean_token_accuracy": 0.8473927974700928, + "num_tokens": 35101673.0, + "step": 906 + }, + { + "epoch": 0.11537972268159268, + "ewc_loss": 0.0028839111328125, + "ewc_loss_diag": 2.60770320892334e-06, + "ewc_loss_parallel": 2.6226043701171875e-06, + "grad_norm": 4.830761432647705, + "learning_rate": 3.8406104281475197e-07, + "loss": 0.4776, + "mean_token_accuracy": 0.8443398475646973, + "num_tokens": 35139769.0, + "step": 907 + }, + { + "epoch": 0.11550693296018319, + "ewc_loss": 0.002899169921875, + "ewc_loss_diag": 2.60770320892334e-06, + "ewc_loss_parallel": 2.637505531311035e-06, + "grad_norm": 4.884254455566406, + "learning_rate": 3.8448495125052987e-07, + "loss": 0.4524, + "mean_token_accuracy": 0.8479505777359009, + "num_tokens": 35179329.0, + "step": 908 + }, + { + "epoch": 0.11563414323877369, + "ewc_loss": 0.002899169921875, + "ewc_loss_diag": 2.60770320892334e-06, + "ewc_loss_parallel": 2.637505531311035e-06, + "grad_norm": 4.631000995635986, + "learning_rate": 3.8490885968630777e-07, + "loss": 0.4036, + "mean_token_accuracy": 0.8660658001899719, + "num_tokens": 35223257.0, + "step": 909 + }, + { + "epoch": 0.1157613535173642, + "ewc_loss": 0.0028533935546875, + "ewc_loss_diag": 2.60770320892334e-06, + "ewc_loss_parallel": 2.592802047729492e-06, + "grad_norm": 4.644137859344482, + "learning_rate": 3.853327681220856e-07, + "loss": 0.4162, + "mean_token_accuracy": 0.85957932472229, + "num_tokens": 35267915.0, + "step": 910 + }, + { + "epoch": 0.11588856379595472, + "ewc_loss": 0.00286865234375, + "ewc_loss_diag": 2.6226043701171875e-06, + "ewc_loss_parallel": 2.60770320892334e-06, + "grad_norm": 4.985711574554443, + "learning_rate": 3.8575667655786346e-07, + "loss": 0.4015, + "mean_token_accuracy": 0.8660765886306763, + "num_tokens": 35300267.0, + "step": 911 + }, + { + "epoch": 0.11601577407454522, + "ewc_loss": 0.0029296875, + "ewc_loss_diag": 2.6226043701171875e-06, + "ewc_loss_parallel": 2.6673078536987305e-06, + "grad_norm": 4.793565273284912, + "learning_rate": 3.8618058499364136e-07, + "loss": 0.422, + "mean_token_accuracy": 0.8607295751571655, + "num_tokens": 35338798.0, + "step": 912 + }, + { + "epoch": 0.11614298435313573, + "ewc_loss": 0.0029144287109375, + "ewc_loss_diag": 2.6226043701171875e-06, + "ewc_loss_parallel": 2.652406692504883e-06, + "grad_norm": 4.675042629241943, + "learning_rate": 3.8660449342941926e-07, + "loss": 0.4631, + "mean_token_accuracy": 0.8517329096794128, + "num_tokens": 35382630.0, + "step": 913 + }, + { + "epoch": 0.11627019463172625, + "ewc_loss": 0.0028839111328125, + "ewc_loss_diag": 2.6226043701171875e-06, + "ewc_loss_parallel": 2.6226043701171875e-06, + "grad_norm": 4.843438625335693, + "learning_rate": 3.870284018651971e-07, + "loss": 0.4344, + "mean_token_accuracy": 0.8573346138000488, + "num_tokens": 35421135.0, + "step": 914 + }, + { + "epoch": 0.11639740491031675, + "ewc_loss": 0.0029144287109375, + "ewc_loss_diag": 2.637505531311035e-06, + "ewc_loss_parallel": 2.652406692504883e-06, + "grad_norm": 4.6906538009643555, + "learning_rate": 3.8745231030097495e-07, + "loss": 0.3949, + "mean_token_accuracy": 0.8681819438934326, + "num_tokens": 35461490.0, + "step": 915 + }, + { + "epoch": 0.11652461518890726, + "ewc_loss": 0.002899169921875, + "ewc_loss_diag": 2.637505531311035e-06, + "ewc_loss_parallel": 2.637505531311035e-06, + "grad_norm": 4.7488579750061035, + "learning_rate": 3.8787621873675285e-07, + "loss": 0.393, + "mean_token_accuracy": 0.8672688007354736, + "num_tokens": 35499572.0, + "step": 916 + }, + { + "epoch": 0.11665182546749778, + "ewc_loss": 0.002899169921875, + "ewc_loss_diag": 2.6673078536987305e-06, + "ewc_loss_parallel": 2.6226043701171875e-06, + "grad_norm": 4.737330436706543, + "learning_rate": 3.883001271725307e-07, + "loss": 0.3897, + "mean_token_accuracy": 0.8704386353492737, + "num_tokens": 35539042.0, + "step": 917 + }, + { + "epoch": 0.11677903574608828, + "ewc_loss": 0.0029144287109375, + "ewc_loss_diag": 2.637505531311035e-06, + "ewc_loss_parallel": 2.652406692504883e-06, + "grad_norm": 4.713362216949463, + "learning_rate": 3.887240356083086e-07, + "loss": 0.3718, + "mean_token_accuracy": 0.8719099760055542, + "num_tokens": 35578882.0, + "step": 918 + }, + { + "epoch": 0.11690624602467879, + "ewc_loss": 0.0029296875, + "ewc_loss_diag": 2.6673078536987305e-06, + "ewc_loss_parallel": 2.652406692504883e-06, + "grad_norm": 5.236861705780029, + "learning_rate": 3.8914794404408644e-07, + "loss": 0.4252, + "mean_token_accuracy": 0.8570573329925537, + "num_tokens": 35607798.0, + "step": 919 + }, + { + "epoch": 0.1170334563032693, + "ewc_loss": 0.00299072265625, + "ewc_loss_diag": 2.6673078536987305e-06, + "ewc_loss_parallel": 2.726912498474121e-06, + "grad_norm": 5.1195573806762695, + "learning_rate": 3.8957185247986434e-07, + "loss": 0.4446, + "mean_token_accuracy": 0.8512372374534607, + "num_tokens": 35641434.0, + "step": 920 + }, + { + "epoch": 0.11716066658185982, + "ewc_loss": 0.002960205078125, + "ewc_loss_diag": 2.6673078536987305e-06, + "ewc_loss_parallel": 2.682209014892578e-06, + "grad_norm": 4.731520175933838, + "learning_rate": 3.899957609156422e-07, + "loss": 0.3988, + "mean_token_accuracy": 0.8680195808410645, + "num_tokens": 35682383.0, + "step": 921 + }, + { + "epoch": 0.11728787686045032, + "ewc_loss": 0.002899169921875, + "ewc_loss_diag": 2.6673078536987305e-06, + "ewc_loss_parallel": 2.6226043701171875e-06, + "grad_norm": 4.748124122619629, + "learning_rate": 3.904196693514201e-07, + "loss": 0.4351, + "mean_token_accuracy": 0.8559337854385376, + "num_tokens": 35727552.0, + "step": 922 + }, + { + "epoch": 0.11741508713904084, + "ewc_loss": 0.0028839111328125, + "ewc_loss_diag": 2.652406692504883e-06, + "ewc_loss_parallel": 2.6226043701171875e-06, + "grad_norm": 4.937248706817627, + "learning_rate": 3.9084357778719793e-07, + "loss": 0.4853, + "mean_token_accuracy": 0.8410834074020386, + "num_tokens": 35766798.0, + "step": 923 + }, + { + "epoch": 0.11754229741763135, + "ewc_loss": 0.002960205078125, + "ewc_loss_diag": 2.6673078536987305e-06, + "ewc_loss_parallel": 2.682209014892578e-06, + "grad_norm": 4.975002765655518, + "learning_rate": 3.9126748622297583e-07, + "loss": 0.4409, + "mean_token_accuracy": 0.8534599542617798, + "num_tokens": 35803377.0, + "step": 924 + }, + { + "epoch": 0.11766950769622185, + "ewc_loss": 0.002960205078125, + "ewc_loss_diag": 2.682209014892578e-06, + "ewc_loss_parallel": 2.682209014892578e-06, + "grad_norm": 4.885330677032471, + "learning_rate": 3.916913946587537e-07, + "loss": 0.4317, + "mean_token_accuracy": 0.8546437621116638, + "num_tokens": 35839352.0, + "step": 925 + }, + { + "epoch": 0.11779671797481236, + "ewc_loss": 0.0029754638671875, + "ewc_loss_diag": 2.682209014892578e-06, + "ewc_loss_parallel": 2.6971101760864258e-06, + "grad_norm": 4.813832759857178, + "learning_rate": 3.921153030945316e-07, + "loss": 0.4541, + "mean_token_accuracy": 0.8489948511123657, + "num_tokens": 35878944.0, + "step": 926 + }, + { + "epoch": 0.11792392825340288, + "ewc_loss": 0.002960205078125, + "ewc_loss_diag": 2.682209014892578e-06, + "ewc_loss_parallel": 2.682209014892578e-06, + "grad_norm": 4.918592929840088, + "learning_rate": 3.925392115303094e-07, + "loss": 0.4048, + "mean_token_accuracy": 0.8685582280158997, + "num_tokens": 35913928.0, + "step": 927 + }, + { + "epoch": 0.11805113853199338, + "ewc_loss": 0.002960205078125, + "ewc_loss_diag": 2.6673078536987305e-06, + "ewc_loss_parallel": 2.6971101760864258e-06, + "grad_norm": 4.9196624755859375, + "learning_rate": 3.929631199660873e-07, + "loss": 0.4096, + "mean_token_accuracy": 0.8584597706794739, + "num_tokens": 35950695.0, + "step": 928 + }, + { + "epoch": 0.1181783488105839, + "ewc_loss": 0.0029754638671875, + "ewc_loss_diag": 2.682209014892578e-06, + "ewc_loss_parallel": 2.6971101760864258e-06, + "grad_norm": 4.888612270355225, + "learning_rate": 3.9338702840186517e-07, + "loss": 0.4104, + "mean_token_accuracy": 0.8601822257041931, + "num_tokens": 35988512.0, + "step": 929 + }, + { + "epoch": 0.11830555908917441, + "ewc_loss": 0.0029296875, + "ewc_loss_diag": 2.6673078536987305e-06, + "ewc_loss_parallel": 2.6673078536987305e-06, + "grad_norm": 4.730544567108154, + "learning_rate": 3.9381093683764307e-07, + "loss": 0.4148, + "mean_token_accuracy": 0.8616093397140503, + "num_tokens": 36031097.0, + "step": 930 + }, + { + "epoch": 0.11843276936776491, + "ewc_loss": 0.002960205078125, + "ewc_loss_diag": 2.6673078536987305e-06, + "ewc_loss_parallel": 2.6971101760864258e-06, + "grad_norm": 4.868631839752197, + "learning_rate": 3.942348452734209e-07, + "loss": 0.4598, + "mean_token_accuracy": 0.8480260968208313, + "num_tokens": 36074671.0, + "step": 931 + }, + { + "epoch": 0.11855997964635542, + "ewc_loss": 0.002960205078125, + "ewc_loss_diag": 2.6673078536987305e-06, + "ewc_loss_parallel": 2.6971101760864258e-06, + "grad_norm": 4.794281482696533, + "learning_rate": 3.946587537091988e-07, + "loss": 0.4224, + "mean_token_accuracy": 0.8590171337127686, + "num_tokens": 36115190.0, + "step": 932 + }, + { + "epoch": 0.11868718992494594, + "ewc_loss": 0.00299072265625, + "ewc_loss_diag": 2.6673078536987305e-06, + "ewc_loss_parallel": 2.7120113372802734e-06, + "grad_norm": 4.806031703948975, + "learning_rate": 3.9508266214497666e-07, + "loss": 0.4194, + "mean_token_accuracy": 0.8619186878204346, + "num_tokens": 36158951.0, + "step": 933 + }, + { + "epoch": 0.11881440020353645, + "ewc_loss": 0.00299072265625, + "ewc_loss_diag": 2.682209014892578e-06, + "ewc_loss_parallel": 2.7120113372802734e-06, + "grad_norm": 4.746286869049072, + "learning_rate": 3.9550657058075456e-07, + "loss": 0.3831, + "mean_token_accuracy": 0.8708727359771729, + "num_tokens": 36199686.0, + "step": 934 + }, + { + "epoch": 0.11894161048212695, + "ewc_loss": 0.0029449462890625, + "ewc_loss_diag": 2.682209014892578e-06, + "ewc_loss_parallel": 2.6673078536987305e-06, + "grad_norm": 4.8666510581970215, + "learning_rate": 3.959304790165324e-07, + "loss": 0.4219, + "mean_token_accuracy": 0.8589863777160645, + "num_tokens": 36237222.0, + "step": 935 + }, + { + "epoch": 0.11906882076071747, + "ewc_loss": 0.0029754638671875, + "ewc_loss_diag": 2.682209014892578e-06, + "ewc_loss_parallel": 2.6971101760864258e-06, + "grad_norm": 4.974552631378174, + "learning_rate": 3.9635438745231025e-07, + "loss": 0.4112, + "mean_token_accuracy": 0.8645843267440796, + "num_tokens": 36271844.0, + "step": 936 + }, + { + "epoch": 0.11919603103930798, + "ewc_loss": 0.0030059814453125, + "ewc_loss_diag": 2.6971101760864258e-06, + "ewc_loss_parallel": 2.726912498474121e-06, + "grad_norm": 4.849393844604492, + "learning_rate": 3.9677829588808815e-07, + "loss": 0.4438, + "mean_token_accuracy": 0.852041482925415, + "num_tokens": 36314991.0, + "step": 937 + }, + { + "epoch": 0.11932324131789848, + "ewc_loss": 0.00299072265625, + "ewc_loss_diag": 2.6971101760864258e-06, + "ewc_loss_parallel": 2.7120113372802734e-06, + "grad_norm": 4.876819133758545, + "learning_rate": 3.9720220432386605e-07, + "loss": 0.4258, + "mean_token_accuracy": 0.857342004776001, + "num_tokens": 36350357.0, + "step": 938 + }, + { + "epoch": 0.119450451596489, + "ewc_loss": 0.00299072265625, + "ewc_loss_diag": 2.6971101760864258e-06, + "ewc_loss_parallel": 2.7120113372802734e-06, + "grad_norm": 4.931324481964111, + "learning_rate": 3.976261127596439e-07, + "loss": 0.3962, + "mean_token_accuracy": 0.8682145476341248, + "num_tokens": 36385489.0, + "step": 939 + }, + { + "epoch": 0.11957766187507951, + "ewc_loss": 0.0029754638671875, + "ewc_loss_diag": 2.7120113372802734e-06, + "ewc_loss_parallel": 2.6971101760864258e-06, + "grad_norm": 4.884304046630859, + "learning_rate": 3.9805002119542174e-07, + "loss": 0.3826, + "mean_token_accuracy": 0.8719781041145325, + "num_tokens": 36420205.0, + "step": 940 + }, + { + "epoch": 0.11970487215367001, + "ewc_loss": 0.0029754638671875, + "ewc_loss_diag": 2.6971101760864258e-06, + "ewc_loss_parallel": 2.6971101760864258e-06, + "grad_norm": 4.896519660949707, + "learning_rate": 3.9847392963119964e-07, + "loss": 0.4541, + "mean_token_accuracy": 0.8526049852371216, + "num_tokens": 36460283.0, + "step": 941 + }, + { + "epoch": 0.11983208243226053, + "ewc_loss": 0.00299072265625, + "ewc_loss_diag": 2.6971101760864258e-06, + "ewc_loss_parallel": 2.7120113372802734e-06, + "grad_norm": 4.8144450187683105, + "learning_rate": 3.9889783806697754e-07, + "loss": 0.4034, + "mean_token_accuracy": 0.8676955699920654, + "num_tokens": 36498827.0, + "step": 942 + }, + { + "epoch": 0.11995929271085104, + "ewc_loss": 0.00299072265625, + "ewc_loss_diag": 2.726912498474121e-06, + "ewc_loss_parallel": 2.7120113372802734e-06, + "grad_norm": 4.909412860870361, + "learning_rate": 3.993217465027554e-07, + "loss": 0.4621, + "mean_token_accuracy": 0.8508119583129883, + "num_tokens": 36541098.0, + "step": 943 + }, + { + "epoch": 0.12008650298944154, + "ewc_loss": 0.00299072265625, + "ewc_loss_diag": 2.7120113372802734e-06, + "ewc_loss_parallel": 2.7120113372802734e-06, + "grad_norm": 4.884511470794678, + "learning_rate": 3.9974565493853323e-07, + "loss": 0.3597, + "mean_token_accuracy": 0.8787095546722412, + "num_tokens": 36576303.0, + "step": 944 + }, + { + "epoch": 0.12021371326803205, + "ewc_loss": 0.0030059814453125, + "ewc_loss_diag": 2.7120113372802734e-06, + "ewc_loss_parallel": 2.726912498474121e-06, + "grad_norm": 4.966195583343506, + "learning_rate": 4.0016956337431113e-07, + "loss": 0.4247, + "mean_token_accuracy": 0.8573004007339478, + "num_tokens": 36612798.0, + "step": 945 + }, + { + "epoch": 0.12034092354662257, + "ewc_loss": 0.0030059814453125, + "ewc_loss_diag": 2.726912498474121e-06, + "ewc_loss_parallel": 2.726912498474121e-06, + "grad_norm": 5.07334566116333, + "learning_rate": 4.0059347181008903e-07, + "loss": 0.4252, + "mean_token_accuracy": 0.8572396039962769, + "num_tokens": 36646259.0, + "step": 946 + }, + { + "epoch": 0.12046813382521308, + "ewc_loss": 0.0030364990234375, + "ewc_loss_diag": 2.7120113372802734e-06, + "ewc_loss_parallel": 2.7567148208618164e-06, + "grad_norm": 4.751588821411133, + "learning_rate": 4.010173802458669e-07, + "loss": 0.3829, + "mean_token_accuracy": 0.8726072907447815, + "num_tokens": 36687562.0, + "step": 947 + }, + { + "epoch": 0.12059534410380358, + "ewc_loss": 0.00299072265625, + "ewc_loss_diag": 2.7120113372802734e-06, + "ewc_loss_parallel": 2.7120113372802734e-06, + "grad_norm": 4.9437384605407715, + "learning_rate": 4.014412886816447e-07, + "loss": 0.4909, + "mean_token_accuracy": 0.8405265212059021, + "num_tokens": 36728244.0, + "step": 948 + }, + { + "epoch": 0.1207225543823941, + "ewc_loss": 0.0030059814453125, + "ewc_loss_diag": 2.726912498474121e-06, + "ewc_loss_parallel": 2.726912498474121e-06, + "grad_norm": 4.902918815612793, + "learning_rate": 4.018651971174226e-07, + "loss": 0.4231, + "mean_token_accuracy": 0.8596211671829224, + "num_tokens": 36765848.0, + "step": 949 + }, + { + "epoch": 0.12084976466098461, + "ewc_loss": 0.003021240234375, + "ewc_loss_diag": 2.7567148208618164e-06, + "ewc_loss_parallel": 2.7418136596679688e-06, + "grad_norm": 4.7851881980896, + "learning_rate": 4.022891055532005e-07, + "loss": 0.4373, + "mean_token_accuracy": 0.855077862739563, + "num_tokens": 36810647.0, + "step": 950 + }, + { + "epoch": 0.12097697493957511, + "ewc_loss": 0.0030059814453125, + "ewc_loss_diag": 2.726912498474121e-06, + "ewc_loss_parallel": 2.726912498474121e-06, + "grad_norm": 4.944195747375488, + "learning_rate": 4.0271301398897837e-07, + "loss": 0.4673, + "mean_token_accuracy": 0.8490435481071472, + "num_tokens": 36853214.0, + "step": 951 + }, + { + "epoch": 0.12110418521816563, + "ewc_loss": 0.0030364990234375, + "ewc_loss_diag": 2.726912498474121e-06, + "ewc_loss_parallel": 2.7567148208618164e-06, + "grad_norm": 5.071811199188232, + "learning_rate": 4.031369224247562e-07, + "loss": 0.4203, + "mean_token_accuracy": 0.8570266366004944, + "num_tokens": 36887281.0, + "step": 952 + }, + { + "epoch": 0.12123139549675614, + "ewc_loss": 0.0030670166015625, + "ewc_loss_diag": 2.7418136596679688e-06, + "ewc_loss_parallel": 2.7865171432495117e-06, + "grad_norm": 4.849329471588135, + "learning_rate": 4.035608308605341e-07, + "loss": 0.4374, + "mean_token_accuracy": 0.8522735238075256, + "num_tokens": 36928363.0, + "step": 953 + }, + { + "epoch": 0.12135860577534664, + "ewc_loss": 0.0030364990234375, + "ewc_loss_diag": 2.7418136596679688e-06, + "ewc_loss_parallel": 2.7567148208618164e-06, + "grad_norm": 4.812028408050537, + "learning_rate": 4.03984739296312e-07, + "loss": 0.4222, + "mean_token_accuracy": 0.8617567420005798, + "num_tokens": 36971760.0, + "step": 954 + }, + { + "epoch": 0.12148581605393716, + "ewc_loss": 0.0030364990234375, + "ewc_loss_diag": 2.7418136596679688e-06, + "ewc_loss_parallel": 2.7567148208618164e-06, + "grad_norm": 4.817921161651611, + "learning_rate": 4.044086477320898e-07, + "loss": 0.4283, + "mean_token_accuracy": 0.8569847345352173, + "num_tokens": 37012874.0, + "step": 955 + }, + { + "epoch": 0.12161302633252767, + "ewc_loss": 0.0030059814453125, + "ewc_loss_diag": 2.7418136596679688e-06, + "ewc_loss_parallel": 2.726912498474121e-06, + "grad_norm": 5.060200214385986, + "learning_rate": 4.048325561678677e-07, + "loss": 0.4496, + "mean_token_accuracy": 0.8512131571769714, + "num_tokens": 37048695.0, + "step": 956 + }, + { + "epoch": 0.12174023661111817, + "ewc_loss": 0.003082275390625, + "ewc_loss_diag": 2.7567148208618164e-06, + "ewc_loss_parallel": 2.8014183044433594e-06, + "grad_norm": 4.9315290451049805, + "learning_rate": 4.052564646036456e-07, + "loss": 0.4371, + "mean_token_accuracy": 0.8563134074211121, + "num_tokens": 37090698.0, + "step": 957 + }, + { + "epoch": 0.12186744688970869, + "ewc_loss": 0.0030517578125, + "ewc_loss_diag": 2.7418136596679688e-06, + "ewc_loss_parallel": 2.771615982055664e-06, + "grad_norm": 4.969611644744873, + "learning_rate": 4.056803730394235e-07, + "loss": 0.4478, + "mean_token_accuracy": 0.8507177829742432, + "num_tokens": 37127323.0, + "step": 958 + }, + { + "epoch": 0.1219946571682992, + "ewc_loss": 0.003082275390625, + "ewc_loss_diag": 2.7418136596679688e-06, + "ewc_loss_parallel": 2.8014183044433594e-06, + "grad_norm": 5.038717746734619, + "learning_rate": 4.061042814752013e-07, + "loss": 0.4333, + "mean_token_accuracy": 0.8569022417068481, + "num_tokens": 37162703.0, + "step": 959 + }, + { + "epoch": 0.12212186744688971, + "ewc_loss": 0.0030975341796875, + "ewc_loss_diag": 2.7418136596679688e-06, + "ewc_loss_parallel": 2.816319465637207e-06, + "grad_norm": 5.159143447875977, + "learning_rate": 4.065281899109792e-07, + "loss": 0.4273, + "mean_token_accuracy": 0.8542484641075134, + "num_tokens": 37195858.0, + "step": 960 + }, + { + "epoch": 0.12224907772548022, + "ewc_loss": 0.0031280517578125, + "ewc_loss_diag": 2.7567148208618164e-06, + "ewc_loss_parallel": 2.8461217880249023e-06, + "grad_norm": 4.821719169616699, + "learning_rate": 4.069520983467571e-07, + "loss": 0.3824, + "mean_token_accuracy": 0.8741862773895264, + "num_tokens": 37235698.0, + "step": 961 + }, + { + "epoch": 0.12237628800407073, + "ewc_loss": 0.0030517578125, + "ewc_loss_diag": 2.771615982055664e-06, + "ewc_loss_parallel": 2.771615982055664e-06, + "grad_norm": 4.972660541534424, + "learning_rate": 4.07376006782535e-07, + "loss": 0.4129, + "mean_token_accuracy": 0.8600438833236694, + "num_tokens": 37270685.0, + "step": 962 + }, + { + "epoch": 0.12250349828266124, + "ewc_loss": 0.0031280517578125, + "ewc_loss_diag": 2.771615982055664e-06, + "ewc_loss_parallel": 2.8461217880249023e-06, + "grad_norm": 5.005914211273193, + "learning_rate": 4.077999152183128e-07, + "loss": 0.3777, + "mean_token_accuracy": 0.8716192841529846, + "num_tokens": 37304667.0, + "step": 963 + }, + { + "epoch": 0.12263070856125174, + "ewc_loss": 0.00311279296875, + "ewc_loss_diag": 2.7865171432495117e-06, + "ewc_loss_parallel": 2.8312206268310547e-06, + "grad_norm": 4.728315830230713, + "learning_rate": 4.082238236540907e-07, + "loss": 0.4241, + "mean_token_accuracy": 0.8609521389007568, + "num_tokens": 37351417.0, + "step": 964 + }, + { + "epoch": 0.12275791883984226, + "ewc_loss": 0.0030670166015625, + "ewc_loss_diag": 2.7865171432495117e-06, + "ewc_loss_parallel": 2.7865171432495117e-06, + "grad_norm": 4.808023929595947, + "learning_rate": 4.086477320898686e-07, + "loss": 0.4033, + "mean_token_accuracy": 0.864007830619812, + "num_tokens": 37392391.0, + "step": 965 + }, + { + "epoch": 0.12288512911843277, + "ewc_loss": 0.003082275390625, + "ewc_loss_diag": 2.7865171432495117e-06, + "ewc_loss_parallel": 2.8014183044433594e-06, + "grad_norm": 5.046517372131348, + "learning_rate": 4.090716405256465e-07, + "loss": 0.4512, + "mean_token_accuracy": 0.8489745855331421, + "num_tokens": 37430629.0, + "step": 966 + }, + { + "epoch": 0.12301233939702327, + "ewc_loss": 0.003143310546875, + "ewc_loss_diag": 2.7865171432495117e-06, + "ewc_loss_parallel": 2.86102294921875e-06, + "grad_norm": 4.944164276123047, + "learning_rate": 4.094955489614243e-07, + "loss": 0.393, + "mean_token_accuracy": 0.8666704297065735, + "num_tokens": 37465686.0, + "step": 967 + }, + { + "epoch": 0.12313954967561379, + "ewc_loss": 0.0031280517578125, + "ewc_loss_diag": 2.7865171432495117e-06, + "ewc_loss_parallel": 2.8461217880249023e-06, + "grad_norm": 4.80442476272583, + "learning_rate": 4.099194573972022e-07, + "loss": 0.4074, + "mean_token_accuracy": 0.8620932102203369, + "num_tokens": 37507997.0, + "step": 968 + }, + { + "epoch": 0.1232667599542043, + "ewc_loss": 0.003082275390625, + "ewc_loss_diag": 2.7865171432495117e-06, + "ewc_loss_parallel": 2.8014183044433594e-06, + "grad_norm": 4.887735366821289, + "learning_rate": 4.1034336583298007e-07, + "loss": 0.4531, + "mean_token_accuracy": 0.850210428237915, + "num_tokens": 37550342.0, + "step": 969 + }, + { + "epoch": 0.1233939702327948, + "ewc_loss": 0.0030975341796875, + "ewc_loss_diag": 2.7865171432495117e-06, + "ewc_loss_parallel": 2.816319465637207e-06, + "grad_norm": 4.866978168487549, + "learning_rate": 4.1076727426875797e-07, + "loss": 0.4399, + "mean_token_accuracy": 0.8533838987350464, + "num_tokens": 37590849.0, + "step": 970 + }, + { + "epoch": 0.12352118051138532, + "ewc_loss": 0.00311279296875, + "ewc_loss_diag": 2.8014183044433594e-06, + "ewc_loss_parallel": 2.8312206268310547e-06, + "grad_norm": 5.0349345207214355, + "learning_rate": 4.1119118270453577e-07, + "loss": 0.4936, + "mean_token_accuracy": 0.8377258777618408, + "num_tokens": 37630506.0, + "step": 971 + }, + { + "epoch": 0.12364839078997583, + "ewc_loss": 0.003143310546875, + "ewc_loss_diag": 2.816319465637207e-06, + "ewc_loss_parallel": 2.8461217880249023e-06, + "grad_norm": 4.915167808532715, + "learning_rate": 4.1161509114031366e-07, + "loss": 0.386, + "mean_token_accuracy": 0.8735295534133911, + "num_tokens": 37669308.0, + "step": 972 + }, + { + "epoch": 0.12377560106856635, + "ewc_loss": 0.003143310546875, + "ewc_loss_diag": 2.816319465637207e-06, + "ewc_loss_parallel": 2.8461217880249023e-06, + "grad_norm": 4.991498947143555, + "learning_rate": 4.1203899957609156e-07, + "loss": 0.4441, + "mean_token_accuracy": 0.8536571264266968, + "num_tokens": 37706783.0, + "step": 973 + }, + { + "epoch": 0.12390281134715685, + "ewc_loss": 0.003143310546875, + "ewc_loss_diag": 2.816319465637207e-06, + "ewc_loss_parallel": 2.86102294921875e-06, + "grad_norm": 5.078942775726318, + "learning_rate": 4.124629080118694e-07, + "loss": 0.4057, + "mean_token_accuracy": 0.8633601665496826, + "num_tokens": 37740243.0, + "step": 974 + }, + { + "epoch": 0.12403002162574736, + "ewc_loss": 0.003143310546875, + "ewc_loss_diag": 2.816319465637207e-06, + "ewc_loss_parallel": 2.8759241104125977e-06, + "grad_norm": 4.806004524230957, + "learning_rate": 4.1288681644764726e-07, + "loss": 0.4316, + "mean_token_accuracy": 0.8567493557929993, + "num_tokens": 37780383.0, + "step": 975 + }, + { + "epoch": 0.12415723190433788, + "ewc_loss": 0.00311279296875, + "ewc_loss_diag": 2.8312206268310547e-06, + "ewc_loss_parallel": 2.816319465637207e-06, + "grad_norm": 5.0406107902526855, + "learning_rate": 4.1331072488342515e-07, + "loss": 0.4517, + "mean_token_accuracy": 0.8529160022735596, + "num_tokens": 37815899.0, + "step": 976 + }, + { + "epoch": 0.12428444218292838, + "ewc_loss": 0.003173828125, + "ewc_loss_diag": 2.8312206268310547e-06, + "ewc_loss_parallel": 2.8908252716064453e-06, + "grad_norm": 4.889780521392822, + "learning_rate": 4.1373463331920305e-07, + "loss": 0.3742, + "mean_token_accuracy": 0.8746917247772217, + "num_tokens": 37852701.0, + "step": 977 + }, + { + "epoch": 0.12441165246151889, + "ewc_loss": 0.003143310546875, + "ewc_loss_diag": 2.8312206268310547e-06, + "ewc_loss_parallel": 2.8759241104125977e-06, + "grad_norm": 4.845447540283203, + "learning_rate": 4.141585417549809e-07, + "loss": 0.4338, + "mean_token_accuracy": 0.8552359342575073, + "num_tokens": 37891227.0, + "step": 978 + }, + { + "epoch": 0.1245388627401094, + "ewc_loss": 0.003143310546875, + "ewc_loss_diag": 2.8312206268310547e-06, + "ewc_loss_parallel": 2.8461217880249023e-06, + "grad_norm": 5.0144219398498535, + "learning_rate": 4.1458245019075875e-07, + "loss": 0.4042, + "mean_token_accuracy": 0.8619095087051392, + "num_tokens": 37924932.0, + "step": 979 + }, + { + "epoch": 0.1246660730186999, + "ewc_loss": 0.003173828125, + "ewc_loss_diag": 2.8312206268310547e-06, + "ewc_loss_parallel": 2.8908252716064453e-06, + "grad_norm": 4.937223434448242, + "learning_rate": 4.1500635862653664e-07, + "loss": 0.4369, + "mean_token_accuracy": 0.8548128604888916, + "num_tokens": 37964465.0, + "step": 980 + }, + { + "epoch": 0.12479328329729042, + "ewc_loss": 0.003173828125, + "ewc_loss_diag": 2.8312206268310547e-06, + "ewc_loss_parallel": 2.8908252716064453e-06, + "grad_norm": 4.8258137702941895, + "learning_rate": 4.1543026706231454e-07, + "loss": 0.3782, + "mean_token_accuracy": 0.8727924823760986, + "num_tokens": 38002543.0, + "step": 981 + }, + { + "epoch": 0.12492049357588093, + "ewc_loss": 0.003143310546875, + "ewc_loss_diag": 2.8312206268310547e-06, + "ewc_loss_parallel": 2.8759241104125977e-06, + "grad_norm": 4.976845741271973, + "learning_rate": 4.158541754980924e-07, + "loss": 0.3842, + "mean_token_accuracy": 0.8725259304046631, + "num_tokens": 38039411.0, + "step": 982 + }, + { + "epoch": 0.12504770385447145, + "ewc_loss": 0.0031585693359375, + "ewc_loss_diag": 2.86102294921875e-06, + "ewc_loss_parallel": 2.8759241104125977e-06, + "grad_norm": 5.082279682159424, + "learning_rate": 4.1627808393387024e-07, + "loss": 0.4396, + "mean_token_accuracy": 0.851865291595459, + "num_tokens": 38076663.0, + "step": 983 + }, + { + "epoch": 0.12517491413306195, + "ewc_loss": 0.0031890869140625, + "ewc_loss_diag": 2.8759241104125977e-06, + "ewc_loss_parallel": 2.905726432800293e-06, + "grad_norm": 5.103217601776123, + "learning_rate": 4.1670199236964813e-07, + "loss": 0.4337, + "mean_token_accuracy": 0.8545790910720825, + "num_tokens": 38113078.0, + "step": 984 + }, + { + "epoch": 0.12530212441165245, + "ewc_loss": 0.003173828125, + "ewc_loss_diag": 2.8759241104125977e-06, + "ewc_loss_parallel": 2.8908252716064453e-06, + "grad_norm": 4.915838718414307, + "learning_rate": 4.1712590080542603e-07, + "loss": 0.3942, + "mean_token_accuracy": 0.8706387877464294, + "num_tokens": 38150399.0, + "step": 985 + }, + { + "epoch": 0.12542933469024298, + "ewc_loss": 0.003173828125, + "ewc_loss_diag": 2.86102294921875e-06, + "ewc_loss_parallel": 2.8908252716064453e-06, + "grad_norm": 4.936354637145996, + "learning_rate": 4.175498092412039e-07, + "loss": 0.4321, + "mean_token_accuracy": 0.8555960655212402, + "num_tokens": 38192285.0, + "step": 986 + }, + { + "epoch": 0.12555654496883348, + "ewc_loss": 0.0031585693359375, + "ewc_loss_diag": 2.8461217880249023e-06, + "ewc_loss_parallel": 2.8759241104125977e-06, + "grad_norm": 4.975035667419434, + "learning_rate": 4.179737176769817e-07, + "loss": 0.3965, + "mean_token_accuracy": 0.8684802055358887, + "num_tokens": 38232135.0, + "step": 987 + }, + { + "epoch": 0.12568375524742398, + "ewc_loss": 0.003173828125, + "ewc_loss_diag": 2.86102294921875e-06, + "ewc_loss_parallel": 2.8908252716064453e-06, + "grad_norm": 4.896719455718994, + "learning_rate": 4.183976261127596e-07, + "loss": 0.4025, + "mean_token_accuracy": 0.8650274276733398, + "num_tokens": 38271246.0, + "step": 988 + }, + { + "epoch": 0.1258109655260145, + "ewc_loss": 0.0031585693359375, + "ewc_loss_diag": 2.86102294921875e-06, + "ewc_loss_parallel": 2.8759241104125977e-06, + "grad_norm": 5.063786506652832, + "learning_rate": 4.1882153454853747e-07, + "loss": 0.4719, + "mean_token_accuracy": 0.8420473337173462, + "num_tokens": 38308362.0, + "step": 989 + }, + { + "epoch": 0.125938175804605, + "ewc_loss": 0.00323486328125, + "ewc_loss_diag": 2.86102294921875e-06, + "ewc_loss_parallel": 2.950429916381836e-06, + "grad_norm": 4.942147731781006, + "learning_rate": 4.1924544298431537e-07, + "loss": 0.4073, + "mean_token_accuracy": 0.8661536574363708, + "num_tokens": 38347834.0, + "step": 990 + }, + { + "epoch": 0.12606538608319554, + "ewc_loss": 0.003173828125, + "ewc_loss_diag": 2.8759241104125977e-06, + "ewc_loss_parallel": 2.8908252716064453e-06, + "grad_norm": 4.867681980133057, + "learning_rate": 4.196693514200932e-07, + "loss": 0.409, + "mean_token_accuracy": 0.8651875853538513, + "num_tokens": 38392746.0, + "step": 991 + }, + { + "epoch": 0.12619259636178604, + "ewc_loss": 0.0031585693359375, + "ewc_loss_diag": 2.8461217880249023e-06, + "ewc_loss_parallel": 2.8759241104125977e-06, + "grad_norm": 5.016178131103516, + "learning_rate": 4.200932598558711e-07, + "loss": 0.4284, + "mean_token_accuracy": 0.8549163341522217, + "num_tokens": 38433722.0, + "step": 992 + }, + { + "epoch": 0.12631980664037654, + "ewc_loss": 0.0032196044921875, + "ewc_loss_diag": 2.8759241104125977e-06, + "ewc_loss_parallel": 2.9355287551879883e-06, + "grad_norm": 4.85698938369751, + "learning_rate": 4.2051716829164896e-07, + "loss": 0.4309, + "mean_token_accuracy": 0.8567707538604736, + "num_tokens": 38479177.0, + "step": 993 + }, + { + "epoch": 0.12644701691896706, + "ewc_loss": 0.003173828125, + "ewc_loss_diag": 2.8759241104125977e-06, + "ewc_loss_parallel": 2.8908252716064453e-06, + "grad_norm": 5.030031681060791, + "learning_rate": 4.2094107672742686e-07, + "loss": 0.4775, + "mean_token_accuracy": 0.8466475605964661, + "num_tokens": 38517530.0, + "step": 994 + }, + { + "epoch": 0.12657422719755757, + "ewc_loss": 0.00323486328125, + "ewc_loss_diag": 2.8759241104125977e-06, + "ewc_loss_parallel": 2.950429916381836e-06, + "grad_norm": 5.0609307289123535, + "learning_rate": 4.2136498516320476e-07, + "loss": 0.3812, + "mean_token_accuracy": 0.8729106187820435, + "num_tokens": 38551279.0, + "step": 995 + }, + { + "epoch": 0.12670143747614807, + "ewc_loss": 0.0032196044921875, + "ewc_loss_diag": 2.8759241104125977e-06, + "ewc_loss_parallel": 2.9355287551879883e-06, + "grad_norm": 4.90908145904541, + "learning_rate": 4.217888935989826e-07, + "loss": 0.439, + "mean_token_accuracy": 0.8539690375328064, + "num_tokens": 38596311.0, + "step": 996 + }, + { + "epoch": 0.1268286477547386, + "ewc_loss": 0.003173828125, + "ewc_loss_diag": 2.8908252716064453e-06, + "ewc_loss_parallel": 2.8908252716064453e-06, + "grad_norm": 4.9702277183532715, + "learning_rate": 4.2221280203476045e-07, + "loss": 0.4284, + "mean_token_accuracy": 0.8586747050285339, + "num_tokens": 38639289.0, + "step": 997 + }, + { + "epoch": 0.1269558580333291, + "ewc_loss": 0.0032196044921875, + "ewc_loss_diag": 2.8908252716064453e-06, + "ewc_loss_parallel": 2.9355287551879883e-06, + "grad_norm": 4.913308143615723, + "learning_rate": 4.2263671047053835e-07, + "loss": 0.4598, + "mean_token_accuracy": 0.8469053506851196, + "num_tokens": 38684874.0, + "step": 998 + }, + { + "epoch": 0.1270830683119196, + "ewc_loss": 0.003173828125, + "ewc_loss_diag": 2.905726432800293e-06, + "ewc_loss_parallel": 2.8908252716064453e-06, + "grad_norm": 4.870070457458496, + "learning_rate": 4.2306061890631625e-07, + "loss": 0.4419, + "mean_token_accuracy": 0.8535719513893127, + "num_tokens": 38727957.0, + "step": 999 + }, + { + "epoch": 0.12721027859051012, + "ewc_loss": 0.003173828125, + "ewc_loss_diag": 2.9206275939941406e-06, + "ewc_loss_parallel": 2.8908252716064453e-06, + "grad_norm": 5.162808895111084, + "learning_rate": 4.234845273420941e-07, + "loss": 0.5172, + "mean_token_accuracy": 0.8358967304229736, + "num_tokens": 38765947.0, + "step": 1000 + }, + { + "epoch": 0.12733748886910062, + "ewc_loss": 0.0033111572265625, + "ewc_loss_diag": 2.9206275939941406e-06, + "ewc_loss_parallel": 3.0249357223510742e-06, + "grad_norm": 4.91011905670166, + "learning_rate": 4.2390843577787194e-07, + "loss": 0.4419, + "mean_token_accuracy": 0.8531085252761841, + "num_tokens": 38810556.0, + "step": 1001 + }, + { + "epoch": 0.12746469914769112, + "ewc_loss": 0.0031585693359375, + "ewc_loss_diag": 2.9206275939941406e-06, + "ewc_loss_parallel": 2.8759241104125977e-06, + "grad_norm": 4.950930595397949, + "learning_rate": 4.2433234421364984e-07, + "loss": 0.418, + "mean_token_accuracy": 0.8608202934265137, + "num_tokens": 38848211.0, + "step": 1002 + }, + { + "epoch": 0.12759190942628165, + "ewc_loss": 0.00323486328125, + "ewc_loss_diag": 2.9206275939941406e-06, + "ewc_loss_parallel": 2.950429916381836e-06, + "grad_norm": 5.040631294250488, + "learning_rate": 4.2475625264942774e-07, + "loss": 0.4107, + "mean_token_accuracy": 0.8653558492660522, + "num_tokens": 38890672.0, + "step": 1003 + }, + { + "epoch": 0.12771911970487215, + "ewc_loss": 0.0032958984375, + "ewc_loss_diag": 2.950429916381836e-06, + "ewc_loss_parallel": 3.0100345611572266e-06, + "grad_norm": 5.336030960083008, + "learning_rate": 4.251801610852056e-07, + "loss": 0.4414, + "mean_token_accuracy": 0.8546609878540039, + "num_tokens": 38922146.0, + "step": 1004 + }, + { + "epoch": 0.12784632998346265, + "ewc_loss": 0.0033416748046875, + "ewc_loss_diag": 2.950429916381836e-06, + "ewc_loss_parallel": 3.0547380447387695e-06, + "grad_norm": 5.0127434730529785, + "learning_rate": 4.2560406952098343e-07, + "loss": 0.4691, + "mean_token_accuracy": 0.8474348783493042, + "num_tokens": 38963408.0, + "step": 1005 + }, + { + "epoch": 0.12797354026205318, + "ewc_loss": 0.003173828125, + "ewc_loss_diag": 2.950429916381836e-06, + "ewc_loss_parallel": 2.8908252716064453e-06, + "grad_norm": 4.9860029220581055, + "learning_rate": 4.2602797795676133e-07, + "loss": 0.4798, + "mean_token_accuracy": 0.8433747887611389, + "num_tokens": 39002881.0, + "step": 1006 + }, + { + "epoch": 0.12810075054064368, + "ewc_loss": 0.0032196044921875, + "ewc_loss_diag": 2.950429916381836e-06, + "ewc_loss_parallel": 2.9355287551879883e-06, + "grad_norm": 5.012963771820068, + "learning_rate": 4.2645188639253923e-07, + "loss": 0.4015, + "mean_token_accuracy": 0.8641239404678345, + "num_tokens": 39039465.0, + "step": 1007 + }, + { + "epoch": 0.12822796081923418, + "ewc_loss": 0.0032501220703125, + "ewc_loss_diag": 2.9653310775756836e-06, + "ewc_loss_parallel": 2.9653310775756836e-06, + "grad_norm": 4.94976806640625, + "learning_rate": 4.26875794828317e-07, + "loss": 0.4173, + "mean_token_accuracy": 0.8604660034179688, + "num_tokens": 39082154.0, + "step": 1008 + }, + { + "epoch": 0.1283551710978247, + "ewc_loss": 0.003204345703125, + "ewc_loss_diag": 2.9653310775756836e-06, + "ewc_loss_parallel": 2.9206275939941406e-06, + "grad_norm": 4.892577171325684, + "learning_rate": 4.272997032640949e-07, + "loss": 0.4446, + "mean_token_accuracy": 0.8518443703651428, + "num_tokens": 39128017.0, + "step": 1009 + }, + { + "epoch": 0.1284823813764152, + "ewc_loss": 0.0032196044921875, + "ewc_loss_diag": 2.9653310775756836e-06, + "ewc_loss_parallel": 2.9355287551879883e-06, + "grad_norm": 4.940908908843994, + "learning_rate": 4.277236116998728e-07, + "loss": 0.4159, + "mean_token_accuracy": 0.8624741435050964, + "num_tokens": 39169983.0, + "step": 1010 + }, + { + "epoch": 0.1286095916550057, + "ewc_loss": 0.003265380859375, + "ewc_loss_diag": 2.9653310775756836e-06, + "ewc_loss_parallel": 2.9802322387695312e-06, + "grad_norm": 5.034049987792969, + "learning_rate": 4.281475201356507e-07, + "loss": 0.4168, + "mean_token_accuracy": 0.8599390983581543, + "num_tokens": 39209086.0, + "step": 1011 + }, + { + "epoch": 0.12873680193359624, + "ewc_loss": 0.0033111572265625, + "ewc_loss_diag": 2.9653310775756836e-06, + "ewc_loss_parallel": 3.0249357223510742e-06, + "grad_norm": 4.982763767242432, + "learning_rate": 4.285714285714285e-07, + "loss": 0.4447, + "mean_token_accuracy": 0.8519983291625977, + "num_tokens": 39249076.0, + "step": 1012 + }, + { + "epoch": 0.12886401221218674, + "ewc_loss": 0.0032501220703125, + "ewc_loss_diag": 2.9653310775756836e-06, + "ewc_loss_parallel": 2.9653310775756836e-06, + "grad_norm": 4.977801322937012, + "learning_rate": 4.289953370072064e-07, + "loss": 0.3992, + "mean_token_accuracy": 0.8667275905609131, + "num_tokens": 39288419.0, + "step": 1013 + }, + { + "epoch": 0.12899122249077727, + "ewc_loss": 0.0032806396484375, + "ewc_loss_diag": 2.9653310775756836e-06, + "ewc_loss_parallel": 2.995133399963379e-06, + "grad_norm": 5.039183139801025, + "learning_rate": 4.294192454429843e-07, + "loss": 0.4441, + "mean_token_accuracy": 0.852714478969574, + "num_tokens": 39328038.0, + "step": 1014 + }, + { + "epoch": 0.12911843276936777, + "ewc_loss": 0.0032958984375, + "ewc_loss_diag": 2.9802322387695312e-06, + "ewc_loss_parallel": 2.995133399963379e-06, + "grad_norm": 4.822667121887207, + "learning_rate": 4.298431538787622e-07, + "loss": 0.3468, + "mean_token_accuracy": 0.8841382265090942, + "num_tokens": 39369595.0, + "step": 1015 + }, + { + "epoch": 0.12924564304795827, + "ewc_loss": 0.0032196044921875, + "ewc_loss_diag": 2.9653310775756836e-06, + "ewc_loss_parallel": 2.9355287551879883e-06, + "grad_norm": 5.162162780761719, + "learning_rate": 4.3026706231454e-07, + "loss": 0.4787, + "mean_token_accuracy": 0.841334342956543, + "num_tokens": 39403508.0, + "step": 1016 + }, + { + "epoch": 0.1293728533265488, + "ewc_loss": 0.003387451171875, + "ewc_loss_diag": 2.9653310775756836e-06, + "ewc_loss_parallel": 3.0994415283203125e-06, + "grad_norm": 5.092677593231201, + "learning_rate": 4.306909707503179e-07, + "loss": 0.4173, + "mean_token_accuracy": 0.8629018068313599, + "num_tokens": 39442423.0, + "step": 1017 + }, + { + "epoch": 0.1295000636051393, + "ewc_loss": 0.0032958984375, + "ewc_loss_diag": 2.9802322387695312e-06, + "ewc_loss_parallel": 2.995133399963379e-06, + "grad_norm": 5.247766971588135, + "learning_rate": 4.311148791860958e-07, + "loss": 0.4198, + "mean_token_accuracy": 0.8577518463134766, + "num_tokens": 39474127.0, + "step": 1018 + }, + { + "epoch": 0.1296272738837298, + "ewc_loss": 0.003387451171875, + "ewc_loss_diag": 3.0100345611572266e-06, + "ewc_loss_parallel": 3.084540367126465e-06, + "grad_norm": 5.112980842590332, + "learning_rate": 4.315387876218737e-07, + "loss": 0.4528, + "mean_token_accuracy": 0.8540636897087097, + "num_tokens": 39510121.0, + "step": 1019 + }, + { + "epoch": 0.12975448416232033, + "ewc_loss": 0.0033111572265625, + "ewc_loss_diag": 3.0100345611572266e-06, + "ewc_loss_parallel": 3.0100345611572266e-06, + "grad_norm": 5.064657688140869, + "learning_rate": 4.319626960576515e-07, + "loss": 0.4516, + "mean_token_accuracy": 0.8475277423858643, + "num_tokens": 39546265.0, + "step": 1020 + }, + { + "epoch": 0.12988169444091083, + "ewc_loss": 0.003265380859375, + "ewc_loss_diag": 3.0100345611572266e-06, + "ewc_loss_parallel": 2.9653310775756836e-06, + "grad_norm": 4.932962417602539, + "learning_rate": 4.323866044934294e-07, + "loss": 0.4027, + "mean_token_accuracy": 0.8621789813041687, + "num_tokens": 39586796.0, + "step": 1021 + }, + { + "epoch": 0.13000890471950133, + "ewc_loss": 0.0033111572265625, + "ewc_loss_diag": 3.0100345611572266e-06, + "ewc_loss_parallel": 3.0100345611572266e-06, + "grad_norm": 4.9178338050842285, + "learning_rate": 4.328105129292073e-07, + "loss": 0.4228, + "mean_token_accuracy": 0.860019862651825, + "num_tokens": 39629783.0, + "step": 1022 + }, + { + "epoch": 0.13013611499809186, + "ewc_loss": 0.0033111572265625, + "ewc_loss_diag": 3.0249357223510742e-06, + "ewc_loss_parallel": 3.0100345611572266e-06, + "grad_norm": 4.97699499130249, + "learning_rate": 4.332344213649852e-07, + "loss": 0.4043, + "mean_token_accuracy": 0.865280270576477, + "num_tokens": 39667409.0, + "step": 1023 + }, + { + "epoch": 0.13026332527668236, + "ewc_loss": 0.0033416748046875, + "ewc_loss_diag": 3.0100345611572266e-06, + "ewc_loss_parallel": 3.039836883544922e-06, + "grad_norm": 5.050768852233887, + "learning_rate": 4.33658329800763e-07, + "loss": 0.4094, + "mean_token_accuracy": 0.865427553653717, + "num_tokens": 39705469.0, + "step": 1024 + }, + { + "epoch": 0.13039053555527286, + "ewc_loss": 0.00335693359375, + "ewc_loss_diag": 2.995133399963379e-06, + "ewc_loss_parallel": 3.0547380447387695e-06, + "grad_norm": 4.883321285247803, + "learning_rate": 4.340822382365409e-07, + "loss": 0.4009, + "mean_token_accuracy": 0.8647882342338562, + "num_tokens": 39749171.0, + "step": 1025 + }, + { + "epoch": 0.13051774583386339, + "ewc_loss": 0.0032958984375, + "ewc_loss_diag": 3.0249357223510742e-06, + "ewc_loss_parallel": 2.995133399963379e-06, + "grad_norm": 4.875891208648682, + "learning_rate": 4.345061466723188e-07, + "loss": 0.3807, + "mean_token_accuracy": 0.8723495006561279, + "num_tokens": 39793756.0, + "step": 1026 + }, + { + "epoch": 0.13064495611245389, + "ewc_loss": 0.00335693359375, + "ewc_loss_diag": 3.0249357223510742e-06, + "ewc_loss_parallel": 3.0547380447387695e-06, + "grad_norm": 5.084554195404053, + "learning_rate": 4.3493005510809663e-07, + "loss": 0.4365, + "mean_token_accuracy": 0.8562809824943542, + "num_tokens": 39830904.0, + "step": 1027 + }, + { + "epoch": 0.1307721663910444, + "ewc_loss": 0.00341796875, + "ewc_loss_diag": 3.0249357223510742e-06, + "ewc_loss_parallel": 3.11434268951416e-06, + "grad_norm": 5.0832061767578125, + "learning_rate": 4.353539635438745e-07, + "loss": 0.4171, + "mean_token_accuracy": 0.861782968044281, + "num_tokens": 39867905.0, + "step": 1028 + }, + { + "epoch": 0.13089937666963491, + "ewc_loss": 0.00335693359375, + "ewc_loss_diag": 3.0249357223510742e-06, + "ewc_loss_parallel": 3.0547380447387695e-06, + "grad_norm": 5.175429344177246, + "learning_rate": 4.357778719796524e-07, + "loss": 0.4605, + "mean_token_accuracy": 0.8445680141448975, + "num_tokens": 39903025.0, + "step": 1029 + }, + { + "epoch": 0.13102658694822542, + "ewc_loss": 0.00335693359375, + "ewc_loss_diag": 3.0249357223510742e-06, + "ewc_loss_parallel": 3.0547380447387695e-06, + "grad_norm": 4.952946662902832, + "learning_rate": 4.362017804154303e-07, + "loss": 0.3737, + "mean_token_accuracy": 0.8753006458282471, + "num_tokens": 39942076.0, + "step": 1030 + }, + { + "epoch": 0.13115379722681592, + "ewc_loss": 0.0032958984375, + "ewc_loss_diag": 3.0249357223510742e-06, + "ewc_loss_parallel": 2.995133399963379e-06, + "grad_norm": 5.115160942077637, + "learning_rate": 4.366256888512081e-07, + "loss": 0.4021, + "mean_token_accuracy": 0.8663979768753052, + "num_tokens": 39977052.0, + "step": 1031 + }, + { + "epoch": 0.13128100750540644, + "ewc_loss": 0.0034027099609375, + "ewc_loss_diag": 3.0249357223510742e-06, + "ewc_loss_parallel": 3.0994415283203125e-06, + "grad_norm": 5.023537635803223, + "learning_rate": 4.3704959728698597e-07, + "loss": 0.4156, + "mean_token_accuracy": 0.8604608774185181, + "num_tokens": 40017501.0, + "step": 1032 + }, + { + "epoch": 0.13140821778399694, + "ewc_loss": 0.003326416015625, + "ewc_loss_diag": 3.0249357223510742e-06, + "ewc_loss_parallel": 3.0249357223510742e-06, + "grad_norm": 5.155886173248291, + "learning_rate": 4.3747350572276386e-07, + "loss": 0.4158, + "mean_token_accuracy": 0.8601016998291016, + "num_tokens": 40051584.0, + "step": 1033 + }, + { + "epoch": 0.13153542806258745, + "ewc_loss": 0.003387451171875, + "ewc_loss_diag": 3.039836883544922e-06, + "ewc_loss_parallel": 3.084540367126465e-06, + "grad_norm": 5.039703369140625, + "learning_rate": 4.3789741415854176e-07, + "loss": 0.4285, + "mean_token_accuracy": 0.8594670295715332, + "num_tokens": 40089046.0, + "step": 1034 + }, + { + "epoch": 0.13166263834117797, + "ewc_loss": 0.003326416015625, + "ewc_loss_diag": 3.039836883544922e-06, + "ewc_loss_parallel": 3.0249357223510742e-06, + "grad_norm": 5.091907024383545, + "learning_rate": 4.383213225943196e-07, + "loss": 0.4324, + "mean_token_accuracy": 0.8569586277008057, + "num_tokens": 40127096.0, + "step": 1035 + }, + { + "epoch": 0.13178984861976847, + "ewc_loss": 0.0033721923828125, + "ewc_loss_diag": 3.0547380447387695e-06, + "ewc_loss_parallel": 3.069639205932617e-06, + "grad_norm": 4.982068061828613, + "learning_rate": 4.3874523103009746e-07, + "loss": 0.3619, + "mean_token_accuracy": 0.8776050806045532, + "num_tokens": 40164250.0, + "step": 1036 + }, + { + "epoch": 0.13191705889835897, + "ewc_loss": 0.0033111572265625, + "ewc_loss_diag": 3.039836883544922e-06, + "ewc_loss_parallel": 3.0100345611572266e-06, + "grad_norm": 5.086309909820557, + "learning_rate": 4.3916913946587536e-07, + "loss": 0.3861, + "mean_token_accuracy": 0.8676320314407349, + "num_tokens": 40199276.0, + "step": 1037 + }, + { + "epoch": 0.1320442691769495, + "ewc_loss": 0.00341796875, + "ewc_loss_diag": 3.039836883544922e-06, + "ewc_loss_parallel": 3.11434268951416e-06, + "grad_norm": 5.055110454559326, + "learning_rate": 4.3959304790165325e-07, + "loss": 0.3898, + "mean_token_accuracy": 0.8666895627975464, + "num_tokens": 40236060.0, + "step": 1038 + }, + { + "epoch": 0.13217147945554, + "ewc_loss": 0.0034027099609375, + "ewc_loss_diag": 3.0547380447387695e-06, + "ewc_loss_parallel": 3.0994415283203125e-06, + "grad_norm": 4.991979122161865, + "learning_rate": 4.400169563374311e-07, + "loss": 0.4215, + "mean_token_accuracy": 0.8606210947036743, + "num_tokens": 40277753.0, + "step": 1039 + }, + { + "epoch": 0.13229868973413053, + "ewc_loss": 0.00335693359375, + "ewc_loss_diag": 3.0547380447387695e-06, + "ewc_loss_parallel": 3.0547380447387695e-06, + "grad_norm": 5.045316696166992, + "learning_rate": 4.4044086477320895e-07, + "loss": 0.4644, + "mean_token_accuracy": 0.8470879197120667, + "num_tokens": 40318612.0, + "step": 1040 + }, + { + "epoch": 0.13242590001272103, + "ewc_loss": 0.00341796875, + "ewc_loss_diag": 3.084540367126465e-06, + "ewc_loss_parallel": 3.11434268951416e-06, + "grad_norm": 5.127716541290283, + "learning_rate": 4.4086477320898685e-07, + "loss": 0.4082, + "mean_token_accuracy": 0.8612367510795593, + "num_tokens": 40353026.0, + "step": 1041 + }, + { + "epoch": 0.13255311029131153, + "ewc_loss": 0.00341796875, + "ewc_loss_diag": 3.084540367126465e-06, + "ewc_loss_parallel": 3.11434268951416e-06, + "grad_norm": 4.951667785644531, + "learning_rate": 4.4128868164476474e-07, + "loss": 0.4079, + "mean_token_accuracy": 0.8654024600982666, + "num_tokens": 40395222.0, + "step": 1042 + }, + { + "epoch": 0.13268032056990206, + "ewc_loss": 0.003387451171875, + "ewc_loss_diag": 3.084540367126465e-06, + "ewc_loss_parallel": 3.084540367126465e-06, + "grad_norm": 4.968715667724609, + "learning_rate": 4.417125900805426e-07, + "loss": 0.4019, + "mean_token_accuracy": 0.8668543696403503, + "num_tokens": 40438032.0, + "step": 1043 + }, + { + "epoch": 0.13280753084849256, + "ewc_loss": 0.0034332275390625, + "ewc_loss_diag": 3.0994415283203125e-06, + "ewc_loss_parallel": 3.129243850708008e-06, + "grad_norm": 5.016808032989502, + "learning_rate": 4.4213649851632044e-07, + "loss": 0.4395, + "mean_token_accuracy": 0.8563352227210999, + "num_tokens": 40480042.0, + "step": 1044 + }, + { + "epoch": 0.13293474112708306, + "ewc_loss": 0.003387451171875, + "ewc_loss_diag": 3.0994415283203125e-06, + "ewc_loss_parallel": 3.084540367126465e-06, + "grad_norm": 5.094612121582031, + "learning_rate": 4.4256040695209834e-07, + "loss": 0.443, + "mean_token_accuracy": 0.8542381525039673, + "num_tokens": 40518043.0, + "step": 1045 + }, + { + "epoch": 0.1330619514056736, + "ewc_loss": 0.00341796875, + "ewc_loss_diag": 3.0994415283203125e-06, + "ewc_loss_parallel": 3.11434268951416e-06, + "grad_norm": 5.254817008972168, + "learning_rate": 4.429843153878762e-07, + "loss": 0.4494, + "mean_token_accuracy": 0.8496657609939575, + "num_tokens": 40551713.0, + "step": 1046 + }, + { + "epoch": 0.1331891616842641, + "ewc_loss": 0.0034637451171875, + "ewc_loss_diag": 3.0994415283203125e-06, + "ewc_loss_parallel": 3.159046173095703e-06, + "grad_norm": 5.0895586013793945, + "learning_rate": 4.434082238236541e-07, + "loss": 0.4704, + "mean_token_accuracy": 0.8428679704666138, + "num_tokens": 40594557.0, + "step": 1047 + }, + { + "epoch": 0.1333163719628546, + "ewc_loss": 0.0034332275390625, + "ewc_loss_diag": 3.11434268951416e-06, + "ewc_loss_parallel": 3.129243850708008e-06, + "grad_norm": 4.9979143142700195, + "learning_rate": 4.4383213225943193e-07, + "loss": 0.4465, + "mean_token_accuracy": 0.8537435531616211, + "num_tokens": 40635930.0, + "step": 1048 + }, + { + "epoch": 0.13344358224144512, + "ewc_loss": 0.00341796875, + "ewc_loss_diag": 3.11434268951416e-06, + "ewc_loss_parallel": 3.11434268951416e-06, + "grad_norm": 5.1204752922058105, + "learning_rate": 4.442560406952098e-07, + "loss": 0.4519, + "mean_token_accuracy": 0.8483610153198242, + "num_tokens": 40674232.0, + "step": 1049 + }, + { + "epoch": 0.13357079252003562, + "ewc_loss": 0.0034637451171875, + "ewc_loss_diag": 3.11434268951416e-06, + "ewc_loss_parallel": 3.159046173095703e-06, + "grad_norm": 4.995112895965576, + "learning_rate": 4.4467994913098767e-07, + "loss": 0.4038, + "mean_token_accuracy": 0.8662184476852417, + "num_tokens": 40716750.0, + "step": 1050 + }, + { + "epoch": 0.13369800279862612, + "ewc_loss": 0.003387451171875, + "ewc_loss_diag": 3.11434268951416e-06, + "ewc_loss_parallel": 3.084540367126465e-06, + "grad_norm": 5.050462245941162, + "learning_rate": 4.4510385756676557e-07, + "loss": 0.4406, + "mean_token_accuracy": 0.8505664467811584, + "num_tokens": 40755911.0, + "step": 1051 + }, + { + "epoch": 0.13382521307721665, + "ewc_loss": 0.0034637451171875, + "ewc_loss_diag": 3.11434268951416e-06, + "ewc_loss_parallel": 3.159046173095703e-06, + "grad_norm": 4.984658718109131, + "learning_rate": 4.455277660025434e-07, + "loss": 0.4067, + "mean_token_accuracy": 0.8646903038024902, + "num_tokens": 40802643.0, + "step": 1052 + }, + { + "epoch": 0.13395242335580715, + "ewc_loss": 0.003448486328125, + "ewc_loss_diag": 3.129243850708008e-06, + "ewc_loss_parallel": 3.129243850708008e-06, + "grad_norm": 5.0897650718688965, + "learning_rate": 4.459516744383213e-07, + "loss": 0.4206, + "mean_token_accuracy": 0.8566430807113647, + "num_tokens": 40839376.0, + "step": 1053 + }, + { + "epoch": 0.13407963363439765, + "ewc_loss": 0.00347900390625, + "ewc_loss_diag": 3.129243850708008e-06, + "ewc_loss_parallel": 3.1739473342895508e-06, + "grad_norm": 5.139175891876221, + "learning_rate": 4.4637558287409916e-07, + "loss": 0.4427, + "mean_token_accuracy": 0.8541302680969238, + "num_tokens": 40880840.0, + "step": 1054 + }, + { + "epoch": 0.13420684391298818, + "ewc_loss": 0.00347900390625, + "ewc_loss_diag": 3.129243850708008e-06, + "ewc_loss_parallel": 3.159046173095703e-06, + "grad_norm": 5.140931129455566, + "learning_rate": 4.4679949130987706e-07, + "loss": 0.4527, + "mean_token_accuracy": 0.8563952445983887, + "num_tokens": 40920515.0, + "step": 1055 + }, + { + "epoch": 0.13433405419157868, + "ewc_loss": 0.003448486328125, + "ewc_loss_diag": 3.129243850708008e-06, + "ewc_loss_parallel": 3.129243850708008e-06, + "grad_norm": 5.124119758605957, + "learning_rate": 4.472233997456549e-07, + "loss": 0.4444, + "mean_token_accuracy": 0.8510420322418213, + "num_tokens": 40958482.0, + "step": 1056 + }, + { + "epoch": 0.13446126447016918, + "ewc_loss": 0.00347900390625, + "ewc_loss_diag": 3.129243850708008e-06, + "ewc_loss_parallel": 3.159046173095703e-06, + "grad_norm": 5.03070592880249, + "learning_rate": 4.476473081814328e-07, + "loss": 0.4283, + "mean_token_accuracy": 0.8581830263137817, + "num_tokens": 41000410.0, + "step": 1057 + }, + { + "epoch": 0.1345884747487597, + "ewc_loss": 0.0034332275390625, + "ewc_loss_diag": 3.1441450119018555e-06, + "ewc_loss_parallel": 3.11434268951416e-06, + "grad_norm": 5.046499252319336, + "learning_rate": 4.4807121661721065e-07, + "loss": 0.3949, + "mean_token_accuracy": 0.8681327104568481, + "num_tokens": 41041070.0, + "step": 1058 + }, + { + "epoch": 0.1347156850273502, + "ewc_loss": 0.00347900390625, + "ewc_loss_diag": 3.1441450119018555e-06, + "ewc_loss_parallel": 3.159046173095703e-06, + "grad_norm": 5.134335994720459, + "learning_rate": 4.4849512505298855e-07, + "loss": 0.4321, + "mean_token_accuracy": 0.8569574356079102, + "num_tokens": 41081667.0, + "step": 1059 + }, + { + "epoch": 0.1348428953059407, + "ewc_loss": 0.0034942626953125, + "ewc_loss_diag": 3.159046173095703e-06, + "ewc_loss_parallel": 3.1739473342895508e-06, + "grad_norm": 5.0759172439575195, + "learning_rate": 4.489190334887664e-07, + "loss": 0.469, + "mean_token_accuracy": 0.8512225151062012, + "num_tokens": 41122685.0, + "step": 1060 + }, + { + "epoch": 0.13497010558453124, + "ewc_loss": 0.0034637451171875, + "ewc_loss_diag": 3.159046173095703e-06, + "ewc_loss_parallel": 3.1441450119018555e-06, + "grad_norm": 5.1149444580078125, + "learning_rate": 4.493429419245443e-07, + "loss": 0.4288, + "mean_token_accuracy": 0.8584376573562622, + "num_tokens": 41159244.0, + "step": 1061 + }, + { + "epoch": 0.13509731586312174, + "ewc_loss": 0.00347900390625, + "ewc_loss_diag": 3.159046173095703e-06, + "ewc_loss_parallel": 3.159046173095703e-06, + "grad_norm": 5.109495639801025, + "learning_rate": 4.4976685036032214e-07, + "loss": 0.3799, + "mean_token_accuracy": 0.8732377290725708, + "num_tokens": 41195242.0, + "step": 1062 + }, + { + "epoch": 0.13522452614171224, + "ewc_loss": 0.0034942626953125, + "ewc_loss_diag": 3.159046173095703e-06, + "ewc_loss_parallel": 3.1739473342895508e-06, + "grad_norm": 5.237494468688965, + "learning_rate": 4.5019075879610004e-07, + "loss": 0.4472, + "mean_token_accuracy": 0.8482094407081604, + "num_tokens": 41230428.0, + "step": 1063 + }, + { + "epoch": 0.13535173642030277, + "ewc_loss": 0.0035247802734375, + "ewc_loss_diag": 3.159046173095703e-06, + "ewc_loss_parallel": 3.203749656677246e-06, + "grad_norm": 5.08057165145874, + "learning_rate": 4.506146672318779e-07, + "loss": 0.4279, + "mean_token_accuracy": 0.8591287136077881, + "num_tokens": 41270081.0, + "step": 1064 + }, + { + "epoch": 0.13547894669889327, + "ewc_loss": 0.0034637451171875, + "ewc_loss_diag": 3.159046173095703e-06, + "ewc_loss_parallel": 3.1441450119018555e-06, + "grad_norm": 5.097653388977051, + "learning_rate": 4.5103857566765573e-07, + "loss": 0.4801, + "mean_token_accuracy": 0.8446781635284424, + "num_tokens": 41311416.0, + "step": 1065 + }, + { + "epoch": 0.1356061569774838, + "ewc_loss": 0.0034942626953125, + "ewc_loss_diag": 3.1739473342895508e-06, + "ewc_loss_parallel": 3.1739473342895508e-06, + "grad_norm": 5.0689167976379395, + "learning_rate": 4.5146248410343363e-07, + "loss": 0.3801, + "mean_token_accuracy": 0.8718488216400146, + "num_tokens": 41348076.0, + "step": 1066 + }, + { + "epoch": 0.1357333672560743, + "ewc_loss": 0.0034942626953125, + "ewc_loss_diag": 3.1739473342895508e-06, + "ewc_loss_parallel": 3.1739473342895508e-06, + "grad_norm": 5.057004451751709, + "learning_rate": 4.5188639253921153e-07, + "loss": 0.3999, + "mean_token_accuracy": 0.866659939289093, + "num_tokens": 41386087.0, + "step": 1067 + }, + { + "epoch": 0.1358605775346648, + "ewc_loss": 0.00347900390625, + "ewc_loss_diag": 3.1739473342895508e-06, + "ewc_loss_parallel": 3.159046173095703e-06, + "grad_norm": 5.024113655090332, + "learning_rate": 4.523103009749894e-07, + "loss": 0.3704, + "mean_token_accuracy": 0.8718778491020203, + "num_tokens": 41424252.0, + "step": 1068 + }, + { + "epoch": 0.13598778781325532, + "ewc_loss": 0.0034942626953125, + "ewc_loss_diag": 3.1739473342895508e-06, + "ewc_loss_parallel": 3.1739473342895508e-06, + "grad_norm": 4.958959579467773, + "learning_rate": 4.527342094107672e-07, + "loss": 0.4156, + "mean_token_accuracy": 0.8602876663208008, + "num_tokens": 41469475.0, + "step": 1069 + }, + { + "epoch": 0.13611499809184582, + "ewc_loss": 0.00347900390625, + "ewc_loss_diag": 3.1739473342895508e-06, + "ewc_loss_parallel": 3.159046173095703e-06, + "grad_norm": 5.105330467224121, + "learning_rate": 4.531581178465451e-07, + "loss": 0.4363, + "mean_token_accuracy": 0.8552747964859009, + "num_tokens": 41509873.0, + "step": 1070 + }, + { + "epoch": 0.13624220837043632, + "ewc_loss": 0.0035247802734375, + "ewc_loss_diag": 3.1739473342895508e-06, + "ewc_loss_parallel": 3.203749656677246e-06, + "grad_norm": 5.159853458404541, + "learning_rate": 4.53582026282323e-07, + "loss": 0.4044, + "mean_token_accuracy": 0.8657336235046387, + "num_tokens": 41550347.0, + "step": 1071 + }, + { + "epoch": 0.13636941864902685, + "ewc_loss": 0.0035247802734375, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.203749656677246e-06, + "grad_norm": 5.2844672203063965, + "learning_rate": 4.5400593471810087e-07, + "loss": 0.4864, + "mean_token_accuracy": 0.8375834226608276, + "num_tokens": 41589720.0, + "step": 1072 + }, + { + "epoch": 0.13649662892761735, + "ewc_loss": 0.0035400390625, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.2186508178710938e-06, + "grad_norm": 5.09052848815918, + "learning_rate": 4.544298431538787e-07, + "loss": 0.4138, + "mean_token_accuracy": 0.8603528738021851, + "num_tokens": 41630192.0, + "step": 1073 + }, + { + "epoch": 0.13662383920620785, + "ewc_loss": 0.003509521484375, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.1888484954833984e-06, + "grad_norm": 5.257213592529297, + "learning_rate": 4.548537515896566e-07, + "loss": 0.4573, + "mean_token_accuracy": 0.8474904298782349, + "num_tokens": 41664889.0, + "step": 1074 + }, + { + "epoch": 0.13675104948479838, + "ewc_loss": 0.0035247802734375, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.203749656677246e-06, + "grad_norm": 5.246909141540527, + "learning_rate": 4.552776600254345e-07, + "loss": 0.4464, + "mean_token_accuracy": 0.8510103225708008, + "num_tokens": 41700998.0, + "step": 1075 + }, + { + "epoch": 0.13687825976338888, + "ewc_loss": 0.003509521484375, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.1888484954833984e-06, + "grad_norm": 5.080288887023926, + "learning_rate": 4.5570156846121236e-07, + "loss": 0.395, + "mean_token_accuracy": 0.8687598705291748, + "num_tokens": 41740240.0, + "step": 1076 + }, + { + "epoch": 0.13700547004197938, + "ewc_loss": 0.003509521484375, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.1888484954833984e-06, + "grad_norm": 5.05112886428833, + "learning_rate": 4.561254768969902e-07, + "loss": 0.4004, + "mean_token_accuracy": 0.8640058040618896, + "num_tokens": 41779530.0, + "step": 1077 + }, + { + "epoch": 0.1371326803205699, + "ewc_loss": 0.0035247802734375, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.203749656677246e-06, + "grad_norm": 5.119545936584473, + "learning_rate": 4.565493853327681e-07, + "loss": 0.4034, + "mean_token_accuracy": 0.8640992641448975, + "num_tokens": 41817938.0, + "step": 1078 + }, + { + "epoch": 0.1372598905991604, + "ewc_loss": 0.0035247802734375, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.203749656677246e-06, + "grad_norm": 5.236123085021973, + "learning_rate": 4.56973293768546e-07, + "loss": 0.4424, + "mean_token_accuracy": 0.8573498725891113, + "num_tokens": 41855388.0, + "step": 1079 + }, + { + "epoch": 0.1373871008777509, + "ewc_loss": 0.0035552978515625, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.2335519790649414e-06, + "grad_norm": 5.190502166748047, + "learning_rate": 4.573972022043238e-07, + "loss": 0.477, + "mean_token_accuracy": 0.8410950303077698, + "num_tokens": 41897055.0, + "step": 1080 + }, + { + "epoch": 0.13751431115634144, + "ewc_loss": 0.0035552978515625, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.2335519790649414e-06, + "grad_norm": 5.010631084442139, + "learning_rate": 4.578211106401017e-07, + "loss": 0.439, + "mean_token_accuracy": 0.8551712036132812, + "num_tokens": 41940982.0, + "step": 1081 + }, + { + "epoch": 0.13764152143493194, + "ewc_loss": 0.0035247802734375, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.203749656677246e-06, + "grad_norm": 5.150810718536377, + "learning_rate": 4.582450190758796e-07, + "loss": 0.4584, + "mean_token_accuracy": 0.8483965396881104, + "num_tokens": 41982750.0, + "step": 1082 + }, + { + "epoch": 0.13776873171352244, + "ewc_loss": 0.0035858154296875, + "ewc_loss_diag": 3.203749656677246e-06, + "ewc_loss_parallel": 3.2633543014526367e-06, + "grad_norm": 5.114426136016846, + "learning_rate": 4.586689275116575e-07, + "loss": 0.3736, + "mean_token_accuracy": 0.8732712268829346, + "num_tokens": 42021563.0, + "step": 1083 + }, + { + "epoch": 0.13789594199211297, + "ewc_loss": 0.0035552978515625, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.2335519790649414e-06, + "grad_norm": 5.175192356109619, + "learning_rate": 4.590928359474353e-07, + "loss": 0.4721, + "mean_token_accuracy": 0.8447037935256958, + "num_tokens": 42058392.0, + "step": 1084 + }, + { + "epoch": 0.13802315227070347, + "ewc_loss": 0.003570556640625, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.248453140258789e-06, + "grad_norm": 5.107876300811768, + "learning_rate": 4.595167443832132e-07, + "loss": 0.4202, + "mean_token_accuracy": 0.860844612121582, + "num_tokens": 42099140.0, + "step": 1085 + }, + { + "epoch": 0.13815036254929397, + "ewc_loss": 0.003570556640625, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.248453140258789e-06, + "grad_norm": 5.053971767425537, + "learning_rate": 4.599406528189911e-07, + "loss": 0.3834, + "mean_token_accuracy": 0.8708220720291138, + "num_tokens": 42137413.0, + "step": 1086 + }, + { + "epoch": 0.1382775728278845, + "ewc_loss": 0.0035400390625, + "ewc_loss_diag": 3.1888484954833984e-06, + "ewc_loss_parallel": 3.2186508178710938e-06, + "grad_norm": 5.089802265167236, + "learning_rate": 4.60364561254769e-07, + "loss": 0.4049, + "mean_token_accuracy": 0.8618305921554565, + "num_tokens": 42174379.0, + "step": 1087 + }, + { + "epoch": 0.138404783106475, + "ewc_loss": 0.0035858154296875, + "ewc_loss_diag": 3.2186508178710938e-06, + "ewc_loss_parallel": 3.2633543014526367e-06, + "grad_norm": 5.15151834487915, + "learning_rate": 4.607884696905468e-07, + "loss": 0.417, + "mean_token_accuracy": 0.860389232635498, + "num_tokens": 42216069.0, + "step": 1088 + }, + { + "epoch": 0.1385319933850655, + "ewc_loss": 0.0035858154296875, + "ewc_loss_diag": 3.2186508178710938e-06, + "ewc_loss_parallel": 3.2633543014526367e-06, + "grad_norm": 5.073086261749268, + "learning_rate": 4.612123781263247e-07, + "loss": 0.4475, + "mean_token_accuracy": 0.8507682085037231, + "num_tokens": 42261895.0, + "step": 1089 + }, + { + "epoch": 0.13865920366365603, + "ewc_loss": 0.0035552978515625, + "ewc_loss_diag": 3.203749656677246e-06, + "ewc_loss_parallel": 3.2335519790649414e-06, + "grad_norm": 5.136262893676758, + "learning_rate": 4.616362865621026e-07, + "loss": 0.3511, + "mean_token_accuracy": 0.8795301914215088, + "num_tokens": 42298616.0, + "step": 1090 + }, + { + "epoch": 0.13878641394224653, + "ewc_loss": 0.0035858154296875, + "ewc_loss_diag": 3.2186508178710938e-06, + "ewc_loss_parallel": 3.2633543014526367e-06, + "grad_norm": 5.121711730957031, + "learning_rate": 4.620601949978805e-07, + "loss": 0.4257, + "mean_token_accuracy": 0.8594785928726196, + "num_tokens": 42337837.0, + "step": 1091 + }, + { + "epoch": 0.13891362422083706, + "ewc_loss": 0.0035400390625, + "ewc_loss_diag": 3.2186508178710938e-06, + "ewc_loss_parallel": 3.2186508178710938e-06, + "grad_norm": 5.245914459228516, + "learning_rate": 4.6248410343365827e-07, + "loss": 0.4873, + "mean_token_accuracy": 0.8389794826507568, + "num_tokens": 42375743.0, + "step": 1092 + }, + { + "epoch": 0.13904083449942756, + "ewc_loss": 0.00360107421875, + "ewc_loss_diag": 3.2186508178710938e-06, + "ewc_loss_parallel": 3.2782554626464844e-06, + "grad_norm": 5.1345672607421875, + "learning_rate": 4.6290801186943617e-07, + "loss": 0.4007, + "mean_token_accuracy": 0.8650549650192261, + "num_tokens": 42415056.0, + "step": 1093 + }, + { + "epoch": 0.13916804477801806, + "ewc_loss": 0.0035858154296875, + "ewc_loss_diag": 3.2186508178710938e-06, + "ewc_loss_parallel": 3.2633543014526367e-06, + "grad_norm": 5.268190383911133, + "learning_rate": 4.6333192030521407e-07, + "loss": 0.4143, + "mean_token_accuracy": 0.8632948994636536, + "num_tokens": 42450359.0, + "step": 1094 + }, + { + "epoch": 0.13929525505660859, + "ewc_loss": 0.00360107421875, + "ewc_loss_diag": 3.2186508178710938e-06, + "ewc_loss_parallel": 3.2782554626464844e-06, + "grad_norm": 5.092133045196533, + "learning_rate": 4.6375582874099196e-07, + "loss": 0.3931, + "mean_token_accuracy": 0.8669265508651733, + "num_tokens": 42492396.0, + "step": 1095 + }, + { + "epoch": 0.1394224653351991, + "ewc_loss": 0.003570556640625, + "ewc_loss_diag": 3.2186508178710938e-06, + "ewc_loss_parallel": 3.248453140258789e-06, + "grad_norm": 5.118654727935791, + "learning_rate": 4.6417973717676976e-07, + "loss": 0.3928, + "mean_token_accuracy": 0.8696615099906921, + "num_tokens": 42529963.0, + "step": 1096 + }, + { + "epoch": 0.1395496756137896, + "ewc_loss": 0.0036163330078125, + "ewc_loss_diag": 3.2335519790649414e-06, + "ewc_loss_parallel": 3.293156623840332e-06, + "grad_norm": 5.238533020019531, + "learning_rate": 4.6460364561254766e-07, + "loss": 0.4405, + "mean_token_accuracy": 0.8516154289245605, + "num_tokens": 42568708.0, + "step": 1097 + }, + { + "epoch": 0.13967688589238011, + "ewc_loss": 0.0036163330078125, + "ewc_loss_diag": 3.2335519790649414e-06, + "ewc_loss_parallel": 3.293156623840332e-06, + "grad_norm": 5.223948955535889, + "learning_rate": 4.6502755404832556e-07, + "loss": 0.3949, + "mean_token_accuracy": 0.8668553233146667, + "num_tokens": 42604493.0, + "step": 1098 + }, + { + "epoch": 0.13980409617097062, + "ewc_loss": 0.0035858154296875, + "ewc_loss_diag": 3.2335519790649414e-06, + "ewc_loss_parallel": 3.2633543014526367e-06, + "grad_norm": 5.359443187713623, + "learning_rate": 4.654514624841034e-07, + "loss": 0.4265, + "mean_token_accuracy": 0.8587630987167358, + "num_tokens": 42636873.0, + "step": 1099 + }, + { + "epoch": 0.13993130644956112, + "ewc_loss": 0.0036468505859375, + "ewc_loss_diag": 3.2335519790649414e-06, + "ewc_loss_parallel": 3.3229589462280273e-06, + "grad_norm": 5.238678455352783, + "learning_rate": 4.6587537091988125e-07, + "loss": 0.4239, + "mean_token_accuracy": 0.8602190017700195, + "num_tokens": 42672684.0, + "step": 1100 + }, + { + "epoch": 0.14005851672815164, + "ewc_loss": 0.0035858154296875, + "ewc_loss_diag": 3.248453140258789e-06, + "ewc_loss_parallel": 3.2633543014526367e-06, + "grad_norm": 5.142381191253662, + "learning_rate": 4.6629927935565915e-07, + "loss": 0.3985, + "mean_token_accuracy": 0.8664014935493469, + "num_tokens": 42709961.0, + "step": 1101 + }, + { + "epoch": 0.14018572700674214, + "ewc_loss": 0.00360107421875, + "ewc_loss_diag": 3.248453140258789e-06, + "ewc_loss_parallel": 3.2782554626464844e-06, + "grad_norm": 5.16873025894165, + "learning_rate": 4.6672318779143705e-07, + "loss": 0.4731, + "mean_token_accuracy": 0.8429285287857056, + "num_tokens": 42752698.0, + "step": 1102 + }, + { + "epoch": 0.14031293728533265, + "ewc_loss": 0.00360107421875, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.2782554626464844e-06, + "grad_norm": 5.366662502288818, + "learning_rate": 4.671470962272149e-07, + "loss": 0.4397, + "mean_token_accuracy": 0.8548961877822876, + "num_tokens": 42786156.0, + "step": 1103 + }, + { + "epoch": 0.14044014756392317, + "ewc_loss": 0.0036163330078125, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.293156623840332e-06, + "grad_norm": 5.146087169647217, + "learning_rate": 4.6757100466299274e-07, + "loss": 0.4347, + "mean_token_accuracy": 0.8561998009681702, + "num_tokens": 42828758.0, + "step": 1104 + }, + { + "epoch": 0.14056735784251367, + "ewc_loss": 0.00360107421875, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.2782554626464844e-06, + "grad_norm": 5.127279758453369, + "learning_rate": 4.6799491309877064e-07, + "loss": 0.4353, + "mean_token_accuracy": 0.8540701270103455, + "num_tokens": 42870862.0, + "step": 1105 + }, + { + "epoch": 0.14069456812110417, + "ewc_loss": 0.00360107421875, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.2782554626464844e-06, + "grad_norm": 5.2321906089782715, + "learning_rate": 4.6841882153454854e-07, + "loss": 0.3836, + "mean_token_accuracy": 0.8706246614456177, + "num_tokens": 42905543.0, + "step": 1106 + }, + { + "epoch": 0.1408217783996947, + "ewc_loss": 0.003631591796875, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.3080577850341797e-06, + "grad_norm": 5.201598167419434, + "learning_rate": 4.688427299703264e-07, + "loss": 0.3709, + "mean_token_accuracy": 0.8708723187446594, + "num_tokens": 42942908.0, + "step": 1107 + }, + { + "epoch": 0.1409489886782852, + "ewc_loss": 0.0036163330078125, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.293156623840332e-06, + "grad_norm": 5.152409076690674, + "learning_rate": 4.6926663840610423e-07, + "loss": 0.4361, + "mean_token_accuracy": 0.855163037776947, + "num_tokens": 42983040.0, + "step": 1108 + }, + { + "epoch": 0.1410761989568757, + "ewc_loss": 0.003631591796875, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.3080577850341797e-06, + "grad_norm": 5.13025426864624, + "learning_rate": 4.6969054684188213e-07, + "loss": 0.4406, + "mean_token_accuracy": 0.8499648571014404, + "num_tokens": 43026148.0, + "step": 1109 + }, + { + "epoch": 0.14120340923546623, + "ewc_loss": 0.00360107421875, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.2782554626464844e-06, + "grad_norm": 5.216026306152344, + "learning_rate": 4.7011445527766003e-07, + "loss": 0.4122, + "mean_token_accuracy": 0.8593683838844299, + "num_tokens": 43061815.0, + "step": 1110 + }, + { + "epoch": 0.14133061951405673, + "ewc_loss": 0.003631591796875, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.3080577850341797e-06, + "grad_norm": 5.26379919052124, + "learning_rate": 4.7053836371343787e-07, + "loss": 0.431, + "mean_token_accuracy": 0.8565501570701599, + "num_tokens": 43100740.0, + "step": 1111 + }, + { + "epoch": 0.14145782979264723, + "ewc_loss": 0.0036468505859375, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.3229589462280273e-06, + "grad_norm": 5.235691070556641, + "learning_rate": 4.709622721492157e-07, + "loss": 0.4056, + "mean_token_accuracy": 0.864560604095459, + "num_tokens": 43137317.0, + "step": 1112 + }, + { + "epoch": 0.14158504007123776, + "ewc_loss": 0.0036468505859375, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.3229589462280273e-06, + "grad_norm": 5.340349197387695, + "learning_rate": 4.713861805849936e-07, + "loss": 0.4192, + "mean_token_accuracy": 0.8614101409912109, + "num_tokens": 43171275.0, + "step": 1113 + }, + { + "epoch": 0.14171225034982826, + "ewc_loss": 0.0037078857421875, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.382563591003418e-06, + "grad_norm": 5.347201347351074, + "learning_rate": 4.718100890207715e-07, + "loss": 0.469, + "mean_token_accuracy": 0.8438287973403931, + "num_tokens": 43208162.0, + "step": 1114 + }, + { + "epoch": 0.1418394606284188, + "ewc_loss": 0.0036773681640625, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.3527612686157227e-06, + "grad_norm": 5.243827819824219, + "learning_rate": 4.7223399745654936e-07, + "loss": 0.4228, + "mean_token_accuracy": 0.8589329123497009, + "num_tokens": 43245840.0, + "step": 1115 + }, + { + "epoch": 0.1419666709070093, + "ewc_loss": 0.003692626953125, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.3676624298095703e-06, + "grad_norm": 5.174007892608643, + "learning_rate": 4.726579058923272e-07, + "loss": 0.4213, + "mean_token_accuracy": 0.8605127334594727, + "num_tokens": 43284000.0, + "step": 1116 + }, + { + "epoch": 0.1420938811855998, + "ewc_loss": 0.003662109375, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.337860107421875e-06, + "grad_norm": 5.255671501159668, + "learning_rate": 4.730818143281051e-07, + "loss": 0.4486, + "mean_token_accuracy": 0.8522462248802185, + "num_tokens": 43328028.0, + "step": 1117 + }, + { + "epoch": 0.14222109146419032, + "ewc_loss": 0.0036773681640625, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.3527612686157227e-06, + "grad_norm": 5.280327320098877, + "learning_rate": 4.7350572276388295e-07, + "loss": 0.3967, + "mean_token_accuracy": 0.8677268028259277, + "num_tokens": 43367157.0, + "step": 1118 + }, + { + "epoch": 0.14234830174278082, + "ewc_loss": 0.0037078857421875, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.382563591003418e-06, + "grad_norm": 5.243381500244141, + "learning_rate": 4.7392963119966085e-07, + "loss": 0.4162, + "mean_token_accuracy": 0.860743522644043, + "num_tokens": 43406154.0, + "step": 1119 + }, + { + "epoch": 0.14247551202137132, + "ewc_loss": 0.003662109375, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.337860107421875e-06, + "grad_norm": 5.088089466094971, + "learning_rate": 4.7435353963543875e-07, + "loss": 0.4144, + "mean_token_accuracy": 0.8656072616577148, + "num_tokens": 43449680.0, + "step": 1120 + }, + { + "epoch": 0.14260272229996185, + "ewc_loss": 0.003692626953125, + "ewc_loss_diag": 3.2633543014526367e-06, + "ewc_loss_parallel": 3.3676624298095703e-06, + "grad_norm": 5.342062473297119, + "learning_rate": 4.747774480712166e-07, + "loss": 0.4416, + "mean_token_accuracy": 0.8555808067321777, + "num_tokens": 43485602.0, + "step": 1121 + }, + { + "epoch": 0.14272993257855235, + "ewc_loss": 0.0037384033203125, + "ewc_loss_diag": 3.293156623840332e-06, + "ewc_loss_parallel": 3.3974647521972656e-06, + "grad_norm": 5.307434558868408, + "learning_rate": 4.7520135650699444e-07, + "loss": 0.4449, + "mean_token_accuracy": 0.8528866767883301, + "num_tokens": 43521852.0, + "step": 1122 + }, + { + "epoch": 0.14285714285714285, + "ewc_loss": 0.003753662109375, + "ewc_loss_diag": 3.293156623840332e-06, + "ewc_loss_parallel": 3.4123659133911133e-06, + "grad_norm": 5.263664245605469, + "learning_rate": 4.7562526494277234e-07, + "loss": 0.4175, + "mean_token_accuracy": 0.8612090945243835, + "num_tokens": 43557759.0, + "step": 1123 + }, + { + "epoch": 0.14298435313573338, + "ewc_loss": 0.003753662109375, + "ewc_loss_diag": 3.293156623840332e-06, + "ewc_loss_parallel": 3.4123659133911133e-06, + "grad_norm": 5.28249454498291, + "learning_rate": 4.7604917337855024e-07, + "loss": 0.3715, + "mean_token_accuracy": 0.8731274604797363, + "num_tokens": 43592836.0, + "step": 1124 + }, + { + "epoch": 0.14311156341432388, + "ewc_loss": 0.003753662109375, + "ewc_loss_diag": 3.293156623840332e-06, + "ewc_loss_parallel": 3.4123659133911133e-06, + "grad_norm": 5.337849140167236, + "learning_rate": 4.764730818143281e-07, + "loss": 0.4158, + "mean_token_accuracy": 0.8571622371673584, + "num_tokens": 43627466.0, + "step": 1125 + }, + { + "epoch": 0.14323877369291438, + "ewc_loss": 0.003753662109375, + "ewc_loss_diag": 3.293156623840332e-06, + "ewc_loss_parallel": 3.4123659133911133e-06, + "grad_norm": 5.171780586242676, + "learning_rate": 4.768969902501059e-07, + "loss": 0.459, + "mean_token_accuracy": 0.8458128571510315, + "num_tokens": 43670168.0, + "step": 1126 + }, + { + "epoch": 0.1433659839715049, + "ewc_loss": 0.00372314453125, + "ewc_loss_diag": 3.293156623840332e-06, + "ewc_loss_parallel": 3.382563591003418e-06, + "grad_norm": 5.145113468170166, + "learning_rate": 4.773208986858838e-07, + "loss": 0.4204, + "mean_token_accuracy": 0.8580578565597534, + "num_tokens": 43710819.0, + "step": 1127 + }, + { + "epoch": 0.1434931942500954, + "ewc_loss": 0.00372314453125, + "ewc_loss_diag": 3.3080577850341797e-06, + "ewc_loss_parallel": 3.382563591003418e-06, + "grad_norm": 5.211186408996582, + "learning_rate": 4.777448071216617e-07, + "loss": 0.4256, + "mean_token_accuracy": 0.8572893142700195, + "num_tokens": 43752506.0, + "step": 1128 + }, + { + "epoch": 0.1436204045286859, + "ewc_loss": 0.003753662109375, + "ewc_loss_diag": 3.3080577850341797e-06, + "ewc_loss_parallel": 3.4123659133911133e-06, + "grad_norm": 5.1254754066467285, + "learning_rate": 4.781687155574396e-07, + "loss": 0.4409, + "mean_token_accuracy": 0.8519695997238159, + "num_tokens": 43797442.0, + "step": 1129 + }, + { + "epoch": 0.14374761480727644, + "ewc_loss": 0.0037841796875, + "ewc_loss_diag": 3.293156623840332e-06, + "ewc_loss_parallel": 3.4421682357788086e-06, + "grad_norm": 5.147645473480225, + "learning_rate": 4.785926239932175e-07, + "loss": 0.3994, + "mean_token_accuracy": 0.8638171553611755, + "num_tokens": 43838209.0, + "step": 1130 + }, + { + "epoch": 0.14387482508586694, + "ewc_loss": 0.003753662109375, + "ewc_loss_diag": 3.293156623840332e-06, + "ewc_loss_parallel": 3.4123659133911133e-06, + "grad_norm": 5.270631313323975, + "learning_rate": 4.790165324289953e-07, + "loss": 0.4171, + "mean_token_accuracy": 0.8605523705482483, + "num_tokens": 43877086.0, + "step": 1131 + }, + { + "epoch": 0.14400203536445744, + "ewc_loss": 0.003753662109375, + "ewc_loss_diag": 3.293156623840332e-06, + "ewc_loss_parallel": 3.4123659133911133e-06, + "grad_norm": 5.247611999511719, + "learning_rate": 4.794404408647732e-07, + "loss": 0.4152, + "mean_token_accuracy": 0.8623624444007874, + "num_tokens": 43914998.0, + "step": 1132 + }, + { + "epoch": 0.14412924564304797, + "ewc_loss": 0.0037689208984375, + "ewc_loss_diag": 3.3229589462280273e-06, + "ewc_loss_parallel": 3.427267074584961e-06, + "grad_norm": 5.111644744873047, + "learning_rate": 4.798643493005511e-07, + "loss": 0.3776, + "mean_token_accuracy": 0.8707486391067505, + "num_tokens": 43955551.0, + "step": 1133 + }, + { + "epoch": 0.14425645592163847, + "ewc_loss": 0.003753662109375, + "ewc_loss_diag": 3.3229589462280273e-06, + "ewc_loss_parallel": 3.4123659133911133e-06, + "grad_norm": 5.303378105163574, + "learning_rate": 4.80288257736329e-07, + "loss": 0.4529, + "mean_token_accuracy": 0.8513711094856262, + "num_tokens": 43994096.0, + "step": 1134 + }, + { + "epoch": 0.14438366620022897, + "ewc_loss": 0.0037841796875, + "ewc_loss_diag": 3.3229589462280273e-06, + "ewc_loss_parallel": 3.4421682357788086e-06, + "grad_norm": 5.266082763671875, + "learning_rate": 4.807121661721068e-07, + "loss": 0.3822, + "mean_token_accuracy": 0.8720699548721313, + "num_tokens": 44029898.0, + "step": 1135 + }, + { + "epoch": 0.1445108764788195, + "ewc_loss": 0.003753662109375, + "ewc_loss_diag": 3.3229589462280273e-06, + "ewc_loss_parallel": 3.4123659133911133e-06, + "grad_norm": 5.158485412597656, + "learning_rate": 4.811360746078847e-07, + "loss": 0.4213, + "mean_token_accuracy": 0.861770510673523, + "num_tokens": 44070205.0, + "step": 1136 + }, + { + "epoch": 0.14463808675741, + "ewc_loss": 0.0037689208984375, + "ewc_loss_diag": 3.3229589462280273e-06, + "ewc_loss_parallel": 3.427267074584961e-06, + "grad_norm": 5.473681449890137, + "learning_rate": 4.815599830436625e-07, + "loss": 0.4174, + "mean_token_accuracy": 0.8601453900337219, + "num_tokens": 44103217.0, + "step": 1137 + }, + { + "epoch": 0.1447652970360005, + "ewc_loss": 0.0037994384765625, + "ewc_loss_diag": 3.3229589462280273e-06, + "ewc_loss_parallel": 3.4570693969726562e-06, + "grad_norm": 5.298386573791504, + "learning_rate": 4.819838914794405e-07, + "loss": 0.3942, + "mean_token_accuracy": 0.8659668564796448, + "num_tokens": 44140537.0, + "step": 1138 + }, + { + "epoch": 0.14489250731459102, + "ewc_loss": 0.0037078857421875, + "ewc_loss_diag": 3.3229589462280273e-06, + "ewc_loss_parallel": 3.3676624298095703e-06, + "grad_norm": 5.370724201202393, + "learning_rate": 4.824077999152183e-07, + "loss": 0.4297, + "mean_token_accuracy": 0.8547964692115784, + "num_tokens": 44174783.0, + "step": 1139 + }, + { + "epoch": 0.14501971759318152, + "ewc_loss": 0.0037841796875, + "ewc_loss_diag": 3.3229589462280273e-06, + "ewc_loss_parallel": 3.4421682357788086e-06, + "grad_norm": 5.221035003662109, + "learning_rate": 4.828317083509962e-07, + "loss": 0.4179, + "mean_token_accuracy": 0.8608227372169495, + "num_tokens": 44213368.0, + "step": 1140 + }, + { + "epoch": 0.14514692787177205, + "ewc_loss": 0.00372314453125, + "ewc_loss_diag": 3.3229589462280273e-06, + "ewc_loss_parallel": 3.382563591003418e-06, + "grad_norm": 5.232266902923584, + "learning_rate": 4.83255616786774e-07, + "loss": 0.3945, + "mean_token_accuracy": 0.8707588911056519, + "num_tokens": 44247075.0, + "step": 1141 + }, + { + "epoch": 0.14527413815036255, + "ewc_loss": 0.0037841796875, + "ewc_loss_diag": 3.3229589462280273e-06, + "ewc_loss_parallel": 3.4421682357788086e-06, + "grad_norm": 5.310890197753906, + "learning_rate": 4.83679525222552e-07, + "loss": 0.3808, + "mean_token_accuracy": 0.8722742795944214, + "num_tokens": 44285244.0, + "step": 1142 + }, + { + "epoch": 0.14540134842895305, + "ewc_loss": 0.0037841796875, + "ewc_loss_diag": 3.3527612686157227e-06, + "ewc_loss_parallel": 3.4421682357788086e-06, + "grad_norm": 5.306795120239258, + "learning_rate": 4.841034336583298e-07, + "loss": 0.4066, + "mean_token_accuracy": 0.8657389879226685, + "num_tokens": 44324318.0, + "step": 1143 + }, + { + "epoch": 0.14552855870754358, + "ewc_loss": 0.0037841796875, + "ewc_loss_diag": 3.3527612686157227e-06, + "ewc_loss_parallel": 3.4421682357788086e-06, + "grad_norm": 5.132187366485596, + "learning_rate": 4.845273420941076e-07, + "loss": 0.3791, + "mean_token_accuracy": 0.8738767504692078, + "num_tokens": 44365294.0, + "step": 1144 + }, + { + "epoch": 0.14565576898613408, + "ewc_loss": 0.0037841796875, + "ewc_loss_diag": 3.3676624298095703e-06, + "ewc_loss_parallel": 3.4421682357788086e-06, + "grad_norm": 5.485114574432373, + "learning_rate": 4.849512505298855e-07, + "loss": 0.4238, + "mean_token_accuracy": 0.8577252626419067, + "num_tokens": 44396314.0, + "step": 1145 + }, + { + "epoch": 0.14578297926472458, + "ewc_loss": 0.003814697265625, + "ewc_loss_diag": 3.382563591003418e-06, + "ewc_loss_parallel": 3.471970558166504e-06, + "grad_norm": 5.1711602210998535, + "learning_rate": 4.853751589656634e-07, + "loss": 0.4002, + "mean_token_accuracy": 0.8682210445404053, + "num_tokens": 44439457.0, + "step": 1146 + }, + { + "epoch": 0.1459101895433151, + "ewc_loss": 0.00372314453125, + "ewc_loss_diag": 3.382563591003418e-06, + "ewc_loss_parallel": 3.382563591003418e-06, + "grad_norm": 5.138467788696289, + "learning_rate": 4.857990674014413e-07, + "loss": 0.4352, + "mean_token_accuracy": 0.8558518886566162, + "num_tokens": 44483715.0, + "step": 1147 + }, + { + "epoch": 0.1460373998219056, + "ewc_loss": 0.0037994384765625, + "ewc_loss_diag": 3.382563591003418e-06, + "ewc_loss_parallel": 3.4570693969726562e-06, + "grad_norm": 5.240936279296875, + "learning_rate": 4.862229758372191e-07, + "loss": 0.4499, + "mean_token_accuracy": 0.849408745765686, + "num_tokens": 44524739.0, + "step": 1148 + }, + { + "epoch": 0.1461646101004961, + "ewc_loss": 0.003814697265625, + "ewc_loss_diag": 3.382563591003418e-06, + "ewc_loss_parallel": 3.471970558166504e-06, + "grad_norm": 5.220460414886475, + "learning_rate": 4.86646884272997e-07, + "loss": 0.4389, + "mean_token_accuracy": 0.8513877391815186, + "num_tokens": 44566541.0, + "step": 1149 + }, + { + "epoch": 0.14629182037908664, + "ewc_loss": 0.0037841796875, + "ewc_loss_diag": 3.382563591003418e-06, + "ewc_loss_parallel": 3.4421682357788086e-06, + "grad_norm": 5.2331342697143555, + "learning_rate": 4.870707927087749e-07, + "loss": 0.4177, + "mean_token_accuracy": 0.8593159914016724, + "num_tokens": 44605603.0, + "step": 1150 + }, + { + "epoch": 0.14641903065767714, + "ewc_loss": 0.0037994384765625, + "ewc_loss_diag": 3.382563591003418e-06, + "ewc_loss_parallel": 3.4570693969726562e-06, + "grad_norm": 5.277559757232666, + "learning_rate": 4.874947011445528e-07, + "loss": 0.4257, + "mean_token_accuracy": 0.8591038584709167, + "num_tokens": 44644129.0, + "step": 1151 + }, + { + "epoch": 0.14654624093626764, + "ewc_loss": 0.0037994384765625, + "ewc_loss_diag": 3.4123659133911133e-06, + "ewc_loss_parallel": 3.4570693969726562e-06, + "grad_norm": 5.203845500946045, + "learning_rate": 4.879186095803306e-07, + "loss": 0.367, + "mean_token_accuracy": 0.8775832056999207, + "num_tokens": 44684786.0, + "step": 1152 + }, + { + "epoch": 0.14667345121485817, + "ewc_loss": 0.0037841796875, + "ewc_loss_diag": 3.4123659133911133e-06, + "ewc_loss_parallel": 3.4421682357788086e-06, + "grad_norm": 5.324742794036865, + "learning_rate": 4.883425180161085e-07, + "loss": 0.3956, + "mean_token_accuracy": 0.8672983646392822, + "num_tokens": 44722286.0, + "step": 1153 + }, + { + "epoch": 0.14680066149344867, + "ewc_loss": 0.0037994384765625, + "ewc_loss_diag": 3.4123659133911133e-06, + "ewc_loss_parallel": 3.4570693969726562e-06, + "grad_norm": 5.324939250946045, + "learning_rate": 4.887664264518864e-07, + "loss": 0.421, + "mean_token_accuracy": 0.863895058631897, + "num_tokens": 44760772.0, + "step": 1154 + }, + { + "epoch": 0.14692787177203917, + "ewc_loss": 0.003814697265625, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.471970558166504e-06, + "grad_norm": 5.215959072113037, + "learning_rate": 4.891903348876643e-07, + "loss": 0.4293, + "mean_token_accuracy": 0.8586261868476868, + "num_tokens": 44800631.0, + "step": 1155 + }, + { + "epoch": 0.1470550820506297, + "ewc_loss": 0.003814697265625, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.471970558166504e-06, + "grad_norm": 5.3345513343811035, + "learning_rate": 4.896142433234421e-07, + "loss": 0.4568, + "mean_token_accuracy": 0.8539276123046875, + "num_tokens": 44838904.0, + "step": 1156 + }, + { + "epoch": 0.1471822923292202, + "ewc_loss": 0.003814697265625, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.471970558166504e-06, + "grad_norm": 5.475808143615723, + "learning_rate": 4.9003815175922e-07, + "loss": 0.4652, + "mean_token_accuracy": 0.8500531911849976, + "num_tokens": 44877454.0, + "step": 1157 + }, + { + "epoch": 0.1473095026078107, + "ewc_loss": 0.003814697265625, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.471970558166504e-06, + "grad_norm": 5.315023899078369, + "learning_rate": 4.904620601949979e-07, + "loss": 0.4169, + "mean_token_accuracy": 0.860934853553772, + "num_tokens": 44917286.0, + "step": 1158 + }, + { + "epoch": 0.14743671288640123, + "ewc_loss": 0.0037841796875, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.4421682357788086e-06, + "grad_norm": 5.439367294311523, + "learning_rate": 4.908859686307758e-07, + "loss": 0.4419, + "mean_token_accuracy": 0.854337215423584, + "num_tokens": 44951147.0, + "step": 1159 + }, + { + "epoch": 0.14756392316499173, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.5017728805541992e-06, + "grad_norm": 5.404142379760742, + "learning_rate": 4.913098770665536e-07, + "loss": 0.4478, + "mean_token_accuracy": 0.8557865619659424, + "num_tokens": 44988907.0, + "step": 1160 + }, + { + "epoch": 0.14769113344358223, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.4868717193603516e-06, + "grad_norm": 5.3560285568237305, + "learning_rate": 4.917337855023314e-07, + "loss": 0.3988, + "mean_token_accuracy": 0.8619430065155029, + "num_tokens": 45030511.0, + "step": 1161 + }, + { + "epoch": 0.14781834372217276, + "ewc_loss": 0.003814697265625, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.471970558166504e-06, + "grad_norm": 5.413672924041748, + "learning_rate": 4.921576939381094e-07, + "loss": 0.4283, + "mean_token_accuracy": 0.8623940944671631, + "num_tokens": 45068500.0, + "step": 1162 + }, + { + "epoch": 0.14794555400076326, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.5017728805541992e-06, + "grad_norm": 5.303444862365723, + "learning_rate": 4.925816023738872e-07, + "loss": 0.4585, + "mean_token_accuracy": 0.8478683233261108, + "num_tokens": 45107287.0, + "step": 1163 + }, + { + "epoch": 0.14807276427935376, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.4868717193603516e-06, + "grad_norm": 5.611172676086426, + "learning_rate": 4.930055108096651e-07, + "loss": 0.3687, + "mean_token_accuracy": 0.8765559196472168, + "num_tokens": 45136778.0, + "step": 1164 + }, + { + "epoch": 0.1481999745579443, + "ewc_loss": 0.003875732421875, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.5315752029418945e-06, + "grad_norm": 5.47198486328125, + "learning_rate": 4.934294192454429e-07, + "loss": 0.4732, + "mean_token_accuracy": 0.8437882661819458, + "num_tokens": 45172297.0, + "step": 1165 + }, + { + "epoch": 0.1483271848365348, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.5017728805541992e-06, + "grad_norm": 5.26603889465332, + "learning_rate": 4.938533276812209e-07, + "loss": 0.4227, + "mean_token_accuracy": 0.8580476641654968, + "num_tokens": 45208483.0, + "step": 1166 + }, + { + "epoch": 0.14845439511512531, + "ewc_loss": 0.003814697265625, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.471970558166504e-06, + "grad_norm": 5.22802734375, + "learning_rate": 4.942772361169987e-07, + "loss": 0.3511, + "mean_token_accuracy": 0.8773229718208313, + "num_tokens": 45249029.0, + "step": 1167 + }, + { + "epoch": 0.14858160539371582, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.4868717193603516e-06, + "grad_norm": 5.195952415466309, + "learning_rate": 4.947011445527766e-07, + "loss": 0.4177, + "mean_token_accuracy": 0.8630258440971375, + "num_tokens": 45292318.0, + "step": 1168 + }, + { + "epoch": 0.14870881567230632, + "ewc_loss": 0.003814697265625, + "ewc_loss_diag": 3.427267074584961e-06, + "ewc_loss_parallel": 3.471970558166504e-06, + "grad_norm": 5.178487777709961, + "learning_rate": 4.951250529885544e-07, + "loss": 0.3821, + "mean_token_accuracy": 0.8712242245674133, + "num_tokens": 45333416.0, + "step": 1169 + }, + { + "epoch": 0.14883602595089684, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.4421682357788086e-06, + "ewc_loss_parallel": 3.4868717193603516e-06, + "grad_norm": 5.4085540771484375, + "learning_rate": 4.955489614243324e-07, + "loss": 0.4944, + "mean_token_accuracy": 0.8337928056716919, + "num_tokens": 45372408.0, + "step": 1170 + }, + { + "epoch": 0.14896323622948734, + "ewc_loss": 0.003875732421875, + "ewc_loss_diag": 3.4421682357788086e-06, + "ewc_loss_parallel": 3.546476364135742e-06, + "grad_norm": 5.248589515686035, + "learning_rate": 4.959728698601102e-07, + "loss": 0.4215, + "mean_token_accuracy": 0.8556429147720337, + "num_tokens": 45414087.0, + "step": 1171 + }, + { + "epoch": 0.14909044650807785, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.4421682357788086e-06, + "ewc_loss_parallel": 3.5017728805541992e-06, + "grad_norm": 5.290245532989502, + "learning_rate": 4.963967782958881e-07, + "loss": 0.4004, + "mean_token_accuracy": 0.8664905428886414, + "num_tokens": 45457865.0, + "step": 1172 + }, + { + "epoch": 0.14921765678666837, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.4421682357788086e-06, + "ewc_loss_parallel": 3.516674041748047e-06, + "grad_norm": 5.284210681915283, + "learning_rate": 4.968206867316659e-07, + "loss": 0.399, + "mean_token_accuracy": 0.8667311668395996, + "num_tokens": 45498542.0, + "step": 1173 + }, + { + "epoch": 0.14934486706525887, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.4421682357788086e-06, + "ewc_loss_parallel": 3.516674041748047e-06, + "grad_norm": 5.517704010009766, + "learning_rate": 4.972445951674439e-07, + "loss": 0.396, + "mean_token_accuracy": 0.8663930296897888, + "num_tokens": 45530415.0, + "step": 1174 + }, + { + "epoch": 0.14947207734384937, + "ewc_loss": 0.003936767578125, + "ewc_loss_diag": 3.4421682357788086e-06, + "ewc_loss_parallel": 3.591179847717285e-06, + "grad_norm": 5.448206424713135, + "learning_rate": 4.976685036032216e-07, + "loss": 0.477, + "mean_token_accuracy": 0.8415933847427368, + "num_tokens": 45569590.0, + "step": 1175 + }, + { + "epoch": 0.1495992876224399, + "ewc_loss": 0.0038604736328125, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.516674041748047e-06, + "grad_norm": 5.158636093139648, + "learning_rate": 4.980924120389996e-07, + "loss": 0.357, + "mean_token_accuracy": 0.8810096979141235, + "num_tokens": 45608849.0, + "step": 1176 + }, + { + "epoch": 0.1497264979010304, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.5017728805541992e-06, + "grad_norm": 5.172869682312012, + "learning_rate": 4.985163204747774e-07, + "loss": 0.3576, + "mean_token_accuracy": 0.8790899515151978, + "num_tokens": 45649760.0, + "step": 1177 + }, + { + "epoch": 0.1498537081796209, + "ewc_loss": 0.0038299560546875, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.471970558166504e-06, + "grad_norm": 5.299200057983398, + "learning_rate": 4.989402289105554e-07, + "loss": 0.4079, + "mean_token_accuracy": 0.8670523166656494, + "num_tokens": 45694374.0, + "step": 1178 + }, + { + "epoch": 0.14998091845821143, + "ewc_loss": 0.0038604736328125, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.516674041748047e-06, + "grad_norm": 5.539764404296875, + "learning_rate": 4.993641373463331e-07, + "loss": 0.4752, + "mean_token_accuracy": 0.8439931273460388, + "num_tokens": 45731452.0, + "step": 1179 + }, + { + "epoch": 0.15010812873680193, + "ewc_loss": 0.003936767578125, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.591179847717285e-06, + "grad_norm": 5.401214122772217, + "learning_rate": 4.997880457821111e-07, + "loss": 0.4256, + "mean_token_accuracy": 0.8562501668930054, + "num_tokens": 45770871.0, + "step": 1180 + }, + { + "epoch": 0.15023533901539243, + "ewc_loss": 0.0038604736328125, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.516674041748047e-06, + "grad_norm": 5.234984874725342, + "learning_rate": 5.002119542178889e-07, + "loss": 0.4228, + "mean_token_accuracy": 0.8594746589660645, + "num_tokens": 45814792.0, + "step": 1181 + }, + { + "epoch": 0.15036254929398296, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.4868717193603516e-06, + "grad_norm": 5.351499557495117, + "learning_rate": 5.006358626536667e-07, + "loss": 0.3817, + "mean_token_accuracy": 0.8714126348495483, + "num_tokens": 45851980.0, + "step": 1182 + }, + { + "epoch": 0.15048975957257346, + "ewc_loss": 0.003875732421875, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.5315752029418945e-06, + "grad_norm": 5.393517017364502, + "learning_rate": 5.010597710894446e-07, + "loss": 0.3995, + "mean_token_accuracy": 0.8637240529060364, + "num_tokens": 45889887.0, + "step": 1183 + }, + { + "epoch": 0.15061696985116396, + "ewc_loss": 0.003875732421875, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.5315752029418945e-06, + "grad_norm": 5.404367446899414, + "learning_rate": 5.014836795252225e-07, + "loss": 0.3838, + "mean_token_accuracy": 0.8711243867874146, + "num_tokens": 45928765.0, + "step": 1184 + }, + { + "epoch": 0.1507441801297545, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.5017728805541992e-06, + "grad_norm": 5.296230316162109, + "learning_rate": 5.019075879610004e-07, + "loss": 0.3857, + "mean_token_accuracy": 0.8747784495353699, + "num_tokens": 45967700.0, + "step": 1185 + }, + { + "epoch": 0.150871390408345, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.4868717193603516e-06, + "grad_norm": 5.220846652984619, + "learning_rate": 5.023314963967783e-07, + "loss": 0.4201, + "mean_token_accuracy": 0.8556191325187683, + "num_tokens": 46010356.0, + "step": 1186 + }, + { + "epoch": 0.1509986006869355, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.5017728805541992e-06, + "grad_norm": 5.5601043701171875, + "learning_rate": 5.027554048325562e-07, + "loss": 0.471, + "mean_token_accuracy": 0.844469428062439, + "num_tokens": 46043030.0, + "step": 1187 + }, + { + "epoch": 0.15112581096552602, + "ewc_loss": 0.00390625, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.56137752532959e-06, + "grad_norm": 5.329238414764404, + "learning_rate": 5.03179313268334e-07, + "loss": 0.4501, + "mean_token_accuracy": 0.8506402373313904, + "num_tokens": 46084813.0, + "step": 1188 + }, + { + "epoch": 0.15125302124411652, + "ewc_loss": 0.00384521484375, + "ewc_loss_diag": 3.4570693969726562e-06, + "ewc_loss_parallel": 3.5017728805541992e-06, + "grad_norm": 5.4115681648254395, + "learning_rate": 5.036032217041119e-07, + "loss": 0.4279, + "mean_token_accuracy": 0.85655677318573, + "num_tokens": 46121978.0, + "step": 1189 + }, + { + "epoch": 0.15138023152270705, + "ewc_loss": 0.0038909912109375, + "ewc_loss_diag": 3.471970558166504e-06, + "ewc_loss_parallel": 3.546476364135742e-06, + "grad_norm": 5.2559285163879395, + "learning_rate": 5.040271301398897e-07, + "loss": 0.4228, + "mean_token_accuracy": 0.857624351978302, + "num_tokens": 46164655.0, + "step": 1190 + }, + { + "epoch": 0.15150744180129755, + "ewc_loss": 0.0038604736328125, + "ewc_loss_diag": 3.471970558166504e-06, + "ewc_loss_parallel": 3.516674041748047e-06, + "grad_norm": 5.1890153884887695, + "learning_rate": 5.044510385756676e-07, + "loss": 0.3936, + "mean_token_accuracy": 0.8663235902786255, + "num_tokens": 46209693.0, + "step": 1191 + }, + { + "epoch": 0.15163465207988805, + "ewc_loss": 0.0038604736328125, + "ewc_loss_diag": 3.471970558166504e-06, + "ewc_loss_parallel": 3.516674041748047e-06, + "grad_norm": 5.316302299499512, + "learning_rate": 5.048749470114455e-07, + "loss": 0.4042, + "mean_token_accuracy": 0.8659896850585938, + "num_tokens": 46247700.0, + "step": 1192 + }, + { + "epoch": 0.15176186235847858, + "ewc_loss": 0.00390625, + "ewc_loss_diag": 3.4868717193603516e-06, + "ewc_loss_parallel": 3.56137752532959e-06, + "grad_norm": 5.648956775665283, + "learning_rate": 5.052988554472234e-07, + "loss": 0.4188, + "mean_token_accuracy": 0.8568596839904785, + "num_tokens": 46279130.0, + "step": 1193 + }, + { + "epoch": 0.15188907263706908, + "ewc_loss": 0.00396728515625, + "ewc_loss_diag": 3.4868717193603516e-06, + "ewc_loss_parallel": 3.635883331298828e-06, + "grad_norm": 5.354831695556641, + "learning_rate": 5.057227638830013e-07, + "loss": 0.4183, + "mean_token_accuracy": 0.861095666885376, + "num_tokens": 46316303.0, + "step": 1194 + }, + { + "epoch": 0.15201628291565958, + "ewc_loss": 0.00390625, + "ewc_loss_diag": 3.4868717193603516e-06, + "ewc_loss_parallel": 3.5762786865234375e-06, + "grad_norm": 5.398101329803467, + "learning_rate": 5.061466723187792e-07, + "loss": 0.4249, + "mean_token_accuracy": 0.8591430187225342, + "num_tokens": 46355537.0, + "step": 1195 + }, + { + "epoch": 0.1521434931942501, + "ewc_loss": 0.00390625, + "ewc_loss_diag": 3.4868717193603516e-06, + "ewc_loss_parallel": 3.5762786865234375e-06, + "grad_norm": 5.624593257904053, + "learning_rate": 5.065705807545569e-07, + "loss": 0.4003, + "mean_token_accuracy": 0.8673827648162842, + "num_tokens": 46389416.0, + "step": 1196 + }, + { + "epoch": 0.1522707034728406, + "ewc_loss": 0.003936767578125, + "ewc_loss_diag": 3.4868717193603516e-06, + "ewc_loss_parallel": 3.591179847717285e-06, + "grad_norm": 5.304222106933594, + "learning_rate": 5.069944891903349e-07, + "loss": 0.4032, + "mean_token_accuracy": 0.863602340221405, + "num_tokens": 46434238.0, + "step": 1197 + }, + { + "epoch": 0.1523979137514311, + "ewc_loss": 0.003875732421875, + "ewc_loss_diag": 3.516674041748047e-06, + "ewc_loss_parallel": 3.5315752029418945e-06, + "grad_norm": 5.307266712188721, + "learning_rate": 5.074183976261127e-07, + "loss": 0.4093, + "mean_token_accuracy": 0.8614529371261597, + "num_tokens": 46472191.0, + "step": 1198 + }, + { + "epoch": 0.15252512403002164, + "ewc_loss": 0.0038909912109375, + "ewc_loss_diag": 3.516674041748047e-06, + "ewc_loss_parallel": 3.546476364135742e-06, + "grad_norm": 5.335350513458252, + "learning_rate": 5.078423060618906e-07, + "loss": 0.4158, + "mean_token_accuracy": 0.8625667095184326, + "num_tokens": 46515159.0, + "step": 1199 + }, + { + "epoch": 0.15265233430861214, + "ewc_loss": 0.00390625, + "ewc_loss_diag": 3.516674041748047e-06, + "ewc_loss_parallel": 3.5762786865234375e-06, + "grad_norm": 5.410508632659912, + "learning_rate": 5.082662144976685e-07, + "loss": 0.4174, + "mean_token_accuracy": 0.8577039241790771, + "num_tokens": 46552211.0, + "step": 1200 + }, + { + "epoch": 0.15277954458720264, + "ewc_loss": 0.00396728515625, + "ewc_loss_diag": 3.516674041748047e-06, + "ewc_loss_parallel": 3.6209821701049805e-06, + "grad_norm": 5.4525532722473145, + "learning_rate": 5.086901229334464e-07, + "loss": 0.4143, + "mean_token_accuracy": 0.8651189208030701, + "num_tokens": 46588174.0, + "step": 1201 + }, + { + "epoch": 0.15290675486579317, + "ewc_loss": 0.00396728515625, + "ewc_loss_diag": 3.516674041748047e-06, + "ewc_loss_parallel": 3.6209821701049805e-06, + "grad_norm": 5.375117778778076, + "learning_rate": 5.091140313692243e-07, + "loss": 0.413, + "mean_token_accuracy": 0.8598178625106812, + "num_tokens": 46625544.0, + "step": 1202 + }, + { + "epoch": 0.15303396514438367, + "ewc_loss": 0.00390625, + "ewc_loss_diag": 3.516674041748047e-06, + "ewc_loss_parallel": 3.5762786865234375e-06, + "grad_norm": 5.375039577484131, + "learning_rate": 5.095379398050022e-07, + "loss": 0.4036, + "mean_token_accuracy": 0.8629077672958374, + "num_tokens": 46664310.0, + "step": 1203 + }, + { + "epoch": 0.15316117542297417, + "ewc_loss": 0.00390625, + "ewc_loss_diag": 3.516674041748047e-06, + "ewc_loss_parallel": 3.5762786865234375e-06, + "grad_norm": 5.58187198638916, + "learning_rate": 5.099618482407799e-07, + "loss": 0.4224, + "mean_token_accuracy": 0.8598231077194214, + "num_tokens": 46697955.0, + "step": 1204 + }, + { + "epoch": 0.1532883857015647, + "ewc_loss": 0.004058837890625, + "ewc_loss_diag": 3.56137752532959e-06, + "ewc_loss_parallel": 3.6954879760742188e-06, + "grad_norm": 5.402242183685303, + "learning_rate": 5.103857566765578e-07, + "loss": 0.4001, + "mean_token_accuracy": 0.8673354983329773, + "num_tokens": 46738326.0, + "step": 1205 + }, + { + "epoch": 0.1534155959801552, + "ewc_loss": 0.00396728515625, + "ewc_loss_diag": 3.591179847717285e-06, + "ewc_loss_parallel": 3.606081008911133e-06, + "grad_norm": 5.244981288909912, + "learning_rate": 5.108096651123357e-07, + "loss": 0.3633, + "mean_token_accuracy": 0.876283586025238, + "num_tokens": 46774847.0, + "step": 1206 + }, + { + "epoch": 0.1535428062587457, + "ewc_loss": 0.00390625, + "ewc_loss_diag": 3.591179847717285e-06, + "ewc_loss_parallel": 3.56137752532959e-06, + "grad_norm": 5.367888927459717, + "learning_rate": 5.112335735481135e-07, + "loss": 0.4119, + "mean_token_accuracy": 0.8613618612289429, + "num_tokens": 46812773.0, + "step": 1207 + }, + { + "epoch": 0.15367001653733622, + "ewc_loss": 0.003997802734375, + "ewc_loss_diag": 3.591179847717285e-06, + "ewc_loss_parallel": 3.635883331298828e-06, + "grad_norm": 5.379398822784424, + "learning_rate": 5.116574819838915e-07, + "loss": 0.3346, + "mean_token_accuracy": 0.8863387703895569, + "num_tokens": 46851420.0, + "step": 1208 + }, + { + "epoch": 0.15379722681592672, + "ewc_loss": 0.003997802734375, + "ewc_loss_diag": 3.591179847717285e-06, + "ewc_loss_parallel": 3.6507844924926758e-06, + "grad_norm": 5.491570949554443, + "learning_rate": 5.120813904196693e-07, + "loss": 0.395, + "mean_token_accuracy": 0.8688982725143433, + "num_tokens": 46888131.0, + "step": 1209 + }, + { + "epoch": 0.15392443709451722, + "ewc_loss": 0.0040283203125, + "ewc_loss_diag": 3.591179847717285e-06, + "ewc_loss_parallel": 3.680586814880371e-06, + "grad_norm": 5.597542762756348, + "learning_rate": 5.125052988554473e-07, + "loss": 0.4664, + "mean_token_accuracy": 0.8494727611541748, + "num_tokens": 46921590.0, + "step": 1210 + }, + { + "epoch": 0.15405164737310775, + "ewc_loss": 0.004058837890625, + "ewc_loss_diag": 3.591179847717285e-06, + "ewc_loss_parallel": 3.6954879760742188e-06, + "grad_norm": 5.510779857635498, + "learning_rate": 5.12929207291225e-07, + "loss": 0.4059, + "mean_token_accuracy": 0.8638455867767334, + "num_tokens": 46955354.0, + "step": 1211 + }, + { + "epoch": 0.15417885765169825, + "ewc_loss": 0.0040283203125, + "ewc_loss_diag": 3.6209821701049805e-06, + "ewc_loss_parallel": 3.6656856536865234e-06, + "grad_norm": 5.305873870849609, + "learning_rate": 5.133531157270029e-07, + "loss": 0.3793, + "mean_token_accuracy": 0.8721023797988892, + "num_tokens": 46996911.0, + "step": 1212 + }, + { + "epoch": 0.15430606793028875, + "ewc_loss": 0.003936767578125, + "ewc_loss_diag": 3.6209821701049805e-06, + "ewc_loss_parallel": 3.591179847717285e-06, + "grad_norm": 5.432338714599609, + "learning_rate": 5.137770241627808e-07, + "loss": 0.3481, + "mean_token_accuracy": 0.8806793093681335, + "num_tokens": 47032820.0, + "step": 1213 + }, + { + "epoch": 0.15443327820887928, + "ewc_loss": 0.003997802734375, + "ewc_loss_diag": 3.6209821701049805e-06, + "ewc_loss_parallel": 3.6507844924926758e-06, + "grad_norm": 5.394162178039551, + "learning_rate": 5.142009325985587e-07, + "loss": 0.3909, + "mean_token_accuracy": 0.8702911734580994, + "num_tokens": 47070799.0, + "step": 1214 + }, + { + "epoch": 0.15456048848746978, + "ewc_loss": 0.0040283203125, + "ewc_loss_diag": 3.6209821701049805e-06, + "ewc_loss_parallel": 3.6656856536865234e-06, + "grad_norm": 5.342956066131592, + "learning_rate": 5.146248410343365e-07, + "loss": 0.4204, + "mean_token_accuracy": 0.8596395254135132, + "num_tokens": 47111730.0, + "step": 1215 + }, + { + "epoch": 0.1546876987660603, + "ewc_loss": 0.00396728515625, + "ewc_loss_diag": 3.6507844924926758e-06, + "ewc_loss_parallel": 3.6209821701049805e-06, + "grad_norm": 5.1988935470581055, + "learning_rate": 5.150487494701145e-07, + "loss": 0.3615, + "mean_token_accuracy": 0.8786638379096985, + "num_tokens": 47155911.0, + "step": 1216 + }, + { + "epoch": 0.1548149090446508, + "ewc_loss": 0.00396728515625, + "ewc_loss_diag": 3.6507844924926758e-06, + "ewc_loss_parallel": 3.606081008911133e-06, + "grad_norm": 5.390436172485352, + "learning_rate": 5.154726579058923e-07, + "loss": 0.4107, + "mean_token_accuracy": 0.8625491857528687, + "num_tokens": 47191946.0, + "step": 1217 + }, + { + "epoch": 0.1549421193232413, + "ewc_loss": 0.004058837890625, + "ewc_loss_diag": 3.6507844924926758e-06, + "ewc_loss_parallel": 3.6954879760742188e-06, + "grad_norm": 5.249130725860596, + "learning_rate": 5.158965663416703e-07, + "loss": 0.3987, + "mean_token_accuracy": 0.8676154017448425, + "num_tokens": 47237989.0, + "step": 1218 + }, + { + "epoch": 0.15506932960183184, + "ewc_loss": 0.003997802734375, + "ewc_loss_diag": 3.6507844924926758e-06, + "ewc_loss_parallel": 3.6507844924926758e-06, + "grad_norm": 5.338109970092773, + "learning_rate": 5.16320474777448e-07, + "loss": 0.3752, + "mean_token_accuracy": 0.8721905946731567, + "num_tokens": 47280093.0, + "step": 1219 + }, + { + "epoch": 0.15519653988042234, + "ewc_loss": 0.0040283203125, + "ewc_loss_diag": 3.6507844924926758e-06, + "ewc_loss_parallel": 3.6656856536865234e-06, + "grad_norm": 5.432811737060547, + "learning_rate": 5.167443832132259e-07, + "loss": 0.4067, + "mean_token_accuracy": 0.861059308052063, + "num_tokens": 47320694.0, + "step": 1220 + }, + { + "epoch": 0.15532375015901284, + "ewc_loss": 0.003997802734375, + "ewc_loss_diag": 3.6656856536865234e-06, + "ewc_loss_parallel": 3.635883331298828e-06, + "grad_norm": 5.38633918762207, + "learning_rate": 5.171682916490038e-07, + "loss": 0.3879, + "mean_token_accuracy": 0.8705238103866577, + "num_tokens": 47356580.0, + "step": 1221 + }, + { + "epoch": 0.15545096043760337, + "ewc_loss": 0.003997802734375, + "ewc_loss_diag": 3.6656856536865234e-06, + "ewc_loss_parallel": 3.635883331298828e-06, + "grad_norm": 5.432325839996338, + "learning_rate": 5.175922000847816e-07, + "loss": 0.3779, + "mean_token_accuracy": 0.8745001554489136, + "num_tokens": 47391888.0, + "step": 1222 + }, + { + "epoch": 0.15557817071619387, + "ewc_loss": 0.003997802734375, + "ewc_loss_diag": 3.6656856536865234e-06, + "ewc_loss_parallel": 3.635883331298828e-06, + "grad_norm": 5.4010419845581055, + "learning_rate": 5.180161085205595e-07, + "loss": 0.3911, + "mean_token_accuracy": 0.8679474592208862, + "num_tokens": 47429235.0, + "step": 1223 + }, + { + "epoch": 0.15570538099478437, + "ewc_loss": 0.0040283203125, + "ewc_loss_diag": 3.6656856536865234e-06, + "ewc_loss_parallel": 3.6507844924926758e-06, + "grad_norm": 5.407121181488037, + "learning_rate": 5.184400169563374e-07, + "loss": 0.4463, + "mean_token_accuracy": 0.8517239689826965, + "num_tokens": 47469663.0, + "step": 1224 + }, + { + "epoch": 0.1558325912733749, + "ewc_loss": 0.004058837890625, + "ewc_loss_diag": 3.680586814880371e-06, + "ewc_loss_parallel": 3.6954879760742188e-06, + "grad_norm": 5.458103656768799, + "learning_rate": 5.188639253921153e-07, + "loss": 0.4011, + "mean_token_accuracy": 0.8654834628105164, + "num_tokens": 47512385.0, + "step": 1225 + }, + { + "epoch": 0.1559598015519654, + "ewc_loss": 0.0040283203125, + "ewc_loss_diag": 3.6954879760742188e-06, + "ewc_loss_parallel": 3.6656856536865234e-06, + "grad_norm": 5.4186882972717285, + "learning_rate": 5.192878338278932e-07, + "loss": 0.4048, + "mean_token_accuracy": 0.8637760877609253, + "num_tokens": 47549440.0, + "step": 1226 + }, + { + "epoch": 0.1560870118305559, + "ewc_loss": 0.003997802734375, + "ewc_loss_diag": 3.6954879760742188e-06, + "ewc_loss_parallel": 3.635883331298828e-06, + "grad_norm": 5.348538875579834, + "learning_rate": 5.19711742263671e-07, + "loss": 0.4447, + "mean_token_accuracy": 0.851204514503479, + "num_tokens": 47595202.0, + "step": 1227 + }, + { + "epoch": 0.15621422210914643, + "ewc_loss": 0.0040283203125, + "ewc_loss_diag": 3.6954879760742188e-06, + "ewc_loss_parallel": 3.6656856536865234e-06, + "grad_norm": 5.537812232971191, + "learning_rate": 5.201356506994488e-07, + "loss": 0.3919, + "mean_token_accuracy": 0.8637783527374268, + "num_tokens": 47627630.0, + "step": 1228 + }, + { + "epoch": 0.15634143238773693, + "ewc_loss": 0.00408935546875, + "ewc_loss_diag": 3.6954879760742188e-06, + "ewc_loss_parallel": 3.725290298461914e-06, + "grad_norm": 5.429952144622803, + "learning_rate": 5.205595591352268e-07, + "loss": 0.4741, + "mean_token_accuracy": 0.8441877365112305, + "num_tokens": 47668680.0, + "step": 1229 + }, + { + "epoch": 0.15646864266632743, + "ewc_loss": 0.0040283203125, + "ewc_loss_diag": 3.7103891372680664e-06, + "ewc_loss_parallel": 3.6507844924926758e-06, + "grad_norm": 5.302420139312744, + "learning_rate": 5.209834675710046e-07, + "loss": 0.405, + "mean_token_accuracy": 0.8638192415237427, + "num_tokens": 47711257.0, + "step": 1230 + }, + { + "epoch": 0.15659585294491796, + "ewc_loss": 0.0040283203125, + "ewc_loss_diag": 3.7103891372680664e-06, + "ewc_loss_parallel": 3.6656856536865234e-06, + "grad_norm": 5.607690811157227, + "learning_rate": 5.214073760067825e-07, + "loss": 0.4802, + "mean_token_accuracy": 0.8426226377487183, + "num_tokens": 47745652.0, + "step": 1231 + }, + { + "epoch": 0.15672306322350846, + "ewc_loss": 0.004119873046875, + "ewc_loss_diag": 3.7103891372680664e-06, + "ewc_loss_parallel": 3.7550926208496094e-06, + "grad_norm": 5.4450364112854, + "learning_rate": 5.218312844425604e-07, + "loss": 0.4085, + "mean_token_accuracy": 0.86391282081604, + "num_tokens": 47781941.0, + "step": 1232 + }, + { + "epoch": 0.15685027350209896, + "ewc_loss": 0.004058837890625, + "ewc_loss_diag": 3.725290298461914e-06, + "ewc_loss_parallel": 3.680586814880371e-06, + "grad_norm": 5.399691581726074, + "learning_rate": 5.222551928783383e-07, + "loss": 0.3871, + "mean_token_accuracy": 0.8663014769554138, + "num_tokens": 47819666.0, + "step": 1233 + }, + { + "epoch": 0.1569774837806895, + "ewc_loss": 0.004058837890625, + "ewc_loss_diag": 3.725290298461914e-06, + "ewc_loss_parallel": 3.680586814880371e-06, + "grad_norm": 5.404106140136719, + "learning_rate": 5.226791013141161e-07, + "loss": 0.446, + "mean_token_accuracy": 0.8510593175888062, + "num_tokens": 47862551.0, + "step": 1234 + }, + { + "epoch": 0.15710469405928, + "ewc_loss": 0.004058837890625, + "ewc_loss_diag": 3.7401914596557617e-06, + "ewc_loss_parallel": 3.680586814880371e-06, + "grad_norm": 5.330544471740723, + "learning_rate": 5.23103009749894e-07, + "loss": 0.3874, + "mean_token_accuracy": 0.8666567206382751, + "num_tokens": 47904601.0, + "step": 1235 + }, + { + "epoch": 0.1572319043378705, + "ewc_loss": 0.004058837890625, + "ewc_loss_diag": 3.7401914596557617e-06, + "ewc_loss_parallel": 3.680586814880371e-06, + "grad_norm": 5.282412528991699, + "learning_rate": 5.235269181856718e-07, + "loss": 0.3956, + "mean_token_accuracy": 0.866584062576294, + "num_tokens": 47950662.0, + "step": 1236 + }, + { + "epoch": 0.15735911461646102, + "ewc_loss": 0.0040283203125, + "ewc_loss_diag": 3.769993782043457e-06, + "ewc_loss_parallel": 3.6656856536865234e-06, + "grad_norm": 5.628488540649414, + "learning_rate": 5.239508266214498e-07, + "loss": 0.4655, + "mean_token_accuracy": 0.8411222696304321, + "num_tokens": 47984969.0, + "step": 1237 + }, + { + "epoch": 0.15748632489505152, + "ewc_loss": 0.004150390625, + "ewc_loss_diag": 3.769993782043457e-06, + "ewc_loss_parallel": 3.7848949432373047e-06, + "grad_norm": 5.443245887756348, + "learning_rate": 5.243747350572276e-07, + "loss": 0.416, + "mean_token_accuracy": 0.8659294843673706, + "num_tokens": 48022621.0, + "step": 1238 + }, + { + "epoch": 0.15761353517364202, + "ewc_loss": 0.004058837890625, + "ewc_loss_diag": 3.769993782043457e-06, + "ewc_loss_parallel": 3.6954879760742188e-06, + "grad_norm": 5.460919380187988, + "learning_rate": 5.247986434930056e-07, + "loss": 0.4125, + "mean_token_accuracy": 0.8660244345664978, + "num_tokens": 48060097.0, + "step": 1239 + }, + { + "epoch": 0.15774074545223254, + "ewc_loss": 0.00408935546875, + "ewc_loss_diag": 3.769993782043457e-06, + "ewc_loss_parallel": 3.725290298461914e-06, + "grad_norm": 5.567368030548096, + "learning_rate": 5.252225519287834e-07, + "loss": 0.4679, + "mean_token_accuracy": 0.8432124853134155, + "num_tokens": 48098360.0, + "step": 1240 + }, + { + "epoch": 0.15786795573082305, + "ewc_loss": 0.004119873046875, + "ewc_loss_diag": 3.769993782043457e-06, + "ewc_loss_parallel": 3.7550926208496094e-06, + "grad_norm": 5.4867844581604, + "learning_rate": 5.256464603645613e-07, + "loss": 0.404, + "mean_token_accuracy": 0.8675559759140015, + "num_tokens": 48136517.0, + "step": 1241 + }, + { + "epoch": 0.15799516600941357, + "ewc_loss": 0.004119873046875, + "ewc_loss_diag": 3.769993782043457e-06, + "ewc_loss_parallel": 3.7550926208496094e-06, + "grad_norm": 5.439849376678467, + "learning_rate": 5.260703688003391e-07, + "loss": 0.4103, + "mean_token_accuracy": 0.8629876375198364, + "num_tokens": 48179411.0, + "step": 1242 + }, + { + "epoch": 0.15812237628800407, + "ewc_loss": 0.00408935546875, + "ewc_loss_diag": 3.769993782043457e-06, + "ewc_loss_parallel": 3.725290298461914e-06, + "grad_norm": 5.543091297149658, + "learning_rate": 5.26494277236117e-07, + "loss": 0.4103, + "mean_token_accuracy": 0.8647924065589905, + "num_tokens": 48212904.0, + "step": 1243 + }, + { + "epoch": 0.15824958656659457, + "ewc_loss": 0.004119873046875, + "ewc_loss_diag": 3.769993782043457e-06, + "ewc_loss_parallel": 3.7550926208496094e-06, + "grad_norm": 5.562036514282227, + "learning_rate": 5.269181856718948e-07, + "loss": 0.4487, + "mean_token_accuracy": 0.8500509262084961, + "num_tokens": 48250693.0, + "step": 1244 + }, + { + "epoch": 0.1583767968451851, + "ewc_loss": 0.004150390625, + "ewc_loss_diag": 3.769993782043457e-06, + "ewc_loss_parallel": 3.7848949432373047e-06, + "grad_norm": 5.457493782043457, + "learning_rate": 5.273420941076727e-07, + "loss": 0.3972, + "mean_token_accuracy": 0.8692665100097656, + "num_tokens": 48286391.0, + "step": 1245 + }, + { + "epoch": 0.1585040071237756, + "ewc_loss": 0.004150390625, + "ewc_loss_diag": 3.769993782043457e-06, + "ewc_loss_parallel": 3.769993782043457e-06, + "grad_norm": 5.315028190612793, + "learning_rate": 5.277660025434506e-07, + "loss": 0.3637, + "mean_token_accuracy": 0.8782914280891418, + "num_tokens": 48326552.0, + "step": 1246 + }, + { + "epoch": 0.1586312174023661, + "ewc_loss": 0.00408935546875, + "ewc_loss_diag": 3.7848949432373047e-06, + "ewc_loss_parallel": 3.7103891372680664e-06, + "grad_norm": 5.319695949554443, + "learning_rate": 5.281899109792285e-07, + "loss": 0.3754, + "mean_token_accuracy": 0.8728365302085876, + "num_tokens": 48367931.0, + "step": 1247 + }, + { + "epoch": 0.15875842768095663, + "ewc_loss": 0.004150390625, + "ewc_loss_diag": 3.7848949432373047e-06, + "ewc_loss_parallel": 3.769993782043457e-06, + "grad_norm": 5.339509010314941, + "learning_rate": 5.286138194150064e-07, + "loss": 0.3962, + "mean_token_accuracy": 0.8697504997253418, + "num_tokens": 48410951.0, + "step": 1248 + }, + { + "epoch": 0.15888563795954713, + "ewc_loss": 0.004150390625, + "ewc_loss_diag": 3.7848949432373047e-06, + "ewc_loss_parallel": 3.769993782043457e-06, + "grad_norm": 5.457839012145996, + "learning_rate": 5.290377278507841e-07, + "loss": 0.4356, + "mean_token_accuracy": 0.8581626415252686, + "num_tokens": 48451770.0, + "step": 1249 + }, + { + "epoch": 0.15901284823813763, + "ewc_loss": 0.004180908203125, + "ewc_loss_diag": 3.7848949432373047e-06, + "ewc_loss_parallel": 3.814697265625e-06, + "grad_norm": 5.413461208343506, + "learning_rate": 5.294616362865621e-07, + "loss": 0.3887, + "mean_token_accuracy": 0.8706813454627991, + "num_tokens": 48494363.0, + "step": 1250 + }, + { + "epoch": 0.15914005851672816, + "ewc_loss": 0.004150390625, + "ewc_loss_diag": 3.814697265625e-06, + "ewc_loss_parallel": 3.7848949432373047e-06, + "grad_norm": 5.535251140594482, + "learning_rate": 5.298855447223399e-07, + "loss": 0.4703, + "mean_token_accuracy": 0.8505991697311401, + "num_tokens": 48537057.0, + "step": 1251 + }, + { + "epoch": 0.15926726879531866, + "ewc_loss": 0.00421142578125, + "ewc_loss_diag": 3.844499588012695e-06, + "ewc_loss_parallel": 3.814697265625e-06, + "grad_norm": 5.519660949707031, + "learning_rate": 5.303094531581178e-07, + "loss": 0.4117, + "mean_token_accuracy": 0.860734224319458, + "num_tokens": 48573548.0, + "step": 1252 + }, + { + "epoch": 0.15939447907390916, + "ewc_loss": 0.004180908203125, + "ewc_loss_diag": 3.844499588012695e-06, + "ewc_loss_parallel": 3.7997961044311523e-06, + "grad_norm": 5.423137187957764, + "learning_rate": 5.307333615938957e-07, + "loss": 0.3635, + "mean_token_accuracy": 0.8772953748703003, + "num_tokens": 48611667.0, + "step": 1253 + }, + { + "epoch": 0.1595216893524997, + "ewc_loss": 0.004180908203125, + "ewc_loss_diag": 3.844499588012695e-06, + "ewc_loss_parallel": 3.7848949432373047e-06, + "grad_norm": 5.6726908683776855, + "learning_rate": 5.311572700296736e-07, + "loss": 0.4275, + "mean_token_accuracy": 0.8554067611694336, + "num_tokens": 48644476.0, + "step": 1254 + }, + { + "epoch": 0.1596488996310902, + "ewc_loss": 0.0042724609375, + "ewc_loss_diag": 3.844499588012695e-06, + "ewc_loss_parallel": 3.874301910400391e-06, + "grad_norm": 5.5254621505737305, + "learning_rate": 5.315811784654515e-07, + "loss": 0.3484, + "mean_token_accuracy": 0.8774720430374146, + "num_tokens": 48681495.0, + "step": 1255 + }, + { + "epoch": 0.1597761099096807, + "ewc_loss": 0.00421142578125, + "ewc_loss_diag": 3.844499588012695e-06, + "ewc_loss_parallel": 3.814697265625e-06, + "grad_norm": 5.290371417999268, + "learning_rate": 5.320050869012294e-07, + "loss": 0.4034, + "mean_token_accuracy": 0.8682247400283813, + "num_tokens": 48728326.0, + "step": 1256 + }, + { + "epoch": 0.15990332018827122, + "ewc_loss": 0.004150390625, + "ewc_loss_diag": 3.874301910400391e-06, + "ewc_loss_parallel": 3.7550926208496094e-06, + "grad_norm": 5.4976677894592285, + "learning_rate": 5.324289953370071e-07, + "loss": 0.4606, + "mean_token_accuracy": 0.8482722640037537, + "num_tokens": 48769898.0, + "step": 1257 + }, + { + "epoch": 0.16003053046686172, + "ewc_loss": 0.0042724609375, + "ewc_loss_diag": 3.874301910400391e-06, + "ewc_loss_parallel": 3.874301910400391e-06, + "grad_norm": 5.459421634674072, + "learning_rate": 5.328529037727851e-07, + "loss": 0.4178, + "mean_token_accuracy": 0.8613017797470093, + "num_tokens": 48813411.0, + "step": 1258 + }, + { + "epoch": 0.16015774074545222, + "ewc_loss": 0.00421142578125, + "ewc_loss_diag": 3.874301910400391e-06, + "ewc_loss_parallel": 3.814697265625e-06, + "grad_norm": 5.454013824462891, + "learning_rate": 5.332768122085629e-07, + "loss": 0.3962, + "mean_token_accuracy": 0.8654389381408691, + "num_tokens": 48852067.0, + "step": 1259 + }, + { + "epoch": 0.16028495102404275, + "ewc_loss": 0.00421142578125, + "ewc_loss_diag": 3.874301910400391e-06, + "ewc_loss_parallel": 3.814697265625e-06, + "grad_norm": 5.5259318351745605, + "learning_rate": 5.337007206443408e-07, + "loss": 0.4208, + "mean_token_accuracy": 0.8590078353881836, + "num_tokens": 48890411.0, + "step": 1260 + }, + { + "epoch": 0.16041216130263325, + "ewc_loss": 0.00421142578125, + "ewc_loss_diag": 3.874301910400391e-06, + "ewc_loss_parallel": 3.814697265625e-06, + "grad_norm": 5.466753959655762, + "learning_rate": 5.341246290801187e-07, + "loss": 0.4043, + "mean_token_accuracy": 0.8659681677818298, + "num_tokens": 48932184.0, + "step": 1261 + }, + { + "epoch": 0.16053937158122375, + "ewc_loss": 0.00421142578125, + "ewc_loss_diag": 3.874301910400391e-06, + "ewc_loss_parallel": 3.814697265625e-06, + "grad_norm": 5.392439365386963, + "learning_rate": 5.345485375158966e-07, + "loss": 0.3854, + "mean_token_accuracy": 0.8705707788467407, + "num_tokens": 48973073.0, + "step": 1262 + }, + { + "epoch": 0.16066658185981428, + "ewc_loss": 0.004180908203125, + "ewc_loss_diag": 3.874301910400391e-06, + "ewc_loss_parallel": 3.7848949432373047e-06, + "grad_norm": 5.578096389770508, + "learning_rate": 5.349724459516745e-07, + "loss": 0.3791, + "mean_token_accuracy": 0.8707347512245178, + "num_tokens": 49010490.0, + "step": 1263 + }, + { + "epoch": 0.16079379213840478, + "ewc_loss": 0.004241943359375, + "ewc_loss_diag": 3.874301910400391e-06, + "ewc_loss_parallel": 3.844499588012695e-06, + "grad_norm": 5.447137355804443, + "learning_rate": 5.353963543874522e-07, + "loss": 0.4275, + "mean_token_accuracy": 0.8569965362548828, + "num_tokens": 49053887.0, + "step": 1264 + }, + { + "epoch": 0.1609210024169953, + "ewc_loss": 0.004180908203125, + "ewc_loss_diag": 3.874301910400391e-06, + "ewc_loss_parallel": 3.7848949432373047e-06, + "grad_norm": 5.686525344848633, + "learning_rate": 5.358202628232301e-07, + "loss": 0.4242, + "mean_token_accuracy": 0.8586680889129639, + "num_tokens": 49084000.0, + "step": 1265 + }, + { + "epoch": 0.1610482126955858, + "ewc_loss": 0.0042724609375, + "ewc_loss_diag": 3.874301910400391e-06, + "ewc_loss_parallel": 3.874301910400391e-06, + "grad_norm": 5.513055801391602, + "learning_rate": 5.36244171259008e-07, + "loss": 0.4008, + "mean_token_accuracy": 0.8627262711524963, + "num_tokens": 49124500.0, + "step": 1266 + }, + { + "epoch": 0.1611754229741763, + "ewc_loss": 0.00421142578125, + "ewc_loss_diag": 3.874301910400391e-06, + "ewc_loss_parallel": 3.814697265625e-06, + "grad_norm": 5.494560241699219, + "learning_rate": 5.366680796947859e-07, + "loss": 0.442, + "mean_token_accuracy": 0.8550163507461548, + "num_tokens": 49165521.0, + "step": 1267 + }, + { + "epoch": 0.16130263325276684, + "ewc_loss": 0.00421142578125, + "ewc_loss_diag": 3.874301910400391e-06, + "ewc_loss_parallel": 3.814697265625e-06, + "grad_norm": 5.849645137786865, + "learning_rate": 5.370919881305637e-07, + "loss": 0.4299, + "mean_token_accuracy": 0.8593519330024719, + "num_tokens": 49200805.0, + "step": 1268 + }, + { + "epoch": 0.16142984353135734, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.567481994628906, + "learning_rate": 5.375158965663417e-07, + "loss": 0.4014, + "mean_token_accuracy": 0.8674227595329285, + "num_tokens": 49237384.0, + "step": 1269 + }, + { + "epoch": 0.16155705380994784, + "ewc_loss": 0.004241943359375, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.844499588012695e-06, + "grad_norm": 5.457461833953857, + "learning_rate": 5.379398050021195e-07, + "loss": 0.4125, + "mean_token_accuracy": 0.8619875907897949, + "num_tokens": 49277811.0, + "step": 1270 + }, + { + "epoch": 0.16168426408853837, + "ewc_loss": 0.00421142578125, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.814697265625e-06, + "grad_norm": 5.6397480964660645, + "learning_rate": 5.383637134378975e-07, + "loss": 0.4001, + "mean_token_accuracy": 0.8655428886413574, + "num_tokens": 49312209.0, + "step": 1271 + }, + { + "epoch": 0.16181147436712887, + "ewc_loss": 0.004302978515625, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.904104232788086e-06, + "grad_norm": 5.446821689605713, + "learning_rate": 5.387876218736752e-07, + "loss": 0.4712, + "mean_token_accuracy": 0.8431539535522461, + "num_tokens": 49358471.0, + "step": 1272 + }, + { + "epoch": 0.16193868464571937, + "ewc_loss": 0.00421142578125, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.814697265625e-06, + "grad_norm": 5.397977828979492, + "learning_rate": 5.392115303094531e-07, + "loss": 0.3705, + "mean_token_accuracy": 0.8741310834884644, + "num_tokens": 49398851.0, + "step": 1273 + }, + { + "epoch": 0.1620658949243099, + "ewc_loss": 0.0042724609375, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.874301910400391e-06, + "grad_norm": 5.417270183563232, + "learning_rate": 5.39635438745231e-07, + "loss": 0.3476, + "mean_token_accuracy": 0.882807731628418, + "num_tokens": 49438702.0, + "step": 1274 + }, + { + "epoch": 0.1621931052029004, + "ewc_loss": 0.0042724609375, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.874301910400391e-06, + "grad_norm": 5.522036552429199, + "learning_rate": 5.400593471810089e-07, + "loss": 0.4422, + "mean_token_accuracy": 0.8528294563293457, + "num_tokens": 49478650.0, + "step": 1275 + }, + { + "epoch": 0.1623203154814909, + "ewc_loss": 0.004302978515625, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.904104232788086e-06, + "grad_norm": 5.749590873718262, + "learning_rate": 5.404832556167867e-07, + "loss": 0.434, + "mean_token_accuracy": 0.8550882935523987, + "num_tokens": 49514303.0, + "step": 1276 + }, + { + "epoch": 0.16244752576008142, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.4662322998046875, + "learning_rate": 5.409071640525647e-07, + "loss": 0.4086, + "mean_token_accuracy": 0.8643248081207275, + "num_tokens": 49554670.0, + "step": 1277 + }, + { + "epoch": 0.16257473603867192, + "ewc_loss": 0.004241943359375, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.844499588012695e-06, + "grad_norm": 5.780008316040039, + "learning_rate": 5.413310724883425e-07, + "loss": 0.4478, + "mean_token_accuracy": 0.8451180458068848, + "num_tokens": 49593683.0, + "step": 1278 + }, + { + "epoch": 0.16270194631726242, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.548991680145264, + "learning_rate": 5.417549809241205e-07, + "loss": 0.4365, + "mean_token_accuracy": 0.8576315641403198, + "num_tokens": 49638252.0, + "step": 1279 + }, + { + "epoch": 0.16282915659585295, + "ewc_loss": 0.0042724609375, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.874301910400391e-06, + "grad_norm": 5.763314247131348, + "learning_rate": 5.421788893598982e-07, + "loss": 0.4425, + "mean_token_accuracy": 0.8543983697891235, + "num_tokens": 49673420.0, + "step": 1280 + }, + { + "epoch": 0.16295636687444345, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.493225574493408, + "learning_rate": 5.42602797795676e-07, + "loss": 0.4542, + "mean_token_accuracy": 0.8493114709854126, + "num_tokens": 49719017.0, + "step": 1281 + }, + { + "epoch": 0.16308357715303395, + "ewc_loss": 0.004241943359375, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.844499588012695e-06, + "grad_norm": 5.485166549682617, + "learning_rate": 5.43026706231454e-07, + "loss": 0.4524, + "mean_token_accuracy": 0.851003885269165, + "num_tokens": 49760970.0, + "step": 1282 + }, + { + "epoch": 0.16321078743162448, + "ewc_loss": 0.0042724609375, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.874301910400391e-06, + "grad_norm": 5.429091930389404, + "learning_rate": 5.434506146672319e-07, + "loss": 0.4429, + "mean_token_accuracy": 0.8538159728050232, + "num_tokens": 49811263.0, + "step": 1283 + }, + { + "epoch": 0.16333799771021498, + "ewc_loss": 0.004241943359375, + "ewc_loss_diag": 3.904104232788086e-06, + "ewc_loss_parallel": 3.844499588012695e-06, + "grad_norm": 5.633678913116455, + "learning_rate": 5.438745231030097e-07, + "loss": 0.4409, + "mean_token_accuracy": 0.8515928983688354, + "num_tokens": 49847981.0, + "step": 1284 + }, + { + "epoch": 0.16346520798880548, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.933906555175781e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.578510761260986, + "learning_rate": 5.442984315387876e-07, + "loss": 0.356, + "mean_token_accuracy": 0.8769721984863281, + "num_tokens": 49883076.0, + "step": 1285 + }, + { + "epoch": 0.163592418267396, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.933906555175781e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.453137397766113, + "learning_rate": 5.447223399745655e-07, + "loss": 0.4004, + "mean_token_accuracy": 0.8683843612670898, + "num_tokens": 49923512.0, + "step": 1286 + }, + { + "epoch": 0.1637196285459865, + "ewc_loss": 0.004302978515625, + "ewc_loss_diag": 3.9637088775634766e-06, + "ewc_loss_parallel": 3.904104232788086e-06, + "grad_norm": 5.647913932800293, + "learning_rate": 5.451462484103433e-07, + "loss": 0.4024, + "mean_token_accuracy": 0.8645035624504089, + "num_tokens": 49958376.0, + "step": 1287 + }, + { + "epoch": 0.163846838824577, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.9637088775634766e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.5549211502075195, + "learning_rate": 5.455701568461212e-07, + "loss": 0.405, + "mean_token_accuracy": 0.8642697930335999, + "num_tokens": 49993657.0, + "step": 1288 + }, + { + "epoch": 0.16397404910316754, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.9637088775634766e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.640716552734375, + "learning_rate": 5.45994065281899e-07, + "loss": 0.472, + "mean_token_accuracy": 0.8434789776802063, + "num_tokens": 50035333.0, + "step": 1289 + }, + { + "epoch": 0.16410125938175804, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.9637088775634766e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.821783065795898, + "learning_rate": 5.46417973717677e-07, + "loss": 0.4248, + "mean_token_accuracy": 0.8576178550720215, + "num_tokens": 50065573.0, + "step": 1290 + }, + { + "epoch": 0.16422846966034857, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 3.9637088775634766e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.720909595489502, + "learning_rate": 5.468418821534548e-07, + "loss": 0.369, + "mean_token_accuracy": 0.8750795125961304, + "num_tokens": 50097891.0, + "step": 1291 + }, + { + "epoch": 0.16435567993893907, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.9637088775634766e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.624051570892334, + "learning_rate": 5.472657905892327e-07, + "loss": 0.3779, + "mean_token_accuracy": 0.873689591884613, + "num_tokens": 50131383.0, + "step": 1292 + }, + { + "epoch": 0.16448289021752957, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.9637088775634766e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.570912837982178, + "learning_rate": 5.476896990250106e-07, + "loss": 0.4773, + "mean_token_accuracy": 0.8429128527641296, + "num_tokens": 50172175.0, + "step": 1293 + }, + { + "epoch": 0.1646101004961201, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.9637088775634766e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.41987419128418, + "learning_rate": 5.481136074607885e-07, + "loss": 0.3668, + "mean_token_accuracy": 0.8725517988204956, + "num_tokens": 50211235.0, + "step": 1294 + }, + { + "epoch": 0.1647373107747106, + "ewc_loss": 0.004302978515625, + "ewc_loss_diag": 3.9637088775634766e-06, + "ewc_loss_parallel": 3.904104232788086e-06, + "grad_norm": 5.3726630210876465, + "learning_rate": 5.485375158965663e-07, + "loss": 0.3764, + "mean_token_accuracy": 0.8750265836715698, + "num_tokens": 50257666.0, + "step": 1295 + }, + { + "epoch": 0.1648645210533011, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.9637088775634766e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.620680332183838, + "learning_rate": 5.489614243323442e-07, + "loss": 0.421, + "mean_token_accuracy": 0.8576565980911255, + "num_tokens": 50296666.0, + "step": 1296 + }, + { + "epoch": 0.16499173133189163, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 3.9637088775634766e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.494966506958008, + "learning_rate": 5.49385332768122e-07, + "loss": 0.3564, + "mean_token_accuracy": 0.8777389526367188, + "num_tokens": 50335628.0, + "step": 1297 + }, + { + "epoch": 0.16511894161048213, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.9637088775634766e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.628222942352295, + "learning_rate": 5.498092412039e-07, + "loss": 0.4639, + "mean_token_accuracy": 0.8495025634765625, + "num_tokens": 50372348.0, + "step": 1298 + }, + { + "epoch": 0.16524615188907263, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 3.9637088775634766e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.366699695587158, + "learning_rate": 5.502331496396778e-07, + "loss": 0.3915, + "mean_token_accuracy": 0.866572916507721, + "num_tokens": 50417252.0, + "step": 1299 + }, + { + "epoch": 0.16537336216766316, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 3.993511199951172e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.509390830993652, + "learning_rate": 5.506570580754557e-07, + "loss": 0.4067, + "mean_token_accuracy": 0.8628050088882446, + "num_tokens": 50457348.0, + "step": 1300 + }, + { + "epoch": 0.16550057244625366, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 4.023313522338867e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.611859321594238, + "learning_rate": 5.510809665112336e-07, + "loss": 0.3861, + "mean_token_accuracy": 0.8701079487800598, + "num_tokens": 50497327.0, + "step": 1301 + }, + { + "epoch": 0.16562778272484416, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 4.023313522338867e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.484340667724609, + "learning_rate": 5.515048749470113e-07, + "loss": 0.4037, + "mean_token_accuracy": 0.866899847984314, + "num_tokens": 50537301.0, + "step": 1302 + }, + { + "epoch": 0.1657549930034347, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 4.023313522338867e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.514309406280518, + "learning_rate": 5.519287833827893e-07, + "loss": 0.3935, + "mean_token_accuracy": 0.8678632378578186, + "num_tokens": 50576684.0, + "step": 1303 + }, + { + "epoch": 0.1658822032820252, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 4.023313522338867e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.7091546058654785, + "learning_rate": 5.523526918185671e-07, + "loss": 0.4239, + "mean_token_accuracy": 0.8558809757232666, + "num_tokens": 50610885.0, + "step": 1304 + }, + { + "epoch": 0.1660094135606157, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 4.023313522338867e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.639639854431152, + "learning_rate": 5.52776600254345e-07, + "loss": 0.3797, + "mean_token_accuracy": 0.8680287003517151, + "num_tokens": 50642725.0, + "step": 1305 + }, + { + "epoch": 0.16613662383920622, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 4.023313522338867e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.50448751449585, + "learning_rate": 5.532005086901229e-07, + "loss": 0.4262, + "mean_token_accuracy": 0.8620092868804932, + "num_tokens": 50682338.0, + "step": 1306 + }, + { + "epoch": 0.16626383411779672, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 4.023313522338867e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.456480979919434, + "learning_rate": 5.536244171259008e-07, + "loss": 0.4008, + "mean_token_accuracy": 0.8651888370513916, + "num_tokens": 50724981.0, + "step": 1307 + }, + { + "epoch": 0.16639104439638722, + "ewc_loss": 0.00433349609375, + "ewc_loss_diag": 4.023313522338867e-06, + "ewc_loss_parallel": 3.933906555175781e-06, + "grad_norm": 5.909828186035156, + "learning_rate": 5.540483255616786e-07, + "loss": 0.4914, + "mean_token_accuracy": 0.8379002213478088, + "num_tokens": 50760164.0, + "step": 1308 + }, + { + "epoch": 0.16651825467497774, + "ewc_loss": 0.004425048828125, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 4.023313522338867e-06, + "grad_norm": 5.624926567077637, + "learning_rate": 5.544722339974566e-07, + "loss": 0.4125, + "mean_token_accuracy": 0.8656114339828491, + "num_tokens": 50794585.0, + "step": 1309 + }, + { + "epoch": 0.16664546495356825, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.57004451751709, + "learning_rate": 5.548961424332343e-07, + "loss": 0.4067, + "mean_token_accuracy": 0.8597183227539062, + "num_tokens": 50830753.0, + "step": 1310 + }, + { + "epoch": 0.16677267523215875, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.50148868560791, + "learning_rate": 5.553200508690123e-07, + "loss": 0.3669, + "mean_token_accuracy": 0.8764421939849854, + "num_tokens": 50871805.0, + "step": 1311 + }, + { + "epoch": 0.16689988551074927, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.641643524169922, + "learning_rate": 5.557439593047901e-07, + "loss": 0.4564, + "mean_token_accuracy": 0.8481355905532837, + "num_tokens": 50907805.0, + "step": 1312 + }, + { + "epoch": 0.16702709578933977, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.5695977210998535, + "learning_rate": 5.56167867740568e-07, + "loss": 0.4135, + "mean_token_accuracy": 0.8624765872955322, + "num_tokens": 50951876.0, + "step": 1313 + }, + { + "epoch": 0.16715430606793028, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.653648376464844, + "learning_rate": 5.565917761763459e-07, + "loss": 0.3881, + "mean_token_accuracy": 0.8699598908424377, + "num_tokens": 50983619.0, + "step": 1314 + }, + { + "epoch": 0.1672815163465208, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.599095344543457, + "learning_rate": 5.570156846121238e-07, + "loss": 0.4223, + "mean_token_accuracy": 0.8567419052124023, + "num_tokens": 51022575.0, + "step": 1315 + }, + { + "epoch": 0.1674087266251113, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.592724800109863, + "learning_rate": 5.574395930479016e-07, + "loss": 0.4293, + "mean_token_accuracy": 0.858370304107666, + "num_tokens": 51060466.0, + "step": 1316 + }, + { + "epoch": 0.16753593690370183, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.667656898498535, + "learning_rate": 5.578635014836796e-07, + "loss": 0.3951, + "mean_token_accuracy": 0.8676173686981201, + "num_tokens": 51095089.0, + "step": 1317 + }, + { + "epoch": 0.16766314718229233, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.792850971221924, + "learning_rate": 5.582874099194573e-07, + "loss": 0.4459, + "mean_token_accuracy": 0.8536614179611206, + "num_tokens": 51127804.0, + "step": 1318 + }, + { + "epoch": 0.16779035746088283, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.670791149139404, + "learning_rate": 5.587113183552353e-07, + "loss": 0.4367, + "mean_token_accuracy": 0.8557820320129395, + "num_tokens": 51167705.0, + "step": 1319 + }, + { + "epoch": 0.16791756773947336, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.463301181793213, + "learning_rate": 5.591352267910131e-07, + "loss": 0.3932, + "mean_token_accuracy": 0.8701624870300293, + "num_tokens": 51209787.0, + "step": 1320 + }, + { + "epoch": 0.16804477801806386, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.662689208984375, + "learning_rate": 5.59559135226791e-07, + "loss": 0.4296, + "mean_token_accuracy": 0.8598019480705261, + "num_tokens": 51249538.0, + "step": 1321 + }, + { + "epoch": 0.16817198829665436, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.609524726867676, + "learning_rate": 5.599830436625689e-07, + "loss": 0.3939, + "mean_token_accuracy": 0.8685940504074097, + "num_tokens": 51288594.0, + "step": 1322 + }, + { + "epoch": 0.1682991985752449, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 6.097485542297363, + "learning_rate": 5.604069520983468e-07, + "loss": 0.4063, + "mean_token_accuracy": 0.8652812242507935, + "num_tokens": 51326269.0, + "step": 1323 + }, + { + "epoch": 0.1684264088538354, + "ewc_loss": 0.0045166015625, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 4.112720489501953e-06, + "grad_norm": 5.741429805755615, + "learning_rate": 5.608308605341246e-07, + "loss": 0.3895, + "mean_token_accuracy": 0.8689043521881104, + "num_tokens": 51360513.0, + "step": 1324 + }, + { + "epoch": 0.1685536191324259, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.62581729888916, + "learning_rate": 5.612547689699024e-07, + "loss": 0.4505, + "mean_token_accuracy": 0.8518030643463135, + "num_tokens": 51396097.0, + "step": 1325 + }, + { + "epoch": 0.16868082941101642, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.6155900955200195, + "learning_rate": 5.616786774056803e-07, + "loss": 0.428, + "mean_token_accuracy": 0.8582934737205505, + "num_tokens": 51437078.0, + "step": 1326 + }, + { + "epoch": 0.16880803968960692, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.469114303588867, + "learning_rate": 5.621025858414582e-07, + "loss": 0.4333, + "mean_token_accuracy": 0.8553466796875, + "num_tokens": 51481017.0, + "step": 1327 + }, + { + "epoch": 0.16893524996819742, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.511163234710693, + "learning_rate": 5.625264942772361e-07, + "loss": 0.4247, + "mean_token_accuracy": 0.8618189096450806, + "num_tokens": 51522092.0, + "step": 1328 + }, + { + "epoch": 0.16906246024678795, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.0531158447265625e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.63370418548584, + "learning_rate": 5.629504027130139e-07, + "loss": 0.4748, + "mean_token_accuracy": 0.8500489592552185, + "num_tokens": 51562534.0, + "step": 1329 + }, + { + "epoch": 0.16918967052537845, + "ewc_loss": 0.004425048828125, + "ewc_loss_diag": 4.082918167114258e-06, + "ewc_loss_parallel": 4.023313522338867e-06, + "grad_norm": 6.526887893676758, + "learning_rate": 5.633743111487919e-07, + "loss": 0.4758, + "mean_token_accuracy": 0.8402534127235413, + "num_tokens": 51602724.0, + "step": 1330 + }, + { + "epoch": 0.16931688080396895, + "ewc_loss": 0.004730224609375, + "ewc_loss_diag": 4.082918167114258e-06, + "ewc_loss_parallel": 4.32133674621582e-06, + "grad_norm": 5.703802585601807, + "learning_rate": 5.637982195845697e-07, + "loss": 0.3612, + "mean_token_accuracy": 0.8800860643386841, + "num_tokens": 51638837.0, + "step": 1331 + }, + { + "epoch": 0.16944409108255948, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.082918167114258e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.464012145996094, + "learning_rate": 5.642221280203476e-07, + "loss": 0.3628, + "mean_token_accuracy": 0.8795385956764221, + "num_tokens": 51675627.0, + "step": 1332 + }, + { + "epoch": 0.16957130136114998, + "ewc_loss": 0.004364013671875, + "ewc_loss_diag": 4.082918167114258e-06, + "ewc_loss_parallel": 3.9637088775634766e-06, + "grad_norm": 5.856522083282471, + "learning_rate": 5.646460364561254e-07, + "loss": 0.3771, + "mean_token_accuracy": 0.8723692297935486, + "num_tokens": 51707716.0, + "step": 1333 + }, + { + "epoch": 0.16969851163974048, + "ewc_loss": 0.004486083984375, + "ewc_loss_diag": 4.082918167114258e-06, + "ewc_loss_parallel": 4.082918167114258e-06, + "grad_norm": 5.69240140914917, + "learning_rate": 5.650699448919033e-07, + "loss": 0.4675, + "mean_token_accuracy": 0.8456375598907471, + "num_tokens": 51751365.0, + "step": 1334 + }, + { + "epoch": 0.169825721918331, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.082918167114258e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.553450584411621, + "learning_rate": 5.654938533276812e-07, + "loss": 0.4068, + "mean_token_accuracy": 0.8634171485900879, + "num_tokens": 51792787.0, + "step": 1335 + }, + { + "epoch": 0.1699529321969215, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.082918167114258e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.740826606750488, + "learning_rate": 5.659177617634591e-07, + "loss": 0.4006, + "mean_token_accuracy": 0.8670347332954407, + "num_tokens": 51830627.0, + "step": 1336 + }, + { + "epoch": 0.170080142475512, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.082918167114258e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.5965375900268555, + "learning_rate": 5.663416701992369e-07, + "loss": 0.4103, + "mean_token_accuracy": 0.8609559535980225, + "num_tokens": 51870206.0, + "step": 1337 + }, + { + "epoch": 0.17020735275410254, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.082918167114258e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.787059783935547, + "learning_rate": 5.667655786350149e-07, + "loss": 0.4699, + "mean_token_accuracy": 0.8393328189849854, + "num_tokens": 51904420.0, + "step": 1338 + }, + { + "epoch": 0.17033456303269304, + "ewc_loss": 0.004425048828125, + "ewc_loss_diag": 4.082918167114258e-06, + "ewc_loss_parallel": 4.023313522338867e-06, + "grad_norm": 5.580832004547119, + "learning_rate": 5.671894870707927e-07, + "loss": 0.403, + "mean_token_accuracy": 0.8663880228996277, + "num_tokens": 51946834.0, + "step": 1339 + }, + { + "epoch": 0.17046177331128357, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.112720489501953e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.626509189605713, + "learning_rate": 5.676133955065705e-07, + "loss": 0.362, + "mean_token_accuracy": 0.8770109415054321, + "num_tokens": 51981576.0, + "step": 1340 + }, + { + "epoch": 0.17058898358987407, + "ewc_loss": 0.00445556640625, + "ewc_loss_diag": 4.112720489501953e-06, + "ewc_loss_parallel": 4.023313522338867e-06, + "grad_norm": 5.613104343414307, + "learning_rate": 5.680373039423484e-07, + "loss": 0.3929, + "mean_token_accuracy": 0.8700482845306396, + "num_tokens": 52021764.0, + "step": 1341 + }, + { + "epoch": 0.17071619386846457, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.112720489501953e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.7072014808654785, + "learning_rate": 5.684612123781263e-07, + "loss": 0.4353, + "mean_token_accuracy": 0.8517237901687622, + "num_tokens": 52063410.0, + "step": 1342 + }, + { + "epoch": 0.1708434041470551, + "ewc_loss": 0.00445556640625, + "ewc_loss_diag": 4.112720489501953e-06, + "ewc_loss_parallel": 4.023313522338867e-06, + "grad_norm": 5.646884918212891, + "learning_rate": 5.688851208139042e-07, + "loss": 0.443, + "mean_token_accuracy": 0.8518074750900269, + "num_tokens": 52103420.0, + "step": 1343 + }, + { + "epoch": 0.1709706144256456, + "ewc_loss": 0.00445556640625, + "ewc_loss_diag": 4.112720489501953e-06, + "ewc_loss_parallel": 4.023313522338867e-06, + "grad_norm": 5.694693565368652, + "learning_rate": 5.69309029249682e-07, + "loss": 0.4186, + "mean_token_accuracy": 0.8634054660797119, + "num_tokens": 52147688.0, + "step": 1344 + }, + { + "epoch": 0.1710978247042361, + "ewc_loss": 0.00445556640625, + "ewc_loss_diag": 4.112720489501953e-06, + "ewc_loss_parallel": 4.023313522338867e-06, + "grad_norm": 5.674103260040283, + "learning_rate": 5.697329376854599e-07, + "loss": 0.4227, + "mean_token_accuracy": 0.8589730858802795, + "num_tokens": 52187023.0, + "step": 1345 + }, + { + "epoch": 0.17122503498282662, + "ewc_loss": 0.00439453125, + "ewc_loss_diag": 4.112720489501953e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.639636993408203, + "learning_rate": 5.701568461212378e-07, + "loss": 0.3732, + "mean_token_accuracy": 0.8752120733261108, + "num_tokens": 52223242.0, + "step": 1346 + }, + { + "epoch": 0.17135224526141712, + "ewc_loss": 0.004425048828125, + "ewc_loss_diag": 4.1425228118896484e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.501419544219971, + "learning_rate": 5.705807545570157e-07, + "loss": 0.3669, + "mean_token_accuracy": 0.8752079010009766, + "num_tokens": 52266141.0, + "step": 1347 + }, + { + "epoch": 0.17147945554000762, + "ewc_loss": 0.004425048828125, + "ewc_loss_diag": 4.1425228118896484e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.617543697357178, + "learning_rate": 5.710046629927934e-07, + "loss": 0.402, + "mean_token_accuracy": 0.8650626540184021, + "num_tokens": 52305892.0, + "step": 1348 + }, + { + "epoch": 0.17160666581859815, + "ewc_loss": 0.00445556640625, + "ewc_loss_diag": 4.1425228118896484e-06, + "ewc_loss_parallel": 4.023313522338867e-06, + "grad_norm": 5.646702766418457, + "learning_rate": 5.714285714285714e-07, + "loss": 0.3797, + "mean_token_accuracy": 0.8735792636871338, + "num_tokens": 52343585.0, + "step": 1349 + }, + { + "epoch": 0.17173387609718865, + "ewc_loss": 0.00445556640625, + "ewc_loss_diag": 4.1425228118896484e-06, + "ewc_loss_parallel": 4.023313522338867e-06, + "grad_norm": 5.867924213409424, + "learning_rate": 5.718524798643492e-07, + "loss": 0.4498, + "mean_token_accuracy": 0.8520172238349915, + "num_tokens": 52378114.0, + "step": 1350 + }, + { + "epoch": 0.17186108637577915, + "ewc_loss": 0.0045166015625, + "ewc_loss_diag": 4.172325134277344e-06, + "ewc_loss_parallel": 4.082918167114258e-06, + "grad_norm": 5.644198417663574, + "learning_rate": 5.722763883001272e-07, + "loss": 0.4389, + "mean_token_accuracy": 0.8542476892471313, + "num_tokens": 52419976.0, + "step": 1351 + }, + { + "epoch": 0.17198829665436968, + "ewc_loss": 0.004425048828125, + "ewc_loss_diag": 4.172325134277344e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.619404315948486, + "learning_rate": 5.72700296735905e-07, + "loss": 0.3947, + "mean_token_accuracy": 0.8678842782974243, + "num_tokens": 52458125.0, + "step": 1352 + }, + { + "epoch": 0.17211550693296018, + "ewc_loss": 0.00445556640625, + "ewc_loss_diag": 4.202127456665039e-06, + "ewc_loss_parallel": 4.023313522338867e-06, + "grad_norm": 5.649077415466309, + "learning_rate": 5.731242051716829e-07, + "loss": 0.4556, + "mean_token_accuracy": 0.8491179943084717, + "num_tokens": 52501643.0, + "step": 1353 + }, + { + "epoch": 0.17224271721155068, + "ewc_loss": 0.00445556640625, + "ewc_loss_diag": 4.202127456665039e-06, + "ewc_loss_parallel": 4.023313522338867e-06, + "grad_norm": 5.674312114715576, + "learning_rate": 5.735481136074608e-07, + "loss": 0.3741, + "mean_token_accuracy": 0.8729195594787598, + "num_tokens": 52537761.0, + "step": 1354 + }, + { + "epoch": 0.1723699274901412, + "ewc_loss": 0.004486083984375, + "ewc_loss_diag": 4.202127456665039e-06, + "ewc_loss_parallel": 4.0531158447265625e-06, + "grad_norm": 5.631596088409424, + "learning_rate": 5.739720220432386e-07, + "loss": 0.387, + "mean_token_accuracy": 0.8674290180206299, + "num_tokens": 52574678.0, + "step": 1355 + }, + { + "epoch": 0.1724971377687317, + "ewc_loss": 0.004486083984375, + "ewc_loss_diag": 4.202127456665039e-06, + "ewc_loss_parallel": 4.0531158447265625e-06, + "grad_norm": 6.851318359375, + "learning_rate": 5.743959304790164e-07, + "loss": 0.4569, + "mean_token_accuracy": 0.8513409495353699, + "num_tokens": 52617181.0, + "step": 1356 + }, + { + "epoch": 0.1726243480473222, + "ewc_loss": 0.004852294921875, + "ewc_loss_diag": 4.202127456665039e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.8185224533081055, + "learning_rate": 5.748198389147944e-07, + "loss": 0.4238, + "mean_token_accuracy": 0.8599011898040771, + "num_tokens": 52656534.0, + "step": 1357 + }, + { + "epoch": 0.17275155832591274, + "ewc_loss": 0.004425048828125, + "ewc_loss_diag": 4.202127456665039e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.441849708557129, + "learning_rate": 5.752437473505722e-07, + "loss": 0.4347, + "mean_token_accuracy": 0.8566062450408936, + "num_tokens": 52700234.0, + "step": 1358 + }, + { + "epoch": 0.17287876860450324, + "ewc_loss": 0.004425048828125, + "ewc_loss_diag": 4.231929779052734e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.801082134246826, + "learning_rate": 5.756676557863502e-07, + "loss": 0.4384, + "mean_token_accuracy": 0.8554651737213135, + "num_tokens": 52737410.0, + "step": 1359 + }, + { + "epoch": 0.17300597888309374, + "ewc_loss": 0.004486083984375, + "ewc_loss_diag": 4.231929779052734e-06, + "ewc_loss_parallel": 4.0531158447265625e-06, + "grad_norm": 5.659875869750977, + "learning_rate": 5.76091564222128e-07, + "loss": 0.3866, + "mean_token_accuracy": 0.8717160224914551, + "num_tokens": 52776605.0, + "step": 1360 + }, + { + "epoch": 0.17313318916168427, + "ewc_loss": 0.004425048828125, + "ewc_loss_diag": 4.231929779052734e-06, + "ewc_loss_parallel": 3.993511199951172e-06, + "grad_norm": 5.654553413391113, + "learning_rate": 5.765154726579059e-07, + "loss": 0.4149, + "mean_token_accuracy": 0.8620029091835022, + "num_tokens": 52815314.0, + "step": 1361 + }, + { + "epoch": 0.17326039944027477, + "ewc_loss": 0.0045166015625, + "ewc_loss_diag": 4.231929779052734e-06, + "ewc_loss_parallel": 4.082918167114258e-06, + "grad_norm": 5.697988033294678, + "learning_rate": 5.769393810936838e-07, + "loss": 0.4086, + "mean_token_accuracy": 0.8647016286849976, + "num_tokens": 52852135.0, + "step": 1362 + }, + { + "epoch": 0.17338760971886527, + "ewc_loss": 0.0045166015625, + "ewc_loss_diag": 4.26173210144043e-06, + "ewc_loss_parallel": 4.082918167114258e-06, + "grad_norm": 5.61292028427124, + "learning_rate": 5.773632895294616e-07, + "loss": 0.448, + "mean_token_accuracy": 0.8487749099731445, + "num_tokens": 52894358.0, + "step": 1363 + }, + { + "epoch": 0.1735148199974558, + "ewc_loss": 0.004486083984375, + "ewc_loss_diag": 4.26173210144043e-06, + "ewc_loss_parallel": 4.0531158447265625e-06, + "grad_norm": 5.683330535888672, + "learning_rate": 5.777871979652394e-07, + "loss": 0.3892, + "mean_token_accuracy": 0.8677492737770081, + "num_tokens": 52931422.0, + "step": 1364 + }, + { + "epoch": 0.1736420302760463, + "ewc_loss": 0.0045166015625, + "ewc_loss_diag": 4.26173210144043e-06, + "ewc_loss_parallel": 4.082918167114258e-06, + "grad_norm": 5.579336166381836, + "learning_rate": 5.782111064010173e-07, + "loss": 0.4174, + "mean_token_accuracy": 0.858529806137085, + "num_tokens": 52976523.0, + "step": 1365 + }, + { + "epoch": 0.17376924055463683, + "ewc_loss": 0.004486083984375, + "ewc_loss_diag": 4.26173210144043e-06, + "ewc_loss_parallel": 4.0531158447265625e-06, + "grad_norm": 5.581609725952148, + "learning_rate": 5.786350148367952e-07, + "loss": 0.4153, + "mean_token_accuracy": 0.8606736660003662, + "num_tokens": 53022035.0, + "step": 1366 + }, + { + "epoch": 0.17389645083322733, + "ewc_loss": 0.0045166015625, + "ewc_loss_diag": 4.26173210144043e-06, + "ewc_loss_parallel": 4.082918167114258e-06, + "grad_norm": 5.644048690795898, + "learning_rate": 5.790589232725731e-07, + "loss": 0.4079, + "mean_token_accuracy": 0.8618965148925781, + "num_tokens": 53065157.0, + "step": 1367 + }, + { + "epoch": 0.17402366111181783, + "ewc_loss": 0.0045166015625, + "ewc_loss_diag": 4.26173210144043e-06, + "ewc_loss_parallel": 4.082918167114258e-06, + "grad_norm": 5.79850435256958, + "learning_rate": 5.79482831708351e-07, + "loss": 0.3971, + "mean_token_accuracy": 0.867176353931427, + "num_tokens": 53099456.0, + "step": 1368 + }, + { + "epoch": 0.17415087139040836, + "ewc_loss": 0.004547119140625, + "ewc_loss_diag": 4.26173210144043e-06, + "ewc_loss_parallel": 4.112720489501953e-06, + "grad_norm": 5.8899993896484375, + "learning_rate": 5.799067401441288e-07, + "loss": 0.4037, + "mean_token_accuracy": 0.8646914958953857, + "num_tokens": 53133761.0, + "step": 1369 + }, + { + "epoch": 0.17427808166899886, + "ewc_loss": 0.004547119140625, + "ewc_loss_diag": 4.26173210144043e-06, + "ewc_loss_parallel": 4.112720489501953e-06, + "grad_norm": 5.761342525482178, + "learning_rate": 5.803306485799068e-07, + "loss": 0.4026, + "mean_token_accuracy": 0.8662118911743164, + "num_tokens": 53169700.0, + "step": 1370 + }, + { + "epoch": 0.17440529194758936, + "ewc_loss": 0.004547119140625, + "ewc_loss_diag": 4.26173210144043e-06, + "ewc_loss_parallel": 4.112720489501953e-06, + "grad_norm": 5.620572566986084, + "learning_rate": 5.807545570156845e-07, + "loss": 0.3715, + "mean_token_accuracy": 0.8741968870162964, + "num_tokens": 53210957.0, + "step": 1371 + }, + { + "epoch": 0.1745325022261799, + "ewc_loss": 0.004486083984375, + "ewc_loss_diag": 4.26173210144043e-06, + "ewc_loss_parallel": 4.0531158447265625e-06, + "grad_norm": 5.872145175933838, + "learning_rate": 5.811784654514624e-07, + "loss": 0.454, + "mean_token_accuracy": 0.8497065901756287, + "num_tokens": 53244967.0, + "step": 1372 + }, + { + "epoch": 0.1746597125047704, + "ewc_loss": 0.004608154296875, + "ewc_loss_diag": 4.26173210144043e-06, + "ewc_loss_parallel": 4.172325134277344e-06, + "grad_norm": 5.64335823059082, + "learning_rate": 5.816023738872403e-07, + "loss": 0.4, + "mean_token_accuracy": 0.8683408498764038, + "num_tokens": 53283747.0, + "step": 1373 + }, + { + "epoch": 0.1747869227833609, + "ewc_loss": 0.0045166015625, + "ewc_loss_diag": 4.26173210144043e-06, + "ewc_loss_parallel": 4.082918167114258e-06, + "grad_norm": 5.574381351470947, + "learning_rate": 5.820262823230182e-07, + "loss": 0.412, + "mean_token_accuracy": 0.8615996837615967, + "num_tokens": 53327261.0, + "step": 1374 + }, + { + "epoch": 0.17491413306195142, + "ewc_loss": 0.0045166015625, + "ewc_loss_diag": 4.26173210144043e-06, + "ewc_loss_parallel": 4.082918167114258e-06, + "grad_norm": 5.820326805114746, + "learning_rate": 5.824501907587961e-07, + "loss": 0.3779, + "mean_token_accuracy": 0.8715764284133911, + "num_tokens": 53362964.0, + "step": 1375 + }, + { + "epoch": 0.17504134334054192, + "ewc_loss": 0.00457763671875, + "ewc_loss_diag": 4.32133674621582e-06, + "ewc_loss_parallel": 4.1425228118896484e-06, + "grad_norm": 5.739121913909912, + "learning_rate": 5.82874099194574e-07, + "loss": 0.4531, + "mean_token_accuracy": 0.8494696617126465, + "num_tokens": 53402236.0, + "step": 1376 + }, + { + "epoch": 0.17516855361913242, + "ewc_loss": 0.00457763671875, + "ewc_loss_diag": 4.32133674621582e-06, + "ewc_loss_parallel": 4.1425228118896484e-06, + "grad_norm": 5.644538402557373, + "learning_rate": 5.832980076303518e-07, + "loss": 0.459, + "mean_token_accuracy": 0.8483991622924805, + "num_tokens": 53444119.0, + "step": 1377 + }, + { + "epoch": 0.17529576389772294, + "ewc_loss": 0.00457763671875, + "ewc_loss_diag": 4.32133674621582e-06, + "ewc_loss_parallel": 4.1425228118896484e-06, + "grad_norm": 5.781431674957275, + "learning_rate": 5.837219160661297e-07, + "loss": 0.3788, + "mean_token_accuracy": 0.8713181018829346, + "num_tokens": 53477890.0, + "step": 1378 + }, + { + "epoch": 0.17542297417631345, + "ewc_loss": 0.004608154296875, + "ewc_loss_diag": 4.32133674621582e-06, + "ewc_loss_parallel": 4.172325134277344e-06, + "grad_norm": 5.6343488693237305, + "learning_rate": 5.841458245019075e-07, + "loss": 0.3612, + "mean_token_accuracy": 0.8778034448623657, + "num_tokens": 53518957.0, + "step": 1379 + }, + { + "epoch": 0.17555018445490395, + "ewc_loss": 0.0045166015625, + "ewc_loss_diag": 4.32133674621582e-06, + "ewc_loss_parallel": 4.082918167114258e-06, + "grad_norm": 5.826598167419434, + "learning_rate": 5.845697329376855e-07, + "loss": 0.4106, + "mean_token_accuracy": 0.8654112815856934, + "num_tokens": 53551083.0, + "step": 1380 + }, + { + "epoch": 0.17567739473349447, + "ewc_loss": 0.004669189453125, + "ewc_loss_diag": 4.32133674621582e-06, + "ewc_loss_parallel": 4.231929779052734e-06, + "grad_norm": 5.702979564666748, + "learning_rate": 5.849936413734633e-07, + "loss": 0.4615, + "mean_token_accuracy": 0.8495450019836426, + "num_tokens": 53592853.0, + "step": 1381 + }, + { + "epoch": 0.17580460501208497, + "ewc_loss": 0.00457763671875, + "ewc_loss_diag": 4.32133674621582e-06, + "ewc_loss_parallel": 4.1425228118896484e-06, + "grad_norm": 5.520991802215576, + "learning_rate": 5.854175498092412e-07, + "loss": 0.3697, + "mean_token_accuracy": 0.8754640817642212, + "num_tokens": 53634972.0, + "step": 1382 + }, + { + "epoch": 0.17593181529067548, + "ewc_loss": 0.0045166015625, + "ewc_loss_diag": 4.32133674621582e-06, + "ewc_loss_parallel": 4.082918167114258e-06, + "grad_norm": 5.553425312042236, + "learning_rate": 5.858414582450191e-07, + "loss": 0.4059, + "mean_token_accuracy": 0.8640404343605042, + "num_tokens": 53680453.0, + "step": 1383 + }, + { + "epoch": 0.176059025569266, + "ewc_loss": 0.0045166015625, + "ewc_loss_diag": 4.32133674621582e-06, + "ewc_loss_parallel": 4.082918167114258e-06, + "grad_norm": 5.7664313316345215, + "learning_rate": 5.86265366680797e-07, + "loss": 0.4324, + "mean_token_accuracy": 0.8559762239456177, + "num_tokens": 53721033.0, + "step": 1384 + }, + { + "epoch": 0.1761862358478565, + "ewc_loss": 0.004669189453125, + "ewc_loss_diag": 4.32133674621582e-06, + "ewc_loss_parallel": 4.231929779052734e-06, + "grad_norm": 5.630690574645996, + "learning_rate": 5.866892751165748e-07, + "loss": 0.4137, + "mean_token_accuracy": 0.8614950180053711, + "num_tokens": 53765236.0, + "step": 1385 + }, + { + "epoch": 0.176313446126447, + "ewc_loss": 0.004547119140625, + "ewc_loss_diag": 4.32133674621582e-06, + "ewc_loss_parallel": 4.112720489501953e-06, + "grad_norm": 5.843618392944336, + "learning_rate": 5.871131835523526e-07, + "loss": 0.4692, + "mean_token_accuracy": 0.8466480374336243, + "num_tokens": 53801319.0, + "step": 1386 + }, + { + "epoch": 0.17644065640503753, + "ewc_loss": 0.004669189453125, + "ewc_loss_diag": 4.351139068603516e-06, + "ewc_loss_parallel": 4.231929779052734e-06, + "grad_norm": 5.661970138549805, + "learning_rate": 5.875370919881305e-07, + "loss": 0.4146, + "mean_token_accuracy": 0.860078752040863, + "num_tokens": 53847248.0, + "step": 1387 + }, + { + "epoch": 0.17656786668362803, + "ewc_loss": 0.0045166015625, + "ewc_loss_diag": 4.351139068603516e-06, + "ewc_loss_parallel": 4.082918167114258e-06, + "grad_norm": 5.767055034637451, + "learning_rate": 5.879610004239084e-07, + "loss": 0.4548, + "mean_token_accuracy": 0.8513959050178528, + "num_tokens": 53887026.0, + "step": 1388 + }, + { + "epoch": 0.17669507696221853, + "ewc_loss": 0.004669189453125, + "ewc_loss_diag": 4.351139068603516e-06, + "ewc_loss_parallel": 4.231929779052734e-06, + "grad_norm": 5.767693996429443, + "learning_rate": 5.883849088596863e-07, + "loss": 0.4425, + "mean_token_accuracy": 0.8531975746154785, + "num_tokens": 53930702.0, + "step": 1389 + }, + { + "epoch": 0.17682228724080906, + "ewc_loss": 0.004638671875, + "ewc_loss_diag": 4.351139068603516e-06, + "ewc_loss_parallel": 4.202127456665039e-06, + "grad_norm": 5.754478454589844, + "learning_rate": 5.888088172954641e-07, + "loss": 0.4109, + "mean_token_accuracy": 0.8635175228118896, + "num_tokens": 53967291.0, + "step": 1390 + }, + { + "epoch": 0.17694949751939956, + "ewc_loss": 0.004608154296875, + "ewc_loss_diag": 4.351139068603516e-06, + "ewc_loss_parallel": 4.172325134277344e-06, + "grad_norm": 5.7246222496032715, + "learning_rate": 5.892327257312421e-07, + "loss": 0.3606, + "mean_token_accuracy": 0.8785936832427979, + "num_tokens": 54003830.0, + "step": 1391 + }, + { + "epoch": 0.1770767077979901, + "ewc_loss": 0.004608154296875, + "ewc_loss_diag": 4.351139068603516e-06, + "ewc_loss_parallel": 4.172325134277344e-06, + "grad_norm": 5.674216270446777, + "learning_rate": 5.896566341670199e-07, + "loss": 0.3691, + "mean_token_accuracy": 0.8705211877822876, + "num_tokens": 54043618.0, + "step": 1392 + }, + { + "epoch": 0.1772039180765806, + "ewc_loss": 0.00457763671875, + "ewc_loss_diag": 4.351139068603516e-06, + "ewc_loss_parallel": 4.1425228118896484e-06, + "grad_norm": 5.8503193855285645, + "learning_rate": 5.900805426027977e-07, + "loss": 0.4088, + "mean_token_accuracy": 0.8630400896072388, + "num_tokens": 54080953.0, + "step": 1393 + }, + { + "epoch": 0.1773311283551711, + "ewc_loss": 0.004608154296875, + "ewc_loss_diag": 4.351139068603516e-06, + "ewc_loss_parallel": 4.172325134277344e-06, + "grad_norm": 5.653087139129639, + "learning_rate": 5.905044510385756e-07, + "loss": 0.3704, + "mean_token_accuracy": 0.8749364614486694, + "num_tokens": 54120481.0, + "step": 1394 + }, + { + "epoch": 0.17745833863376162, + "ewc_loss": 0.004547119140625, + "ewc_loss_diag": 4.351139068603516e-06, + "ewc_loss_parallel": 4.112720489501953e-06, + "grad_norm": 5.850491523742676, + "learning_rate": 5.909283594743535e-07, + "loss": 0.4622, + "mean_token_accuracy": 0.8426647186279297, + "num_tokens": 54158964.0, + "step": 1395 + }, + { + "epoch": 0.17758554891235212, + "ewc_loss": 0.0047607421875, + "ewc_loss_diag": 4.351139068603516e-06, + "ewc_loss_parallel": 4.32133674621582e-06, + "grad_norm": 5.790659427642822, + "learning_rate": 5.913522679101314e-07, + "loss": 0.411, + "mean_token_accuracy": 0.8598969578742981, + "num_tokens": 54197485.0, + "step": 1396 + }, + { + "epoch": 0.17771275919094262, + "ewc_loss": 0.00469970703125, + "ewc_loss_diag": 4.351139068603516e-06, + "ewc_loss_parallel": 4.26173210144043e-06, + "grad_norm": 5.687134265899658, + "learning_rate": 5.917761763459093e-07, + "loss": 0.4337, + "mean_token_accuracy": 0.8526257872581482, + "num_tokens": 54239053.0, + "step": 1397 + }, + { + "epoch": 0.17783996946953315, + "ewc_loss": 0.004669189453125, + "ewc_loss_diag": 4.351139068603516e-06, + "ewc_loss_parallel": 4.231929779052734e-06, + "grad_norm": 5.715907096862793, + "learning_rate": 5.922000847816871e-07, + "loss": 0.3741, + "mean_token_accuracy": 0.8720825910568237, + "num_tokens": 54277225.0, + "step": 1398 + }, + { + "epoch": 0.17796717974812365, + "ewc_loss": 0.004669189453125, + "ewc_loss_diag": 4.351139068603516e-06, + "ewc_loss_parallel": 4.231929779052734e-06, + "grad_norm": 5.682163238525391, + "learning_rate": 5.926239932174651e-07, + "loss": 0.4468, + "mean_token_accuracy": 0.8510036468505859, + "num_tokens": 54322925.0, + "step": 1399 + }, + { + "epoch": 0.17809439002671415, + "ewc_loss": 0.004669189453125, + "ewc_loss_diag": 4.380941390991211e-06, + "ewc_loss_parallel": 4.231929779052734e-06, + "grad_norm": 5.765899658203125, + "learning_rate": 5.930479016532429e-07, + "loss": 0.4148, + "mean_token_accuracy": 0.8596639633178711, + "num_tokens": 54359720.0, + "step": 1400 + }, + { + "epoch": 0.17822160030530468, + "ewc_loss": 0.004669189453125, + "ewc_loss_diag": 4.380941390991211e-06, + "ewc_loss_parallel": 4.231929779052734e-06, + "grad_norm": 5.733645439147949, + "learning_rate": 5.934718100890207e-07, + "loss": 0.3827, + "mean_token_accuracy": 0.870540976524353, + "num_tokens": 54400278.0, + "step": 1401 + }, + { + "epoch": 0.17834881058389518, + "ewc_loss": 0.004669189453125, + "ewc_loss_diag": 4.380941390991211e-06, + "ewc_loss_parallel": 4.231929779052734e-06, + "grad_norm": 5.825935363769531, + "learning_rate": 5.938957185247986e-07, + "loss": 0.427, + "mean_token_accuracy": 0.855265736579895, + "num_tokens": 54438215.0, + "step": 1402 + }, + { + "epoch": 0.17847602086248568, + "ewc_loss": 0.00469970703125, + "ewc_loss_diag": 4.380941390991211e-06, + "ewc_loss_parallel": 4.26173210144043e-06, + "grad_norm": 5.8945417404174805, + "learning_rate": 5.943196269605765e-07, + "loss": 0.3927, + "mean_token_accuracy": 0.8642931580543518, + "num_tokens": 54473344.0, + "step": 1403 + }, + { + "epoch": 0.1786032311410762, + "ewc_loss": 0.004669189453125, + "ewc_loss_diag": 4.380941390991211e-06, + "ewc_loss_parallel": 4.231929779052734e-06, + "grad_norm": 5.89610481262207, + "learning_rate": 5.947435353963544e-07, + "loss": 0.4434, + "mean_token_accuracy": 0.8548201322555542, + "num_tokens": 54506801.0, + "step": 1404 + }, + { + "epoch": 0.1787304414196667, + "ewc_loss": 0.004669189453125, + "ewc_loss_diag": 4.380941390991211e-06, + "ewc_loss_parallel": 4.231929779052734e-06, + "grad_norm": 5.817648410797119, + "learning_rate": 5.951674438321323e-07, + "loss": 0.4231, + "mean_token_accuracy": 0.8600454330444336, + "num_tokens": 54541730.0, + "step": 1405 + }, + { + "epoch": 0.1788576516982572, + "ewc_loss": 0.004669189453125, + "ewc_loss_diag": 4.380941390991211e-06, + "ewc_loss_parallel": 4.231929779052734e-06, + "grad_norm": 5.872320652008057, + "learning_rate": 5.955913522679101e-07, + "loss": 0.4285, + "mean_token_accuracy": 0.8524870872497559, + "num_tokens": 54579224.0, + "step": 1406 + }, + { + "epoch": 0.17898486197684774, + "ewc_loss": 0.0047607421875, + "ewc_loss_diag": 4.380941390991211e-06, + "ewc_loss_parallel": 4.32133674621582e-06, + "grad_norm": 5.929079055786133, + "learning_rate": 5.96015260703688e-07, + "loss": 0.4808, + "mean_token_accuracy": 0.8403588533401489, + "num_tokens": 54621331.0, + "step": 1407 + }, + { + "epoch": 0.17911207225543824, + "ewc_loss": 0.0047607421875, + "ewc_loss_diag": 4.380941390991211e-06, + "ewc_loss_parallel": 4.32133674621582e-06, + "grad_norm": 5.645598411560059, + "learning_rate": 5.964391691394659e-07, + "loss": 0.4239, + "mean_token_accuracy": 0.8613471984863281, + "num_tokens": 54669476.0, + "step": 1408 + }, + { + "epoch": 0.17923928253402874, + "ewc_loss": 0.004669189453125, + "ewc_loss_diag": 4.380941390991211e-06, + "ewc_loss_parallel": 4.231929779052734e-06, + "grad_norm": 5.824272632598877, + "learning_rate": 5.968630775752436e-07, + "loss": 0.4482, + "mean_token_accuracy": 0.8497344255447388, + "num_tokens": 54704665.0, + "step": 1409 + }, + { + "epoch": 0.17936649281261927, + "ewc_loss": 0.0047607421875, + "ewc_loss_diag": 4.410743713378906e-06, + "ewc_loss_parallel": 4.32133674621582e-06, + "grad_norm": 5.894944667816162, + "learning_rate": 5.972869860110216e-07, + "loss": 0.4417, + "mean_token_accuracy": 0.8509006500244141, + "num_tokens": 54740048.0, + "step": 1410 + }, + { + "epoch": 0.17949370309120977, + "ewc_loss": 0.004791259765625, + "ewc_loss_diag": 4.410743713378906e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.848517417907715, + "learning_rate": 5.977108944467994e-07, + "loss": 0.485, + "mean_token_accuracy": 0.8454174995422363, + "num_tokens": 54780317.0, + "step": 1411 + }, + { + "epoch": 0.17962091336980027, + "ewc_loss": 0.0047607421875, + "ewc_loss_diag": 4.410743713378906e-06, + "ewc_loss_parallel": 4.32133674621582e-06, + "grad_norm": 5.686787128448486, + "learning_rate": 5.981348028825774e-07, + "loss": 0.3956, + "mean_token_accuracy": 0.8719103336334229, + "num_tokens": 54823747.0, + "step": 1412 + }, + { + "epoch": 0.1797481236483908, + "ewc_loss": 0.0047607421875, + "ewc_loss_diag": 4.410743713378906e-06, + "ewc_loss_parallel": 4.32133674621582e-06, + "grad_norm": 5.819018840789795, + "learning_rate": 5.985587113183552e-07, + "loss": 0.4379, + "mean_token_accuracy": 0.8553292751312256, + "num_tokens": 54863799.0, + "step": 1413 + }, + { + "epoch": 0.1798753339269813, + "ewc_loss": 0.004791259765625, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.32133674621582e-06, + "grad_norm": 5.80547571182251, + "learning_rate": 5.989826197541331e-07, + "loss": 0.4255, + "mean_token_accuracy": 0.8618326187133789, + "num_tokens": 54902126.0, + "step": 1414 + }, + { + "epoch": 0.18000254420557182, + "ewc_loss": 0.004791259765625, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.32133674621582e-06, + "grad_norm": 5.669166088104248, + "learning_rate": 5.99406528189911e-07, + "loss": 0.4505, + "mean_token_accuracy": 0.8538945317268372, + "num_tokens": 54948996.0, + "step": 1415 + }, + { + "epoch": 0.18012975448416232, + "ewc_loss": 0.004791259765625, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.32133674621582e-06, + "grad_norm": 5.6620869636535645, + "learning_rate": 5.998304366256888e-07, + "loss": 0.4148, + "mean_token_accuracy": 0.8631882071495056, + "num_tokens": 54992275.0, + "step": 1416 + }, + { + "epoch": 0.18025696476275282, + "ewc_loss": 0.004791259765625, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.32133674621582e-06, + "grad_norm": 5.812435150146484, + "learning_rate": 6.002543450614666e-07, + "loss": 0.3523, + "mean_token_accuracy": 0.8833011984825134, + "num_tokens": 55028067.0, + "step": 1417 + }, + { + "epoch": 0.18038417504134335, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.762382984161377, + "learning_rate": 6.006782534972446e-07, + "loss": 0.395, + "mean_token_accuracy": 0.8666316866874695, + "num_tokens": 55066945.0, + "step": 1418 + }, + { + "epoch": 0.18051138531993385, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.837053298950195, + "learning_rate": 6.011021619330224e-07, + "loss": 0.4294, + "mean_token_accuracy": 0.8581031560897827, + "num_tokens": 55106103.0, + "step": 1419 + }, + { + "epoch": 0.18063859559852435, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 6.011318206787109, + "learning_rate": 6.015260703688004e-07, + "loss": 0.4382, + "mean_token_accuracy": 0.8520139455795288, + "num_tokens": 55142007.0, + "step": 1420 + }, + { + "epoch": 0.18076580587711488, + "ewc_loss": 0.004852294921875, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.380941390991211e-06, + "grad_norm": 5.938088893890381, + "learning_rate": 6.019499788045782e-07, + "loss": 0.4201, + "mean_token_accuracy": 0.8624142408370972, + "num_tokens": 55178198.0, + "step": 1421 + }, + { + "epoch": 0.18089301615570538, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.911412239074707, + "learning_rate": 6.023738872403561e-07, + "loss": 0.4019, + "mean_token_accuracy": 0.8645793199539185, + "num_tokens": 55208348.0, + "step": 1422 + }, + { + "epoch": 0.18102022643429588, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.770089626312256, + "learning_rate": 6.02797795676134e-07, + "loss": 0.4115, + "mean_token_accuracy": 0.8648145198822021, + "num_tokens": 55246798.0, + "step": 1423 + }, + { + "epoch": 0.1811474367128864, + "ewc_loss": 0.004852294921875, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.380941390991211e-06, + "grad_norm": 5.83531379699707, + "learning_rate": 6.032217041119118e-07, + "loss": 0.4313, + "mean_token_accuracy": 0.8509136438369751, + "num_tokens": 55283556.0, + "step": 1424 + }, + { + "epoch": 0.1812746469914769, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.891650199890137, + "learning_rate": 6.036456125476896e-07, + "loss": 0.4607, + "mean_token_accuracy": 0.850492000579834, + "num_tokens": 55319241.0, + "step": 1425 + }, + { + "epoch": 0.1814018572700674, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.699831962585449, + "learning_rate": 6.040695209834675e-07, + "loss": 0.3637, + "mean_token_accuracy": 0.8755683302879333, + "num_tokens": 55362353.0, + "step": 1426 + }, + { + "epoch": 0.18152906754865794, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.837791442871094, + "learning_rate": 6.044934294192454e-07, + "loss": 0.4488, + "mean_token_accuracy": 0.8532105088233948, + "num_tokens": 55400423.0, + "step": 1427 + }, + { + "epoch": 0.18165627782724844, + "ewc_loss": 0.004852294921875, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.380941390991211e-06, + "grad_norm": 5.785719394683838, + "learning_rate": 6.049173378550233e-07, + "loss": 0.4087, + "mean_token_accuracy": 0.8621090650558472, + "num_tokens": 55446793.0, + "step": 1428 + }, + { + "epoch": 0.18178348810583894, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.4405460357666016e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.727775573730469, + "learning_rate": 6.053412462908012e-07, + "loss": 0.3845, + "mean_token_accuracy": 0.8696909546852112, + "num_tokens": 55486452.0, + "step": 1429 + }, + { + "epoch": 0.18191069838442947, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.470348358154297e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.815568447113037, + "learning_rate": 6.05765154726579e-07, + "loss": 0.4436, + "mean_token_accuracy": 0.8531014919281006, + "num_tokens": 55528996.0, + "step": 1430 + }, + { + "epoch": 0.18203790866301997, + "ewc_loss": 0.0048828125, + "ewc_loss_diag": 4.470348358154297e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.688301086425781, + "learning_rate": 6.061890631623569e-07, + "loss": 0.3653, + "mean_token_accuracy": 0.8758302927017212, + "num_tokens": 55573082.0, + "step": 1431 + }, + { + "epoch": 0.18216511894161047, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.470348358154297e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.7538161277771, + "learning_rate": 6.066129715981347e-07, + "loss": 0.4076, + "mean_token_accuracy": 0.8625427484512329, + "num_tokens": 55615370.0, + "step": 1432 + }, + { + "epoch": 0.182292329220201, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.470348358154297e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.847127437591553, + "learning_rate": 6.070368800339126e-07, + "loss": 0.3704, + "mean_token_accuracy": 0.8755673170089722, + "num_tokens": 55647618.0, + "step": 1433 + }, + { + "epoch": 0.1824195394987915, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.470348358154297e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.854427814483643, + "learning_rate": 6.074607884696905e-07, + "loss": 0.3907, + "mean_token_accuracy": 0.8678863048553467, + "num_tokens": 55685389.0, + "step": 1434 + }, + { + "epoch": 0.182546749777382, + "ewc_loss": 0.004852294921875, + "ewc_loss_diag": 4.470348358154297e-06, + "ewc_loss_parallel": 4.380941390991211e-06, + "grad_norm": 5.9674506187438965, + "learning_rate": 6.078846969054684e-07, + "loss": 0.4239, + "mean_token_accuracy": 0.8580151796340942, + "num_tokens": 55719793.0, + "step": 1435 + }, + { + "epoch": 0.18267396005597253, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.470348358154297e-06, + "ewc_loss_parallel": 4.4405460357666016e-06, + "grad_norm": 5.781393051147461, + "learning_rate": 6.083086053412463e-07, + "loss": 0.4044, + "mean_token_accuracy": 0.8657718896865845, + "num_tokens": 55760268.0, + "step": 1436 + }, + { + "epoch": 0.18280117033456303, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.470348358154297e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.830922603607178, + "learning_rate": 6.087325137770242e-07, + "loss": 0.3932, + "mean_token_accuracy": 0.8671578168869019, + "num_tokens": 55796102.0, + "step": 1437 + }, + { + "epoch": 0.18292838061315353, + "ewc_loss": 0.0048828125, + "ewc_loss_diag": 4.470348358154297e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.8103556632995605, + "learning_rate": 6.09156422212802e-07, + "loss": 0.4328, + "mean_token_accuracy": 0.8597046136856079, + "num_tokens": 55840977.0, + "step": 1438 + }, + { + "epoch": 0.18305559089174406, + "ewc_loss": 0.0048828125, + "ewc_loss_diag": 4.470348358154297e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.803515434265137, + "learning_rate": 6.095803306485799e-07, + "loss": 0.4131, + "mean_token_accuracy": 0.8610538244247437, + "num_tokens": 55882314.0, + "step": 1439 + }, + { + "epoch": 0.18318280117033456, + "ewc_loss": 0.004852294921875, + "ewc_loss_diag": 4.470348358154297e-06, + "ewc_loss_parallel": 4.380941390991211e-06, + "grad_norm": 5.778182506561279, + "learning_rate": 6.100042390843577e-07, + "loss": 0.4233, + "mean_token_accuracy": 0.8583048582077026, + "num_tokens": 55923986.0, + "step": 1440 + }, + { + "epoch": 0.1833100114489251, + "ewc_loss": 0.0048828125, + "ewc_loss_diag": 4.500150680541992e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.801022052764893, + "learning_rate": 6.104281475201356e-07, + "loss": 0.3913, + "mean_token_accuracy": 0.8669005036354065, + "num_tokens": 55961787.0, + "step": 1441 + }, + { + "epoch": 0.1834372217275156, + "ewc_loss": 0.004852294921875, + "ewc_loss_diag": 4.500150680541992e-06, + "ewc_loss_parallel": 4.380941390991211e-06, + "grad_norm": 5.992379665374756, + "learning_rate": 6.108520559559135e-07, + "loss": 0.4421, + "mean_token_accuracy": 0.8512781858444214, + "num_tokens": 55995899.0, + "step": 1442 + }, + { + "epoch": 0.1835644320061061, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.500150680541992e-06, + "ewc_loss_parallel": 4.4405460357666016e-06, + "grad_norm": 5.773286819458008, + "learning_rate": 6.112759643916914e-07, + "loss": 0.3568, + "mean_token_accuracy": 0.881682276725769, + "num_tokens": 56031926.0, + "step": 1443 + }, + { + "epoch": 0.18369164228469662, + "ewc_loss": 0.004852294921875, + "ewc_loss_diag": 4.5299530029296875e-06, + "ewc_loss_parallel": 4.380941390991211e-06, + "grad_norm": 5.9654436111450195, + "learning_rate": 6.116998728274693e-07, + "loss": 0.4235, + "mean_token_accuracy": 0.8638774156570435, + "num_tokens": 56070872.0, + "step": 1444 + }, + { + "epoch": 0.18381885256328712, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.5299530029296875e-06, + "ewc_loss_parallel": 4.470348358154297e-06, + "grad_norm": 5.857804298400879, + "learning_rate": 6.121237812632472e-07, + "loss": 0.433, + "mean_token_accuracy": 0.8527669906616211, + "num_tokens": 56111001.0, + "step": 1445 + }, + { + "epoch": 0.18394606284187762, + "ewc_loss": 0.0048828125, + "ewc_loss_diag": 4.5299530029296875e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.708886623382568, + "learning_rate": 6.125476896990249e-07, + "loss": 0.3819, + "mean_token_accuracy": 0.8741135001182556, + "num_tokens": 56151242.0, + "step": 1446 + }, + { + "epoch": 0.18407327312046814, + "ewc_loss": 0.00482177734375, + "ewc_loss_diag": 4.5299530029296875e-06, + "ewc_loss_parallel": 4.351139068603516e-06, + "grad_norm": 5.8584160804748535, + "learning_rate": 6.129715981348028e-07, + "loss": 0.4182, + "mean_token_accuracy": 0.8642481565475464, + "num_tokens": 56189749.0, + "step": 1447 + }, + { + "epoch": 0.18420048339905865, + "ewc_loss": 0.0048828125, + "ewc_loss_diag": 4.5299530029296875e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.854194164276123, + "learning_rate": 6.133955065705807e-07, + "loss": 0.4309, + "mean_token_accuracy": 0.854641318321228, + "num_tokens": 56230025.0, + "step": 1448 + }, + { + "epoch": 0.18432769367764915, + "ewc_loss": 0.004852294921875, + "ewc_loss_diag": 4.5299530029296875e-06, + "ewc_loss_parallel": 4.380941390991211e-06, + "grad_norm": 5.814987659454346, + "learning_rate": 6.138194150063585e-07, + "loss": 0.3983, + "mean_token_accuracy": 0.8683038949966431, + "num_tokens": 56270341.0, + "step": 1449 + }, + { + "epoch": 0.18445490395623967, + "ewc_loss": 0.0048828125, + "ewc_loss_diag": 4.5299530029296875e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.839358806610107, + "learning_rate": 6.142433234421365e-07, + "loss": 0.438, + "mean_token_accuracy": 0.8531913757324219, + "num_tokens": 56309752.0, + "step": 1450 + }, + { + "epoch": 0.18458211423483017, + "ewc_loss": 0.0048828125, + "ewc_loss_diag": 4.5299530029296875e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.866951942443848, + "learning_rate": 6.146672318779143e-07, + "loss": 0.4193, + "mean_token_accuracy": 0.8582996726036072, + "num_tokens": 56349114.0, + "step": 1451 + }, + { + "epoch": 0.18470932451342068, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.5299530029296875e-06, + "ewc_loss_parallel": 4.4405460357666016e-06, + "grad_norm": 5.899899482727051, + "learning_rate": 6.150911403136923e-07, + "loss": 0.4149, + "mean_token_accuracy": 0.8619334697723389, + "num_tokens": 56384786.0, + "step": 1452 + }, + { + "epoch": 0.1848365347920112, + "ewc_loss": 0.0048828125, + "ewc_loss_diag": 4.5299530029296875e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.864931583404541, + "learning_rate": 6.155150487494701e-07, + "loss": 0.3613, + "mean_token_accuracy": 0.8792619705200195, + "num_tokens": 56419882.0, + "step": 1453 + }, + { + "epoch": 0.1849637450706017, + "ewc_loss": 0.0048828125, + "ewc_loss_diag": 4.559755325317383e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.752021312713623, + "learning_rate": 6.159389571852479e-07, + "loss": 0.3987, + "mean_token_accuracy": 0.8672749400138855, + "num_tokens": 56464124.0, + "step": 1454 + }, + { + "epoch": 0.1850909553491922, + "ewc_loss": 0.0048828125, + "ewc_loss_diag": 4.589557647705078e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.709442615509033, + "learning_rate": 6.163628656210258e-07, + "loss": 0.4023, + "mean_token_accuracy": 0.8660238981246948, + "num_tokens": 56512113.0, + "step": 1455 + }, + { + "epoch": 0.18521816562778273, + "ewc_loss": 0.004852294921875, + "ewc_loss_diag": 4.589557647705078e-06, + "ewc_loss_parallel": 4.380941390991211e-06, + "grad_norm": 6.009512901306152, + "learning_rate": 6.167867740568037e-07, + "loss": 0.4667, + "mean_token_accuracy": 0.8460421562194824, + "num_tokens": 56549147.0, + "step": 1456 + }, + { + "epoch": 0.18534537590637323, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.589557647705078e-06, + "ewc_loss_parallel": 4.470348358154297e-06, + "grad_norm": 5.906950950622559, + "learning_rate": 6.172106824925815e-07, + "loss": 0.403, + "mean_token_accuracy": 0.8637529611587524, + "num_tokens": 56591016.0, + "step": 1457 + }, + { + "epoch": 0.18547258618496373, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.589557647705078e-06, + "ewc_loss_parallel": 4.4405460357666016e-06, + "grad_norm": 5.825936317443848, + "learning_rate": 6.176345909283595e-07, + "loss": 0.4161, + "mean_token_accuracy": 0.8610478639602661, + "num_tokens": 56630160.0, + "step": 1458 + }, + { + "epoch": 0.18559979646355426, + "ewc_loss": 0.00494384765625, + "ewc_loss_diag": 4.589557647705078e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.96613883972168, + "learning_rate": 6.180584993641373e-07, + "loss": 0.3896, + "mean_token_accuracy": 0.8677464723587036, + "num_tokens": 56668435.0, + "step": 1459 + }, + { + "epoch": 0.18572700674214476, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.589557647705078e-06, + "ewc_loss_parallel": 4.4405460357666016e-06, + "grad_norm": 5.857922554016113, + "learning_rate": 6.184824077999153e-07, + "loss": 0.3808, + "mean_token_accuracy": 0.8736478090286255, + "num_tokens": 56704556.0, + "step": 1460 + }, + { + "epoch": 0.18585421702073526, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.6193599700927734e-06, + "ewc_loss_parallel": 4.4405460357666016e-06, + "grad_norm": 5.984401226043701, + "learning_rate": 6.189063162356931e-07, + "loss": 0.457, + "mean_token_accuracy": 0.8478183746337891, + "num_tokens": 56740353.0, + "step": 1461 + }, + { + "epoch": 0.1859814272993258, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.6193599700927734e-06, + "ewc_loss_parallel": 4.470348358154297e-06, + "grad_norm": 5.920382022857666, + "learning_rate": 6.193302246714709e-07, + "loss": 0.4001, + "mean_token_accuracy": 0.8663166761398315, + "num_tokens": 56779548.0, + "step": 1462 + }, + { + "epoch": 0.1861086375779163, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.6193599700927734e-06, + "ewc_loss_parallel": 4.470348358154297e-06, + "grad_norm": 5.931034564971924, + "learning_rate": 6.197541331072488e-07, + "loss": 0.4306, + "mean_token_accuracy": 0.8544706106185913, + "num_tokens": 56815001.0, + "step": 1463 + }, + { + "epoch": 0.1862358478565068, + "ewc_loss": 0.0048828125, + "ewc_loss_diag": 4.6193599700927734e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.872878074645996, + "learning_rate": 6.201780415430267e-07, + "loss": 0.3994, + "mean_token_accuracy": 0.8665671348571777, + "num_tokens": 56855324.0, + "step": 1464 + }, + { + "epoch": 0.18636305813509732, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.6193599700927734e-06, + "ewc_loss_parallel": 4.470348358154297e-06, + "grad_norm": 5.950620174407959, + "learning_rate": 6.206019499788045e-07, + "loss": 0.4403, + "mean_token_accuracy": 0.8551884889602661, + "num_tokens": 56893645.0, + "step": 1465 + }, + { + "epoch": 0.18649026841368782, + "ewc_loss": 0.00494384765625, + "ewc_loss_diag": 4.6193599700927734e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.817555904388428, + "learning_rate": 6.210258584145825e-07, + "loss": 0.3919, + "mean_token_accuracy": 0.8666888475418091, + "num_tokens": 56931330.0, + "step": 1466 + }, + { + "epoch": 0.18661747869227835, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.6193599700927734e-06, + "ewc_loss_parallel": 4.470348358154297e-06, + "grad_norm": 5.890742778778076, + "learning_rate": 6.214497668503603e-07, + "loss": 0.4278, + "mean_token_accuracy": 0.8548828363418579, + "num_tokens": 56970202.0, + "step": 1467 + }, + { + "epoch": 0.18674468897086885, + "ewc_loss": 0.00494384765625, + "ewc_loss_diag": 4.649162292480469e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 6.165334701538086, + "learning_rate": 6.218736752861383e-07, + "loss": 0.3983, + "mean_token_accuracy": 0.8673419952392578, + "num_tokens": 56998672.0, + "step": 1468 + }, + { + "epoch": 0.18687189924945935, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.649162292480469e-06, + "ewc_loss_parallel": 4.470348358154297e-06, + "grad_norm": 5.918165683746338, + "learning_rate": 6.22297583721916e-07, + "loss": 0.4307, + "mean_token_accuracy": 0.8593091368675232, + "num_tokens": 57038975.0, + "step": 1469 + }, + { + "epoch": 0.18699910952804988, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.649162292480469e-06, + "ewc_loss_parallel": 4.4405460357666016e-06, + "grad_norm": 5.906802177429199, + "learning_rate": 6.227214921576938e-07, + "loss": 0.4538, + "mean_token_accuracy": 0.8521952033042908, + "num_tokens": 57077925.0, + "step": 1470 + }, + { + "epoch": 0.18712631980664038, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.649162292480469e-06, + "ewc_loss_parallel": 4.4405460357666016e-06, + "grad_norm": 6.003296375274658, + "learning_rate": 6.231454005934718e-07, + "loss": 0.3714, + "mean_token_accuracy": 0.8755435943603516, + "num_tokens": 57112329.0, + "step": 1471 + }, + { + "epoch": 0.18725353008523088, + "ewc_loss": 0.00494384765625, + "ewc_loss_diag": 4.649162292480469e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.928067207336426, + "learning_rate": 6.235693090292496e-07, + "loss": 0.4665, + "mean_token_accuracy": 0.848635196685791, + "num_tokens": 57152458.0, + "step": 1472 + }, + { + "epoch": 0.1873807403638214, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.649162292480469e-06, + "ewc_loss_parallel": 4.470348358154297e-06, + "grad_norm": 5.745450973510742, + "learning_rate": 6.239932174650275e-07, + "loss": 0.3608, + "mean_token_accuracy": 0.8780667781829834, + "num_tokens": 57193375.0, + "step": 1473 + }, + { + "epoch": 0.1875079506424119, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.649162292480469e-06, + "ewc_loss_parallel": 4.470348358154297e-06, + "grad_norm": 6.112890720367432, + "learning_rate": 6.244171259008054e-07, + "loss": 0.4315, + "mean_token_accuracy": 0.8552868366241455, + "num_tokens": 57228422.0, + "step": 1474 + }, + { + "epoch": 0.1876351609210024, + "ewc_loss": 0.00494384765625, + "ewc_loss_diag": 4.649162292480469e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.882360458374023, + "learning_rate": 6.248410343365833e-07, + "loss": 0.3802, + "mean_token_accuracy": 0.871502161026001, + "num_tokens": 57266509.0, + "step": 1475 + }, + { + "epoch": 0.18776237119959294, + "ewc_loss": 0.00494384765625, + "ewc_loss_diag": 4.649162292480469e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.738903999328613, + "learning_rate": 6.252649427723612e-07, + "loss": 0.3948, + "mean_token_accuracy": 0.8713716268539429, + "num_tokens": 57310042.0, + "step": 1476 + }, + { + "epoch": 0.18788958147818344, + "ewc_loss": 0.00494384765625, + "ewc_loss_diag": 4.708766937255859e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.906569480895996, + "learning_rate": 6.25688851208139e-07, + "loss": 0.3959, + "mean_token_accuracy": 0.8679466247558594, + "num_tokens": 57348758.0, + "step": 1477 + }, + { + "epoch": 0.18801679175677394, + "ewc_loss": 0.00494384765625, + "ewc_loss_diag": 4.708766937255859e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.7918243408203125, + "learning_rate": 6.261127596439168e-07, + "loss": 0.4328, + "mean_token_accuracy": 0.8529521226882935, + "num_tokens": 57395056.0, + "step": 1478 + }, + { + "epoch": 0.18814400203536447, + "ewc_loss": 0.004974365234375, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.768895149230957, + "learning_rate": 6.265366680796948e-07, + "loss": 0.3598, + "mean_token_accuracy": 0.875239372253418, + "num_tokens": 57435470.0, + "step": 1479 + }, + { + "epoch": 0.18827121231395497, + "ewc_loss": 0.0050048828125, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.5299530029296875e-06, + "grad_norm": 5.798138618469238, + "learning_rate": 6.269605765154726e-07, + "loss": 0.403, + "mean_token_accuracy": 0.8635482788085938, + "num_tokens": 57479546.0, + "step": 1480 + }, + { + "epoch": 0.18839842259254547, + "ewc_loss": 0.004974365234375, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.926037788391113, + "learning_rate": 6.273844849512505e-07, + "loss": 0.4101, + "mean_token_accuracy": 0.862228512763977, + "num_tokens": 57520250.0, + "step": 1481 + }, + { + "epoch": 0.188525632871136, + "ewc_loss": 0.0050048828125, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.5299530029296875e-06, + "grad_norm": 5.7661261558532715, + "learning_rate": 6.278083933870284e-07, + "loss": 0.3454, + "mean_token_accuracy": 0.8840387463569641, + "num_tokens": 57564030.0, + "step": 1482 + }, + { + "epoch": 0.1886528431497265, + "ewc_loss": 0.004974365234375, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.958672523498535, + "learning_rate": 6.282323018228063e-07, + "loss": 0.3606, + "mean_token_accuracy": 0.8802403211593628, + "num_tokens": 57600808.0, + "step": 1483 + }, + { + "epoch": 0.188780053428317, + "ewc_loss": 0.0050048828125, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.5299530029296875e-06, + "grad_norm": 5.859752178192139, + "learning_rate": 6.286562102585841e-07, + "loss": 0.446, + "mean_token_accuracy": 0.8554730415344238, + "num_tokens": 57646910.0, + "step": 1484 + }, + { + "epoch": 0.18890726370690752, + "ewc_loss": 0.004974365234375, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.910045146942139, + "learning_rate": 6.29080118694362e-07, + "loss": 0.3556, + "mean_token_accuracy": 0.8773829936981201, + "num_tokens": 57681683.0, + "step": 1485 + }, + { + "epoch": 0.18903447398549802, + "ewc_loss": 0.0050048828125, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.5299530029296875e-06, + "grad_norm": 6.030506610870361, + "learning_rate": 6.295040271301398e-07, + "loss": 0.3469, + "mean_token_accuracy": 0.8794554471969604, + "num_tokens": 57718266.0, + "step": 1486 + }, + { + "epoch": 0.18916168426408853, + "ewc_loss": 0.004974365234375, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.766750812530518, + "learning_rate": 6.299279355659178e-07, + "loss": 0.3898, + "mean_token_accuracy": 0.8699668049812317, + "num_tokens": 57764762.0, + "step": 1487 + }, + { + "epoch": 0.18928889454267905, + "ewc_loss": 0.004913330078125, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.410743713378906e-06, + "grad_norm": 5.951908111572266, + "learning_rate": 6.303518440016956e-07, + "loss": 0.4106, + "mean_token_accuracy": 0.8611814975738525, + "num_tokens": 57800258.0, + "step": 1488 + }, + { + "epoch": 0.18941610482126955, + "ewc_loss": 0.00494384765625, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.470348358154297e-06, + "grad_norm": 6.048149585723877, + "learning_rate": 6.307757524374735e-07, + "loss": 0.4042, + "mean_token_accuracy": 0.8686186671257019, + "num_tokens": 57835212.0, + "step": 1489 + }, + { + "epoch": 0.18954331509986008, + "ewc_loss": 0.00494384765625, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.470348358154297e-06, + "grad_norm": 5.943185329437256, + "learning_rate": 6.311996608732514e-07, + "loss": 0.4385, + "mean_token_accuracy": 0.8602074384689331, + "num_tokens": 57873294.0, + "step": 1490 + }, + { + "epoch": 0.18967052537845058, + "ewc_loss": 0.00494384765625, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.4405460357666016e-06, + "grad_norm": 6.0065531730651855, + "learning_rate": 6.316235693090292e-07, + "loss": 0.4311, + "mean_token_accuracy": 0.8547998666763306, + "num_tokens": 57909001.0, + "step": 1491 + }, + { + "epoch": 0.18979773565704108, + "ewc_loss": 0.004974365234375, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.951761722564697, + "learning_rate": 6.320474777448071e-07, + "loss": 0.4274, + "mean_token_accuracy": 0.8564451932907104, + "num_tokens": 57946752.0, + "step": 1492 + }, + { + "epoch": 0.1899249459356316, + "ewc_loss": 0.00494384765625, + "ewc_loss_diag": 4.76837158203125e-06, + "ewc_loss_parallel": 4.470348358154297e-06, + "grad_norm": 5.863122463226318, + "learning_rate": 6.324713861805849e-07, + "loss": 0.3907, + "mean_token_accuracy": 0.868811845779419, + "num_tokens": 57987260.0, + "step": 1493 + }, + { + "epoch": 0.1900521562142221, + "ewc_loss": 0.004974365234375, + "ewc_loss_diag": 4.798173904418945e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 6.106955528259277, + "learning_rate": 6.328952946163628e-07, + "loss": 0.4707, + "mean_token_accuracy": 0.8469926714897156, + "num_tokens": 58022450.0, + "step": 1494 + }, + { + "epoch": 0.1901793664928126, + "ewc_loss": 0.004974365234375, + "ewc_loss_diag": 4.798173904418945e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.859132289886475, + "learning_rate": 6.333192030521407e-07, + "loss": 0.363, + "mean_token_accuracy": 0.8747027516365051, + "num_tokens": 58058924.0, + "step": 1495 + }, + { + "epoch": 0.19030657677140314, + "ewc_loss": 0.004974365234375, + "ewc_loss_diag": 4.798173904418945e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.8329386711120605, + "learning_rate": 6.337431114879186e-07, + "loss": 0.404, + "mean_token_accuracy": 0.8635965585708618, + "num_tokens": 58098048.0, + "step": 1496 + }, + { + "epoch": 0.19043378704999364, + "ewc_loss": 0.00494384765625, + "ewc_loss_diag": 4.798173904418945e-06, + "ewc_loss_parallel": 4.470348358154297e-06, + "grad_norm": 5.911489009857178, + "learning_rate": 6.341670199236965e-07, + "loss": 0.3867, + "mean_token_accuracy": 0.8716816902160645, + "num_tokens": 58136601.0, + "step": 1497 + }, + { + "epoch": 0.19056099732858414, + "ewc_loss": 0.004974365234375, + "ewc_loss_diag": 4.798173904418945e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.7238240242004395, + "learning_rate": 6.345909283594744e-07, + "loss": 0.3708, + "mean_token_accuracy": 0.8736588954925537, + "num_tokens": 58183228.0, + "step": 1498 + }, + { + "epoch": 0.19068820760717467, + "ewc_loss": 0.0050048828125, + "ewc_loss_diag": 4.827976226806641e-06, + "ewc_loss_parallel": 4.5299530029296875e-06, + "grad_norm": 6.0095648765563965, + "learning_rate": 6.350148367952522e-07, + "loss": 0.4112, + "mean_token_accuracy": 0.856261670589447, + "num_tokens": 58219823.0, + "step": 1499 + }, + { + "epoch": 0.19081541788576517, + "ewc_loss": 0.005096435546875, + "ewc_loss_diag": 4.798173904418945e-06, + "ewc_loss_parallel": 4.6193599700927734e-06, + "grad_norm": 5.985231399536133, + "learning_rate": 6.354387452310301e-07, + "loss": 0.382, + "mean_token_accuracy": 0.868524432182312, + "num_tokens": 58254785.0, + "step": 1500 + }, + { + "epoch": 0.19094262816435567, + "ewc_loss": 0.0050048828125, + "ewc_loss_diag": 4.798173904418945e-06, + "ewc_loss_parallel": 4.5299530029296875e-06, + "grad_norm": 5.983182430267334, + "learning_rate": 6.358626536668079e-07, + "loss": 0.3828, + "mean_token_accuracy": 0.8705646991729736, + "num_tokens": 58291132.0, + "step": 1501 + }, + { + "epoch": 0.1910698384429462, + "ewc_loss": 0.0050048828125, + "ewc_loss_diag": 4.798173904418945e-06, + "ewc_loss_parallel": 4.5299530029296875e-06, + "grad_norm": 5.813236713409424, + "learning_rate": 6.362865621025858e-07, + "loss": 0.3568, + "mean_token_accuracy": 0.8792513608932495, + "num_tokens": 58335323.0, + "step": 1502 + }, + { + "epoch": 0.1911970487215367, + "ewc_loss": 0.004974365234375, + "ewc_loss_diag": 4.798173904418945e-06, + "ewc_loss_parallel": 4.500150680541992e-06, + "grad_norm": 5.93528413772583, + "learning_rate": 6.367104705383637e-07, + "loss": 0.3981, + "mean_token_accuracy": 0.8648621439933777, + "num_tokens": 58373706.0, + "step": 1503 + }, + { + "epoch": 0.1913242590001272, + "ewc_loss": 0.0050048828125, + "ewc_loss_diag": 4.827976226806641e-06, + "ewc_loss_parallel": 4.5299530029296875e-06, + "grad_norm": 5.922893524169922, + "learning_rate": 6.371343789741416e-07, + "loss": 0.4091, + "mean_token_accuracy": 0.8620833158493042, + "num_tokens": 58415873.0, + "step": 1504 + }, + { + "epoch": 0.19145146927871773, + "ewc_loss": 0.0050048828125, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.5299530029296875e-06, + "grad_norm": 5.885263442993164, + "learning_rate": 6.375582874099195e-07, + "loss": 0.4013, + "mean_token_accuracy": 0.8654220104217529, + "num_tokens": 58454609.0, + "step": 1505 + }, + { + "epoch": 0.19157867955730823, + "ewc_loss": 0.0050048828125, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.5299530029296875e-06, + "grad_norm": 6.120552062988281, + "learning_rate": 6.379821958456974e-07, + "loss": 0.4384, + "mean_token_accuracy": 0.854295551776886, + "num_tokens": 58490392.0, + "step": 1506 + }, + { + "epoch": 0.19170588983589873, + "ewc_loss": 0.005096435546875, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.6193599700927734e-06, + "grad_norm": 6.160540580749512, + "learning_rate": 6.384061042814751e-07, + "loss": 0.4266, + "mean_token_accuracy": 0.8574148416519165, + "num_tokens": 58527261.0, + "step": 1507 + }, + { + "epoch": 0.19183310011448926, + "ewc_loss": 0.005035400390625, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.559755325317383e-06, + "grad_norm": 5.899484634399414, + "learning_rate": 6.38830012717253e-07, + "loss": 0.4108, + "mean_token_accuracy": 0.8620474934577942, + "num_tokens": 58568751.0, + "step": 1508 + }, + { + "epoch": 0.19196031039307976, + "ewc_loss": 0.005035400390625, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.559755325317383e-06, + "grad_norm": 5.883081912994385, + "learning_rate": 6.392539211530309e-07, + "loss": 0.4108, + "mean_token_accuracy": 0.8619508147239685, + "num_tokens": 58615380.0, + "step": 1509 + }, + { + "epoch": 0.19208752067167026, + "ewc_loss": 0.0050048828125, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.5299530029296875e-06, + "grad_norm": 5.97089147567749, + "learning_rate": 6.396778295888087e-07, + "loss": 0.3805, + "mean_token_accuracy": 0.8728172779083252, + "num_tokens": 58654130.0, + "step": 1510 + }, + { + "epoch": 0.1922147309502608, + "ewc_loss": 0.00506591796875, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.589557647705078e-06, + "grad_norm": 6.063029766082764, + "learning_rate": 6.401017380245867e-07, + "loss": 0.4129, + "mean_token_accuracy": 0.8626922369003296, + "num_tokens": 58694714.0, + "step": 1511 + }, + { + "epoch": 0.1923419412288513, + "ewc_loss": 0.005096435546875, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.6193599700927734e-06, + "grad_norm": 5.885412216186523, + "learning_rate": 6.405256464603645e-07, + "loss": 0.4132, + "mean_token_accuracy": 0.8592953681945801, + "num_tokens": 58737394.0, + "step": 1512 + }, + { + "epoch": 0.1924691515074418, + "ewc_loss": 0.005035400390625, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.559755325317383e-06, + "grad_norm": 5.961343288421631, + "learning_rate": 6.409495548961425e-07, + "loss": 0.3899, + "mean_token_accuracy": 0.8706623315811157, + "num_tokens": 58774472.0, + "step": 1513 + }, + { + "epoch": 0.19259636178603232, + "ewc_loss": 0.005035400390625, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.559755325317383e-06, + "grad_norm": 5.965317726135254, + "learning_rate": 6.413734633319203e-07, + "loss": 0.3948, + "mean_token_accuracy": 0.8697028160095215, + "num_tokens": 58811109.0, + "step": 1514 + }, + { + "epoch": 0.19272357206462282, + "ewc_loss": 0.00506591796875, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.589557647705078e-06, + "grad_norm": 5.968068599700928, + "learning_rate": 6.417973717676981e-07, + "loss": 0.3717, + "mean_token_accuracy": 0.8733769655227661, + "num_tokens": 58846290.0, + "step": 1515 + }, + { + "epoch": 0.19285078234321335, + "ewc_loss": 0.00506591796875, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.589557647705078e-06, + "grad_norm": 5.756193161010742, + "learning_rate": 6.42221280203476e-07, + "loss": 0.4226, + "mean_token_accuracy": 0.8586381673812866, + "num_tokens": 58897185.0, + "step": 1516 + }, + { + "epoch": 0.19297799262180385, + "ewc_loss": 0.0050048828125, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.5299530029296875e-06, + "grad_norm": 5.97787618637085, + "learning_rate": 6.426451886392539e-07, + "loss": 0.4561, + "mean_token_accuracy": 0.8529961109161377, + "num_tokens": 58937525.0, + "step": 1517 + }, + { + "epoch": 0.19310520290039435, + "ewc_loss": 0.005096435546875, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.6193599700927734e-06, + "grad_norm": 6.036740779876709, + "learning_rate": 6.430690970750317e-07, + "loss": 0.3832, + "mean_token_accuracy": 0.8692554235458374, + "num_tokens": 58976954.0, + "step": 1518 + }, + { + "epoch": 0.19323241317898487, + "ewc_loss": 0.00506591796875, + "ewc_loss_diag": 4.857778549194336e-06, + "ewc_loss_parallel": 4.589557647705078e-06, + "grad_norm": 6.154398441314697, + "learning_rate": 6.434930055108097e-07, + "loss": 0.43, + "mean_token_accuracy": 0.8558490872383118, + "num_tokens": 59007442.0, + "step": 1519 + }, + { + "epoch": 0.19335962345757537, + "ewc_loss": 0.005096435546875, + "ewc_loss_diag": 4.887580871582031e-06, + "ewc_loss_parallel": 4.6193599700927734e-06, + "grad_norm": 5.997862339019775, + "learning_rate": 6.439169139465875e-07, + "loss": 0.4297, + "mean_token_accuracy": 0.8564854860305786, + "num_tokens": 59045421.0, + "step": 1520 + }, + { + "epoch": 0.19348683373616588, + "ewc_loss": 0.005035400390625, + "ewc_loss_diag": 4.9173831939697266e-06, + "ewc_loss_parallel": 4.559755325317383e-06, + "grad_norm": 5.9186296463012695, + "learning_rate": 6.443408223823655e-07, + "loss": 0.4065, + "mean_token_accuracy": 0.8632651567459106, + "num_tokens": 59081122.0, + "step": 1521 + }, + { + "epoch": 0.1936140440147564, + "ewc_loss": 0.00506591796875, + "ewc_loss_diag": 4.9173831939697266e-06, + "ewc_loss_parallel": 4.589557647705078e-06, + "grad_norm": 5.951056003570557, + "learning_rate": 6.447647308181432e-07, + "loss": 0.4021, + "mean_token_accuracy": 0.8642755150794983, + "num_tokens": 59121048.0, + "step": 1522 + }, + { + "epoch": 0.1937412542933469, + "ewc_loss": 0.00506591796875, + "ewc_loss_diag": 4.9173831939697266e-06, + "ewc_loss_parallel": 4.589557647705078e-06, + "grad_norm": 5.883667469024658, + "learning_rate": 6.451886392539211e-07, + "loss": 0.3884, + "mean_token_accuracy": 0.8688251972198486, + "num_tokens": 59163085.0, + "step": 1523 + }, + { + "epoch": 0.1938684645719374, + "ewc_loss": 0.005126953125, + "ewc_loss_diag": 4.9173831939697266e-06, + "ewc_loss_parallel": 4.649162292480469e-06, + "grad_norm": 6.02532434463501, + "learning_rate": 6.45612547689699e-07, + "loss": 0.4118, + "mean_token_accuracy": 0.8637065291404724, + "num_tokens": 59202875.0, + "step": 1524 + }, + { + "epoch": 0.19399567485052793, + "ewc_loss": 0.005157470703125, + "ewc_loss_diag": 4.9173831939697266e-06, + "ewc_loss_parallel": 4.678964614868164e-06, + "grad_norm": 5.92724609375, + "learning_rate": 6.460364561254769e-07, + "loss": 0.3707, + "mean_token_accuracy": 0.8755910396575928, + "num_tokens": 59243467.0, + "step": 1525 + }, + { + "epoch": 0.19412288512911843, + "ewc_loss": 0.005096435546875, + "ewc_loss_diag": 4.9173831939697266e-06, + "ewc_loss_parallel": 4.6193599700927734e-06, + "grad_norm": 5.90783166885376, + "learning_rate": 6.464603645612547e-07, + "loss": 0.3663, + "mean_token_accuracy": 0.8763325214385986, + "num_tokens": 59281828.0, + "step": 1526 + }, + { + "epoch": 0.19425009540770893, + "ewc_loss": 0.005096435546875, + "ewc_loss_diag": 4.9173831939697266e-06, + "ewc_loss_parallel": 4.6193599700927734e-06, + "grad_norm": 5.891116142272949, + "learning_rate": 6.468842729970327e-07, + "loss": 0.4012, + "mean_token_accuracy": 0.8675045967102051, + "num_tokens": 59324106.0, + "step": 1527 + }, + { + "epoch": 0.19437730568629946, + "ewc_loss": 0.005157470703125, + "ewc_loss_diag": 4.9173831939697266e-06, + "ewc_loss_parallel": 4.678964614868164e-06, + "grad_norm": 6.053345203399658, + "learning_rate": 6.473081814328105e-07, + "loss": 0.4333, + "mean_token_accuracy": 0.8594191670417786, + "num_tokens": 59366550.0, + "step": 1528 + }, + { + "epoch": 0.19450451596488996, + "ewc_loss": 0.005126953125, + "ewc_loss_diag": 4.9173831939697266e-06, + "ewc_loss_parallel": 4.649162292480469e-06, + "grad_norm": 5.90958833694458, + "learning_rate": 6.477320898685885e-07, + "loss": 0.4154, + "mean_token_accuracy": 0.861745297908783, + "num_tokens": 59410940.0, + "step": 1529 + }, + { + "epoch": 0.19463172624348046, + "ewc_loss": 0.005126953125, + "ewc_loss_diag": 4.976987838745117e-06, + "ewc_loss_parallel": 4.649162292480469e-06, + "grad_norm": 5.996178150177002, + "learning_rate": 6.481559983043662e-07, + "loss": 0.4232, + "mean_token_accuracy": 0.8562081456184387, + "num_tokens": 59451032.0, + "step": 1530 + }, + { + "epoch": 0.194758936522071, + "ewc_loss": 0.005126953125, + "ewc_loss_diag": 4.976987838745117e-06, + "ewc_loss_parallel": 4.649162292480469e-06, + "grad_norm": 5.93691873550415, + "learning_rate": 6.48579906740144e-07, + "loss": 0.3776, + "mean_token_accuracy": 0.8731327652931213, + "num_tokens": 59492217.0, + "step": 1531 + }, + { + "epoch": 0.1948861468006615, + "ewc_loss": 0.005096435546875, + "ewc_loss_diag": 4.976987838745117e-06, + "ewc_loss_parallel": 4.6193599700927734e-06, + "grad_norm": 5.9237494468688965, + "learning_rate": 6.49003815175922e-07, + "loss": 0.3569, + "mean_token_accuracy": 0.8814805150032043, + "num_tokens": 59528588.0, + "step": 1532 + }, + { + "epoch": 0.195013357079252, + "ewc_loss": 0.005126953125, + "ewc_loss_diag": 4.976987838745117e-06, + "ewc_loss_parallel": 4.649162292480469e-06, + "grad_norm": 6.045069217681885, + "learning_rate": 6.494277236116998e-07, + "loss": 0.3493, + "mean_token_accuracy": 0.8804159164428711, + "num_tokens": 59563519.0, + "step": 1533 + }, + { + "epoch": 0.19514056735784252, + "ewc_loss": 0.005157470703125, + "ewc_loss_diag": 4.976987838745117e-06, + "ewc_loss_parallel": 4.678964614868164e-06, + "grad_norm": 5.983330726623535, + "learning_rate": 6.498516320474777e-07, + "loss": 0.4753, + "mean_token_accuracy": 0.8446245193481445, + "num_tokens": 59611192.0, + "step": 1534 + }, + { + "epoch": 0.19526777763643302, + "ewc_loss": 0.005157470703125, + "ewc_loss_diag": 4.976987838745117e-06, + "ewc_loss_parallel": 4.678964614868164e-06, + "grad_norm": 6.042212009429932, + "learning_rate": 6.502755404832556e-07, + "loss": 0.3885, + "mean_token_accuracy": 0.8681958913803101, + "num_tokens": 59649055.0, + "step": 1535 + }, + { + "epoch": 0.19539498791502352, + "ewc_loss": 0.005126953125, + "ewc_loss_diag": 4.976987838745117e-06, + "ewc_loss_parallel": 4.649162292480469e-06, + "grad_norm": 6.08386754989624, + "learning_rate": 6.506994489190335e-07, + "loss": 0.4472, + "mean_token_accuracy": 0.8557320833206177, + "num_tokens": 59688898.0, + "step": 1536 + }, + { + "epoch": 0.19552219819361405, + "ewc_loss": 0.005126953125, + "ewc_loss_diag": 4.976987838745117e-06, + "ewc_loss_parallel": 4.649162292480469e-06, + "grad_norm": 5.939119338989258, + "learning_rate": 6.511233573548114e-07, + "loss": 0.3705, + "mean_token_accuracy": 0.875262975692749, + "num_tokens": 59729585.0, + "step": 1537 + }, + { + "epoch": 0.19564940847220455, + "ewc_loss": 0.005126953125, + "ewc_loss_diag": 4.976987838745117e-06, + "ewc_loss_parallel": 4.649162292480469e-06, + "grad_norm": 6.112229824066162, + "learning_rate": 6.515472657905892e-07, + "loss": 0.4014, + "mean_token_accuracy": 0.8642223477363586, + "num_tokens": 59761753.0, + "step": 1538 + }, + { + "epoch": 0.19577661875079505, + "ewc_loss": 0.00518798828125, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.708766937255859e-06, + "grad_norm": 6.165615558624268, + "learning_rate": 6.51971174226367e-07, + "loss": 0.4536, + "mean_token_accuracy": 0.8526848554611206, + "num_tokens": 59799740.0, + "step": 1539 + }, + { + "epoch": 0.19590382902938558, + "ewc_loss": 0.005157470703125, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.678964614868164e-06, + "grad_norm": 5.99790096282959, + "learning_rate": 6.52395082662145e-07, + "loss": 0.4439, + "mean_token_accuracy": 0.8519093990325928, + "num_tokens": 59839616.0, + "step": 1540 + }, + { + "epoch": 0.19603103930797608, + "ewc_loss": 0.005126953125, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.649162292480469e-06, + "grad_norm": 6.034433841705322, + "learning_rate": 6.528189910979228e-07, + "loss": 0.3877, + "mean_token_accuracy": 0.86980140209198, + "num_tokens": 59876750.0, + "step": 1541 + }, + { + "epoch": 0.1961582495865666, + "ewc_loss": 0.005157470703125, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.678964614868164e-06, + "grad_norm": 5.903759002685547, + "learning_rate": 6.532428995337007e-07, + "loss": 0.3948, + "mean_token_accuracy": 0.8674855828285217, + "num_tokens": 59920047.0, + "step": 1542 + }, + { + "epoch": 0.1962854598651571, + "ewc_loss": 0.005126953125, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.649162292480469e-06, + "grad_norm": 6.018199443817139, + "learning_rate": 6.536668079694786e-07, + "loss": 0.4136, + "mean_token_accuracy": 0.8592764735221863, + "num_tokens": 59956887.0, + "step": 1543 + }, + { + "epoch": 0.1964126701437476, + "ewc_loss": 0.005157470703125, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.678964614868164e-06, + "grad_norm": 5.959700584411621, + "learning_rate": 6.540907164052565e-07, + "loss": 0.4093, + "mean_token_accuracy": 0.8613816499710083, + "num_tokens": 60000387.0, + "step": 1544 + }, + { + "epoch": 0.19653988042233814, + "ewc_loss": 0.005157470703125, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.678964614868164e-06, + "grad_norm": 5.953958511352539, + "learning_rate": 6.545146248410343e-07, + "loss": 0.4101, + "mean_token_accuracy": 0.8612926006317139, + "num_tokens": 60043716.0, + "step": 1545 + }, + { + "epoch": 0.19666709070092864, + "ewc_loss": 0.005157470703125, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.678964614868164e-06, + "grad_norm": 6.022550582885742, + "learning_rate": 6.549385332768122e-07, + "loss": 0.4254, + "mean_token_accuracy": 0.8619884252548218, + "num_tokens": 60085350.0, + "step": 1546 + }, + { + "epoch": 0.19679430097951914, + "ewc_loss": 0.005157470703125, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.678964614868164e-06, + "grad_norm": 5.987746715545654, + "learning_rate": 6.5536244171259e-07, + "loss": 0.4182, + "mean_token_accuracy": 0.8576265573501587, + "num_tokens": 60126107.0, + "step": 1547 + }, + { + "epoch": 0.19692151125810967, + "ewc_loss": 0.005157470703125, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.678964614868164e-06, + "grad_norm": 6.040847301483154, + "learning_rate": 6.55786350148368e-07, + "loss": 0.4264, + "mean_token_accuracy": 0.858232855796814, + "num_tokens": 60164055.0, + "step": 1548 + }, + { + "epoch": 0.19704872153670017, + "ewc_loss": 0.005218505859375, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.738569259643555e-06, + "grad_norm": 5.982326030731201, + "learning_rate": 6.562102585841458e-07, + "loss": 0.3801, + "mean_token_accuracy": 0.8692129254341125, + "num_tokens": 60203818.0, + "step": 1549 + }, + { + "epoch": 0.19717593181529067, + "ewc_loss": 0.00518798828125, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.708766937255859e-06, + "grad_norm": 5.966630935668945, + "learning_rate": 6.566341670199236e-07, + "loss": 0.3834, + "mean_token_accuracy": 0.8704637289047241, + "num_tokens": 60244467.0, + "step": 1550 + }, + { + "epoch": 0.1973031420938812, + "ewc_loss": 0.005218505859375, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.738569259643555e-06, + "grad_norm": 5.931909084320068, + "learning_rate": 6.570580754557016e-07, + "loss": 0.3703, + "mean_token_accuracy": 0.8767067193984985, + "num_tokens": 60284788.0, + "step": 1551 + }, + { + "epoch": 0.1974303523724717, + "ewc_loss": 0.005157470703125, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.678964614868164e-06, + "grad_norm": 6.088113784790039, + "learning_rate": 6.574819838914794e-07, + "loss": 0.3835, + "mean_token_accuracy": 0.871519148349762, + "num_tokens": 60319168.0, + "step": 1552 + }, + { + "epoch": 0.1975575626510622, + "ewc_loss": 0.0052490234375, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.76837158203125e-06, + "grad_norm": 5.988832473754883, + "learning_rate": 6.579058923272573e-07, + "loss": 0.4184, + "mean_token_accuracy": 0.8596848249435425, + "num_tokens": 60360768.0, + "step": 1553 + }, + { + "epoch": 0.19768477292965272, + "ewc_loss": 0.005218505859375, + "ewc_loss_diag": 5.0067901611328125e-06, + "ewc_loss_parallel": 4.738569259643555e-06, + "grad_norm": 6.165142059326172, + "learning_rate": 6.583298007630351e-07, + "loss": 0.4158, + "mean_token_accuracy": 0.8622553944587708, + "num_tokens": 60393356.0, + "step": 1554 + }, + { + "epoch": 0.19781198320824323, + "ewc_loss": 0.00531005859375, + "ewc_loss_diag": 5.036592483520508e-06, + "ewc_loss_parallel": 4.798173904418945e-06, + "grad_norm": 5.976004123687744, + "learning_rate": 6.58753709198813e-07, + "loss": 0.4169, + "mean_token_accuracy": 0.8612867593765259, + "num_tokens": 60437717.0, + "step": 1555 + }, + { + "epoch": 0.19793919348683373, + "ewc_loss": 0.00518798828125, + "ewc_loss_diag": 5.036592483520508e-06, + "ewc_loss_parallel": 4.678964614868164e-06, + "grad_norm": 5.920495510101318, + "learning_rate": 6.591776176345909e-07, + "loss": 0.3899, + "mean_token_accuracy": 0.8689596652984619, + "num_tokens": 60481868.0, + "step": 1556 + }, + { + "epoch": 0.19806640376542425, + "ewc_loss": 0.00518798828125, + "ewc_loss_diag": 5.036592483520508e-06, + "ewc_loss_parallel": 4.708766937255859e-06, + "grad_norm": 5.90730094909668, + "learning_rate": 6.596015260703688e-07, + "loss": 0.3905, + "mean_token_accuracy": 0.8704116940498352, + "num_tokens": 60529552.0, + "step": 1557 + }, + { + "epoch": 0.19819361404401475, + "ewc_loss": 0.00518798828125, + "ewc_loss_diag": 5.036592483520508e-06, + "ewc_loss_parallel": 4.708766937255859e-06, + "grad_norm": 5.964028835296631, + "learning_rate": 6.600254345061466e-07, + "loss": 0.3802, + "mean_token_accuracy": 0.8711917400360107, + "num_tokens": 60570118.0, + "step": 1558 + }, + { + "epoch": 0.19832082432260525, + "ewc_loss": 0.0052490234375, + "ewc_loss_diag": 5.036592483520508e-06, + "ewc_loss_parallel": 4.738569259643555e-06, + "grad_norm": 6.002220153808594, + "learning_rate": 6.604493429419246e-07, + "loss": 0.4347, + "mean_token_accuracy": 0.855108380317688, + "num_tokens": 60614180.0, + "step": 1559 + }, + { + "epoch": 0.19844803460119578, + "ewc_loss": 0.0052490234375, + "ewc_loss_diag": 5.036592483520508e-06, + "ewc_loss_parallel": 4.76837158203125e-06, + "grad_norm": 5.952991485595703, + "learning_rate": 6.608732513777023e-07, + "loss": 0.4227, + "mean_token_accuracy": 0.863156795501709, + "num_tokens": 60662834.0, + "step": 1560 + }, + { + "epoch": 0.19857524487978628, + "ewc_loss": 0.005218505859375, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.708766937255859e-06, + "grad_norm": 5.995691299438477, + "learning_rate": 6.612971598134803e-07, + "loss": 0.4006, + "mean_token_accuracy": 0.8640349507331848, + "num_tokens": 60701486.0, + "step": 1561 + }, + { + "epoch": 0.19870245515837678, + "ewc_loss": 0.0052490234375, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.738569259643555e-06, + "grad_norm": 6.188603401184082, + "learning_rate": 6.617210682492581e-07, + "loss": 0.4397, + "mean_token_accuracy": 0.8533936738967896, + "num_tokens": 60739601.0, + "step": 1562 + }, + { + "epoch": 0.1988296654369673, + "ewc_loss": 0.005279541015625, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.76837158203125e-06, + "grad_norm": 6.01418399810791, + "learning_rate": 6.62144976685036e-07, + "loss": 0.3755, + "mean_token_accuracy": 0.8731169104576111, + "num_tokens": 60776925.0, + "step": 1563 + }, + { + "epoch": 0.1989568757155578, + "ewc_loss": 0.005218505859375, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.708766937255859e-06, + "grad_norm": 5.976019859313965, + "learning_rate": 6.625688851208139e-07, + "loss": 0.412, + "mean_token_accuracy": 0.8612103462219238, + "num_tokens": 60815963.0, + "step": 1564 + }, + { + "epoch": 0.19908408599414834, + "ewc_loss": 0.005279541015625, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.76837158203125e-06, + "grad_norm": 6.15159797668457, + "learning_rate": 6.629927935565918e-07, + "loss": 0.4155, + "mean_token_accuracy": 0.8581522703170776, + "num_tokens": 60851390.0, + "step": 1565 + }, + { + "epoch": 0.19921129627273884, + "ewc_loss": 0.00531005859375, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.798173904418945e-06, + "grad_norm": 6.09093713760376, + "learning_rate": 6.634167019923696e-07, + "loss": 0.4566, + "mean_token_accuracy": 0.8541179895401001, + "num_tokens": 60891437.0, + "step": 1566 + }, + { + "epoch": 0.19933850655132934, + "ewc_loss": 0.005279541015625, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.76837158203125e-06, + "grad_norm": 6.178053379058838, + "learning_rate": 6.638406104281476e-07, + "loss": 0.4073, + "mean_token_accuracy": 0.8647490739822388, + "num_tokens": 60924266.0, + "step": 1567 + }, + { + "epoch": 0.19946571682991987, + "ewc_loss": 0.00531005859375, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.798173904418945e-06, + "grad_norm": 5.94624137878418, + "learning_rate": 6.642645188639253e-07, + "loss": 0.3811, + "mean_token_accuracy": 0.8712565898895264, + "num_tokens": 60967481.0, + "step": 1568 + }, + { + "epoch": 0.19959292710851037, + "ewc_loss": 0.005218505859375, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.708766937255859e-06, + "grad_norm": 6.047961235046387, + "learning_rate": 6.646884272997032e-07, + "loss": 0.4311, + "mean_token_accuracy": 0.8534700870513916, + "num_tokens": 61007374.0, + "step": 1569 + }, + { + "epoch": 0.19972013738710087, + "ewc_loss": 0.00531005859375, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.798173904418945e-06, + "grad_norm": 6.064600467681885, + "learning_rate": 6.651123357354811e-07, + "loss": 0.4039, + "mean_token_accuracy": 0.863979160785675, + "num_tokens": 61044717.0, + "step": 1570 + }, + { + "epoch": 0.1998473476656914, + "ewc_loss": 0.00531005859375, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.798173904418945e-06, + "grad_norm": 6.165848255157471, + "learning_rate": 6.655362441712589e-07, + "loss": 0.436, + "mean_token_accuracy": 0.8538325428962708, + "num_tokens": 61079948.0, + "step": 1571 + }, + { + "epoch": 0.1999745579442819, + "ewc_loss": 0.005340576171875, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.827976226806641e-06, + "grad_norm": 6.004271030426025, + "learning_rate": 6.659601526070369e-07, + "loss": 0.4554, + "mean_token_accuracy": 0.8525184392929077, + "num_tokens": 61123953.0, + "step": 1572 + }, + { + "epoch": 0.2001017682228724, + "ewc_loss": 0.00531005859375, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.798173904418945e-06, + "grad_norm": 6.331509590148926, + "learning_rate": 6.663840610428147e-07, + "loss": 0.4844, + "mean_token_accuracy": 0.8454149961471558, + "num_tokens": 61155486.0, + "step": 1573 + }, + { + "epoch": 0.20022897850146293, + "ewc_loss": 0.00537109375, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.857778549194336e-06, + "grad_norm": 5.997081279754639, + "learning_rate": 6.668079694785926e-07, + "loss": 0.36, + "mean_token_accuracy": 0.8800420761108398, + "num_tokens": 61196606.0, + "step": 1574 + }, + { + "epoch": 0.20035618878005343, + "ewc_loss": 0.00531005859375, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.798173904418945e-06, + "grad_norm": 6.111839771270752, + "learning_rate": 6.672318779143704e-07, + "loss": 0.3995, + "mean_token_accuracy": 0.86466383934021, + "num_tokens": 61231028.0, + "step": 1575 + }, + { + "epoch": 0.20048339905864393, + "ewc_loss": 0.00537109375, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.857778549194336e-06, + "grad_norm": 6.317401885986328, + "learning_rate": 6.676557863501483e-07, + "loss": 0.4134, + "mean_token_accuracy": 0.8636929988861084, + "num_tokens": 61263513.0, + "step": 1576 + }, + { + "epoch": 0.20061060933723446, + "ewc_loss": 0.00543212890625, + "ewc_loss_diag": 5.066394805908203e-06, + "ewc_loss_parallel": 4.9173831939697266e-06, + "grad_norm": 6.013064384460449, + "learning_rate": 6.680796947859262e-07, + "loss": 0.4069, + "mean_token_accuracy": 0.8647182583808899, + "num_tokens": 61304885.0, + "step": 1577 + }, + { + "epoch": 0.20073781961582496, + "ewc_loss": 0.00537109375, + "ewc_loss_diag": 5.0961971282958984e-06, + "ewc_loss_parallel": 4.857778549194336e-06, + "grad_norm": 6.085648059844971, + "learning_rate": 6.685036032217041e-07, + "loss": 0.4053, + "mean_token_accuracy": 0.8659348487854004, + "num_tokens": 61341921.0, + "step": 1578 + }, + { + "epoch": 0.20086502989441546, + "ewc_loss": 0.00537109375, + "ewc_loss_diag": 5.0961971282958984e-06, + "ewc_loss_parallel": 4.857778549194336e-06, + "grad_norm": 6.080630302429199, + "learning_rate": 6.689275116574819e-07, + "loss": 0.3867, + "mean_token_accuracy": 0.8719425201416016, + "num_tokens": 61378831.0, + "step": 1579 + }, + { + "epoch": 0.200992240173006, + "ewc_loss": 0.00537109375, + "ewc_loss_diag": 5.0961971282958984e-06, + "ewc_loss_parallel": 4.857778549194336e-06, + "grad_norm": 5.977169036865234, + "learning_rate": 6.693514200932599e-07, + "loss": 0.3981, + "mean_token_accuracy": 0.8677172660827637, + "num_tokens": 61420531.0, + "step": 1580 + }, + { + "epoch": 0.2011194504515965, + "ewc_loss": 0.00537109375, + "ewc_loss_diag": 5.0961971282958984e-06, + "ewc_loss_parallel": 4.857778549194336e-06, + "grad_norm": 6.062674045562744, + "learning_rate": 6.697753285290377e-07, + "loss": 0.4175, + "mean_token_accuracy": 0.8609153032302856, + "num_tokens": 61463165.0, + "step": 1581 + }, + { + "epoch": 0.201246660730187, + "ewc_loss": 0.00537109375, + "ewc_loss_diag": 5.125999450683594e-06, + "ewc_loss_parallel": 4.857778549194336e-06, + "grad_norm": 6.136831760406494, + "learning_rate": 6.701992369648156e-07, + "loss": 0.3401, + "mean_token_accuracy": 0.8850703239440918, + "num_tokens": 61500200.0, + "step": 1582 + }, + { + "epoch": 0.20137387100877752, + "ewc_loss": 0.005401611328125, + "ewc_loss_diag": 5.0961971282958984e-06, + "ewc_loss_parallel": 4.887580871582031e-06, + "grad_norm": 6.010866165161133, + "learning_rate": 6.706231454005934e-07, + "loss": 0.3665, + "mean_token_accuracy": 0.8747535347938538, + "num_tokens": 61541444.0, + "step": 1583 + }, + { + "epoch": 0.20150108128736802, + "ewc_loss": 0.005340576171875, + "ewc_loss_diag": 5.125999450683594e-06, + "ewc_loss_parallel": 4.827976226806641e-06, + "grad_norm": 6.047325611114502, + "learning_rate": 6.710470538363713e-07, + "loss": 0.3922, + "mean_token_accuracy": 0.8684941530227661, + "num_tokens": 61581481.0, + "step": 1584 + }, + { + "epoch": 0.20162829156595852, + "ewc_loss": 0.005401611328125, + "ewc_loss_diag": 5.125999450683594e-06, + "ewc_loss_parallel": 4.887580871582031e-06, + "grad_norm": 6.360403537750244, + "learning_rate": 6.714709622721492e-07, + "loss": 0.4405, + "mean_token_accuracy": 0.8548972010612488, + "num_tokens": 61613479.0, + "step": 1585 + }, + { + "epoch": 0.20175550184454905, + "ewc_loss": 0.00543212890625, + "ewc_loss_diag": 5.125999450683594e-06, + "ewc_loss_parallel": 4.9173831939697266e-06, + "grad_norm": 6.0954813957214355, + "learning_rate": 6.718948707079271e-07, + "loss": 0.3966, + "mean_token_accuracy": 0.8639533519744873, + "num_tokens": 61654295.0, + "step": 1586 + }, + { + "epoch": 0.20188271212313955, + "ewc_loss": 0.00537109375, + "ewc_loss_diag": 5.125999450683594e-06, + "ewc_loss_parallel": 4.857778549194336e-06, + "grad_norm": 6.0319647789001465, + "learning_rate": 6.723187791437049e-07, + "loss": 0.3885, + "mean_token_accuracy": 0.8685443997383118, + "num_tokens": 61692816.0, + "step": 1587 + }, + { + "epoch": 0.20200992240173005, + "ewc_loss": 0.00537109375, + "ewc_loss_diag": 5.125999450683594e-06, + "ewc_loss_parallel": 4.857778549194336e-06, + "grad_norm": 6.090620040893555, + "learning_rate": 6.727426875794829e-07, + "loss": 0.439, + "mean_token_accuracy": 0.8549302816390991, + "num_tokens": 61734900.0, + "step": 1588 + }, + { + "epoch": 0.20213713268032057, + "ewc_loss": 0.00537109375, + "ewc_loss_diag": 5.125999450683594e-06, + "ewc_loss_parallel": 4.857778549194336e-06, + "grad_norm": 5.962200164794922, + "learning_rate": 6.731665960152607e-07, + "loss": 0.405, + "mean_token_accuracy": 0.8621554374694824, + "num_tokens": 61780196.0, + "step": 1589 + }, + { + "epoch": 0.20226434295891108, + "ewc_loss": 0.00531005859375, + "ewc_loss_diag": 5.125999450683594e-06, + "ewc_loss_parallel": 4.798173904418945e-06, + "grad_norm": 6.153162002563477, + "learning_rate": 6.735905044510385e-07, + "loss": 0.4346, + "mean_token_accuracy": 0.8494548201560974, + "num_tokens": 61819252.0, + "step": 1590 + }, + { + "epoch": 0.2023915532375016, + "ewc_loss": 0.00537109375, + "ewc_loss_diag": 5.125999450683594e-06, + "ewc_loss_parallel": 4.857778549194336e-06, + "grad_norm": 6.192855358123779, + "learning_rate": 6.740144128868164e-07, + "loss": 0.4277, + "mean_token_accuracy": 0.8555896282196045, + "num_tokens": 61856727.0, + "step": 1591 + }, + { + "epoch": 0.2025187635160921, + "ewc_loss": 0.005401611328125, + "ewc_loss_diag": 5.125999450683594e-06, + "ewc_loss_parallel": 4.887580871582031e-06, + "grad_norm": 6.096778869628906, + "learning_rate": 6.744383213225942e-07, + "loss": 0.4487, + "mean_token_accuracy": 0.8489221334457397, + "num_tokens": 61896707.0, + "step": 1592 + }, + { + "epoch": 0.2026459737946826, + "ewc_loss": 0.00537109375, + "ewc_loss_diag": 5.125999450683594e-06, + "ewc_loss_parallel": 4.857778549194336e-06, + "grad_norm": 6.145684719085693, + "learning_rate": 6.748622297583722e-07, + "loss": 0.3929, + "mean_token_accuracy": 0.8668311834335327, + "num_tokens": 61929758.0, + "step": 1593 + }, + { + "epoch": 0.20277318407327313, + "ewc_loss": 0.005401611328125, + "ewc_loss_diag": 5.185604095458984e-06, + "ewc_loss_parallel": 4.887580871582031e-06, + "grad_norm": 6.08939266204834, + "learning_rate": 6.7528613819415e-07, + "loss": 0.3915, + "mean_token_accuracy": 0.8680598139762878, + "num_tokens": 61967733.0, + "step": 1594 + }, + { + "epoch": 0.20290039435186363, + "ewc_loss": 0.005401611328125, + "ewc_loss_diag": 5.185604095458984e-06, + "ewc_loss_parallel": 4.887580871582031e-06, + "grad_norm": 6.026584148406982, + "learning_rate": 6.757100466299279e-07, + "loss": 0.408, + "mean_token_accuracy": 0.8643550872802734, + "num_tokens": 62009238.0, + "step": 1595 + }, + { + "epoch": 0.20302760463045413, + "ewc_loss": 0.005340576171875, + "ewc_loss_diag": 5.185604095458984e-06, + "ewc_loss_parallel": 4.827976226806641e-06, + "grad_norm": 6.423089504241943, + "learning_rate": 6.761339550657058e-07, + "loss": 0.4257, + "mean_token_accuracy": 0.8550811409950256, + "num_tokens": 62037173.0, + "step": 1596 + }, + { + "epoch": 0.20315481490904466, + "ewc_loss": 0.005462646484375, + "ewc_loss_diag": 5.185604095458984e-06, + "ewc_loss_parallel": 4.947185516357422e-06, + "grad_norm": 6.069338798522949, + "learning_rate": 6.765578635014837e-07, + "loss": 0.4005, + "mean_token_accuracy": 0.867901086807251, + "num_tokens": 62077671.0, + "step": 1597 + }, + { + "epoch": 0.20328202518763516, + "ewc_loss": 0.005401611328125, + "ewc_loss_diag": 5.21540641784668e-06, + "ewc_loss_parallel": 4.887580871582031e-06, + "grad_norm": 6.212724208831787, + "learning_rate": 6.769817719372614e-07, + "loss": 0.3891, + "mean_token_accuracy": 0.869122326374054, + "num_tokens": 62112722.0, + "step": 1598 + }, + { + "epoch": 0.20340923546622566, + "ewc_loss": 0.005462646484375, + "ewc_loss_diag": 5.245208740234375e-06, + "ewc_loss_parallel": 4.947185516357422e-06, + "grad_norm": 6.196064472198486, + "learning_rate": 6.774056803730394e-07, + "loss": 0.4206, + "mean_token_accuracy": 0.8601609468460083, + "num_tokens": 62147682.0, + "step": 1599 + }, + { + "epoch": 0.2035364457448162, + "ewc_loss": 0.00543212890625, + "ewc_loss_diag": 5.245208740234375e-06, + "ewc_loss_parallel": 4.9173831939697266e-06, + "grad_norm": 6.036263942718506, + "learning_rate": 6.778295888088172e-07, + "loss": 0.4014, + "mean_token_accuracy": 0.8648130297660828, + "num_tokens": 62186528.0, + "step": 1600 + }, + { + "epoch": 0.2036636560234067, + "ewc_loss": 0.00543212890625, + "ewc_loss_diag": 5.245208740234375e-06, + "ewc_loss_parallel": 4.9173831939697266e-06, + "grad_norm": 6.1012420654296875, + "learning_rate": 6.782534972445952e-07, + "loss": 0.4145, + "mean_token_accuracy": 0.8597425222396851, + "num_tokens": 62224125.0, + "step": 1601 + }, + { + "epoch": 0.2037908663019972, + "ewc_loss": 0.00543212890625, + "ewc_loss_diag": 5.245208740234375e-06, + "ewc_loss_parallel": 4.9173831939697266e-06, + "grad_norm": 6.268595218658447, + "learning_rate": 6.78677405680373e-07, + "loss": 0.4168, + "mean_token_accuracy": 0.8610999584197998, + "num_tokens": 62264887.0, + "step": 1602 + }, + { + "epoch": 0.20391807658058772, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.245208740234375e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.126204013824463, + "learning_rate": 6.791013141161509e-07, + "loss": 0.4104, + "mean_token_accuracy": 0.8613526225090027, + "num_tokens": 62304774.0, + "step": 1603 + }, + { + "epoch": 0.20404528685917822, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.245208740234375e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.098347187042236, + "learning_rate": 6.795252225519288e-07, + "loss": 0.4348, + "mean_token_accuracy": 0.8534947633743286, + "num_tokens": 62348268.0, + "step": 1604 + }, + { + "epoch": 0.20417249713776872, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.27501106262207e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.112362861633301, + "learning_rate": 6.799491309877067e-07, + "loss": 0.4018, + "mean_token_accuracy": 0.862622857093811, + "num_tokens": 62386227.0, + "step": 1605 + }, + { + "epoch": 0.20429970741635925, + "ewc_loss": 0.005462646484375, + "ewc_loss_diag": 5.245208740234375e-06, + "ewc_loss_parallel": 4.947185516357422e-06, + "grad_norm": 6.321292877197266, + "learning_rate": 6.803730394234844e-07, + "loss": 0.3827, + "mean_token_accuracy": 0.8702839612960815, + "num_tokens": 62419625.0, + "step": 1606 + }, + { + "epoch": 0.20442691769494975, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.245208740234375e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.166925430297852, + "learning_rate": 6.807969478592624e-07, + "loss": 0.4042, + "mean_token_accuracy": 0.867073655128479, + "num_tokens": 62456424.0, + "step": 1607 + }, + { + "epoch": 0.20455412797354025, + "ewc_loss": 0.005462646484375, + "ewc_loss_diag": 5.245208740234375e-06, + "ewc_loss_parallel": 4.947185516357422e-06, + "grad_norm": 6.076776027679443, + "learning_rate": 6.812208562950402e-07, + "loss": 0.438, + "mean_token_accuracy": 0.8592698574066162, + "num_tokens": 62499888.0, + "step": 1608 + }, + { + "epoch": 0.20468133825213078, + "ewc_loss": 0.005462646484375, + "ewc_loss_diag": 5.27501106262207e-06, + "ewc_loss_parallel": 4.947185516357422e-06, + "grad_norm": 6.088483810424805, + "learning_rate": 6.816447647308182e-07, + "loss": 0.4615, + "mean_token_accuracy": 0.847052812576294, + "num_tokens": 62545209.0, + "step": 1609 + }, + { + "epoch": 0.20480854853072128, + "ewc_loss": 0.005462646484375, + "ewc_loss_diag": 5.27501106262207e-06, + "ewc_loss_parallel": 4.947185516357422e-06, + "grad_norm": 6.249088764190674, + "learning_rate": 6.82068673166596e-07, + "loss": 0.4202, + "mean_token_accuracy": 0.861052930355072, + "num_tokens": 62580406.0, + "step": 1610 + }, + { + "epoch": 0.20493575880931178, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.27501106262207e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.119684219360352, + "learning_rate": 6.824925816023738e-07, + "loss": 0.3995, + "mean_token_accuracy": 0.864145040512085, + "num_tokens": 62620070.0, + "step": 1611 + }, + { + "epoch": 0.2050629690879023, + "ewc_loss": 0.005462646484375, + "ewc_loss_diag": 5.27501106262207e-06, + "ewc_loss_parallel": 4.947185516357422e-06, + "grad_norm": 6.009908199310303, + "learning_rate": 6.829164900381518e-07, + "loss": 0.3847, + "mean_token_accuracy": 0.8706623315811157, + "num_tokens": 62665818.0, + "step": 1612 + }, + { + "epoch": 0.2051901793664928, + "ewc_loss": 0.00543212890625, + "ewc_loss_diag": 5.27501106262207e-06, + "ewc_loss_parallel": 4.9173831939697266e-06, + "grad_norm": 6.129003524780273, + "learning_rate": 6.833403984739295e-07, + "loss": 0.3913, + "mean_token_accuracy": 0.8668539524078369, + "num_tokens": 62702577.0, + "step": 1613 + }, + { + "epoch": 0.2053173896450833, + "ewc_loss": 0.005462646484375, + "ewc_loss_diag": 5.27501106262207e-06, + "ewc_loss_parallel": 4.947185516357422e-06, + "grad_norm": 6.2157673835754395, + "learning_rate": 6.837643069097074e-07, + "loss": 0.4693, + "mean_token_accuracy": 0.8466298580169678, + "num_tokens": 62741198.0, + "step": 1614 + }, + { + "epoch": 0.20544459992367384, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.27501106262207e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.138378143310547, + "learning_rate": 6.841882153454853e-07, + "loss": 0.3828, + "mean_token_accuracy": 0.8706502914428711, + "num_tokens": 62777227.0, + "step": 1615 + }, + { + "epoch": 0.20557181020226434, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.27501106262207e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.080423831939697, + "learning_rate": 6.846121237812632e-07, + "loss": 0.3632, + "mean_token_accuracy": 0.8769016861915588, + "num_tokens": 62814312.0, + "step": 1616 + }, + { + "epoch": 0.20569902048085487, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.27501106262207e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 5.973503112792969, + "learning_rate": 6.850360322170411e-07, + "loss": 0.3853, + "mean_token_accuracy": 0.8702453970909119, + "num_tokens": 62857779.0, + "step": 1617 + }, + { + "epoch": 0.20582623075944537, + "ewc_loss": 0.00543212890625, + "ewc_loss_diag": 5.27501106262207e-06, + "ewc_loss_parallel": 4.9173831939697266e-06, + "grad_norm": 6.352527141571045, + "learning_rate": 6.85459940652819e-07, + "loss": 0.4043, + "mean_token_accuracy": 0.8645634651184082, + "num_tokens": 62891240.0, + "step": 1618 + }, + { + "epoch": 0.20595344103803587, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.27501106262207e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.2129225730896, + "learning_rate": 6.858838490885968e-07, + "loss": 0.4517, + "mean_token_accuracy": 0.8495949506759644, + "num_tokens": 62933177.0, + "step": 1619 + }, + { + "epoch": 0.2060806513166264, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.27501106262207e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.215043544769287, + "learning_rate": 6.863077575243748e-07, + "loss": 0.454, + "mean_token_accuracy": 0.8508703708648682, + "num_tokens": 62971431.0, + "step": 1620 + }, + { + "epoch": 0.2062078615952169, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.262197494506836, + "learning_rate": 6.867316659601525e-07, + "loss": 0.4036, + "mean_token_accuracy": 0.8632416725158691, + "num_tokens": 63007597.0, + "step": 1621 + }, + { + "epoch": 0.2063350718738074, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.03057861328125, + "learning_rate": 6.871555743959304e-07, + "loss": 0.4098, + "mean_token_accuracy": 0.8644163608551025, + "num_tokens": 63049810.0, + "step": 1622 + }, + { + "epoch": 0.20646228215239792, + "ewc_loss": 0.00543212890625, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 4.9173831939697266e-06, + "grad_norm": 6.14601993560791, + "learning_rate": 6.875794828317083e-07, + "loss": 0.3854, + "mean_token_accuracy": 0.8685116171836853, + "num_tokens": 63089331.0, + "step": 1623 + }, + { + "epoch": 0.20658949243098843, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.04490327835083, + "learning_rate": 6.880033912674862e-07, + "loss": 0.3744, + "mean_token_accuracy": 0.874204158782959, + "num_tokens": 63136150.0, + "step": 1624 + }, + { + "epoch": 0.20671670270957893, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.183587074279785, + "learning_rate": 6.884272997032641e-07, + "loss": 0.4732, + "mean_token_accuracy": 0.8434329032897949, + "num_tokens": 63177089.0, + "step": 1625 + }, + { + "epoch": 0.20684391298816945, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.223351955413818, + "learning_rate": 6.88851208139042e-07, + "loss": 0.3744, + "mean_token_accuracy": 0.8752951622009277, + "num_tokens": 63212555.0, + "step": 1626 + }, + { + "epoch": 0.20697112326675995, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 5.036592483520508e-06, + "grad_norm": 6.1535139083862305, + "learning_rate": 6.892751165748198e-07, + "loss": 0.4199, + "mean_token_accuracy": 0.8593984842300415, + "num_tokens": 63254033.0, + "step": 1627 + }, + { + "epoch": 0.20709833354535045, + "ewc_loss": 0.005462646484375, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 4.947185516357422e-06, + "grad_norm": 6.127734661102295, + "learning_rate": 6.896990250105978e-07, + "loss": 0.3816, + "mean_token_accuracy": 0.874426543712616, + "num_tokens": 63294817.0, + "step": 1628 + }, + { + "epoch": 0.20722554382394098, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.408119201660156, + "learning_rate": 6.901229334463755e-07, + "loss": 0.3995, + "mean_token_accuracy": 0.8649258613586426, + "num_tokens": 63329714.0, + "step": 1629 + }, + { + "epoch": 0.20735275410253148, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.133054256439209, + "learning_rate": 6.905468418821534e-07, + "loss": 0.4145, + "mean_token_accuracy": 0.8680101633071899, + "num_tokens": 63370724.0, + "step": 1630 + }, + { + "epoch": 0.20747996438112198, + "ewc_loss": 0.005462646484375, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 4.947185516357422e-06, + "grad_norm": 6.169742107391357, + "learning_rate": 6.909707503179313e-07, + "loss": 0.4063, + "mean_token_accuracy": 0.8609780073165894, + "num_tokens": 63410946.0, + "step": 1631 + }, + { + "epoch": 0.2076071746597125, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.117840766906738, + "learning_rate": 6.913946587537091e-07, + "loss": 0.3987, + "mean_token_accuracy": 0.8659399151802063, + "num_tokens": 63449898.0, + "step": 1632 + }, + { + "epoch": 0.207734384938303, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.272558689117432, + "learning_rate": 6.918185671894871e-07, + "loss": 0.4652, + "mean_token_accuracy": 0.844770073890686, + "num_tokens": 63488466.0, + "step": 1633 + }, + { + "epoch": 0.2078615952168935, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.484106063842773, + "learning_rate": 6.922424756252649e-07, + "loss": 0.4505, + "mean_token_accuracy": 0.8603677153587341, + "num_tokens": 63523873.0, + "step": 1634 + }, + { + "epoch": 0.20798880549548404, + "ewc_loss": 0.005584716796875, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 5.066394805908203e-06, + "grad_norm": 6.119683742523193, + "learning_rate": 6.926663840610428e-07, + "loss": 0.3604, + "mean_token_accuracy": 0.8771561980247498, + "num_tokens": 63562774.0, + "step": 1635 + }, + { + "epoch": 0.20811601577407454, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.304813385009766e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.129828453063965, + "learning_rate": 6.930902924968206e-07, + "loss": 0.3934, + "mean_token_accuracy": 0.8633259534835815, + "num_tokens": 63602353.0, + "step": 1636 + }, + { + "epoch": 0.20824322605266504, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.334615707397461e-06, + "ewc_loss_parallel": 5.036592483520508e-06, + "grad_norm": 6.256511688232422, + "learning_rate": 6.935142009325985e-07, + "loss": 0.4118, + "mean_token_accuracy": 0.8620789051055908, + "num_tokens": 63644352.0, + "step": 1637 + }, + { + "epoch": 0.20837043633125557, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.364418029785156e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.124650478363037, + "learning_rate": 6.939381093683764e-07, + "loss": 0.4262, + "mean_token_accuracy": 0.8586480617523193, + "num_tokens": 63684883.0, + "step": 1638 + }, + { + "epoch": 0.20849764660984607, + "ewc_loss": 0.0054931640625, + "ewc_loss_diag": 5.364418029785156e-06, + "ewc_loss_parallel": 4.947185516357422e-06, + "grad_norm": 6.071757793426514, + "learning_rate": 6.943620178041543e-07, + "loss": 0.364, + "mean_token_accuracy": 0.8772110939025879, + "num_tokens": 63724677.0, + "step": 1639 + }, + { + "epoch": 0.2086248568884366, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.364418029785156e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.2642621994018555, + "learning_rate": 6.947859262399321e-07, + "loss": 0.4001, + "mean_token_accuracy": 0.8729603290557861, + "num_tokens": 63761861.0, + "step": 1640 + }, + { + "epoch": 0.2087520671670271, + "ewc_loss": 0.005615234375, + "ewc_loss_diag": 5.334615707397461e-06, + "ewc_loss_parallel": 5.066394805908203e-06, + "grad_norm": 6.209449291229248, + "learning_rate": 6.952098346757101e-07, + "loss": 0.4027, + "mean_token_accuracy": 0.8634772300720215, + "num_tokens": 63799207.0, + "step": 1641 + }, + { + "epoch": 0.2088792774456176, + "ewc_loss": 0.005584716796875, + "ewc_loss_diag": 5.364418029785156e-06, + "ewc_loss_parallel": 5.036592483520508e-06, + "grad_norm": 6.1339216232299805, + "learning_rate": 6.956337431114879e-07, + "loss": 0.3852, + "mean_token_accuracy": 0.8725076913833618, + "num_tokens": 63838839.0, + "step": 1642 + }, + { + "epoch": 0.20900648772420813, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.364418029785156e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.331306457519531, + "learning_rate": 6.960576515472658e-07, + "loss": 0.4203, + "mean_token_accuracy": 0.8619334101676941, + "num_tokens": 63880293.0, + "step": 1643 + }, + { + "epoch": 0.20913369800279863, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.364418029785156e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.257074356079102, + "learning_rate": 6.964815599830436e-07, + "loss": 0.4145, + "mean_token_accuracy": 0.8599879741668701, + "num_tokens": 63915510.0, + "step": 1644 + }, + { + "epoch": 0.20926090828138913, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.364418029785156e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.160720348358154, + "learning_rate": 6.969054684188215e-07, + "loss": 0.4627, + "mean_token_accuracy": 0.8441938161849976, + "num_tokens": 63959353.0, + "step": 1645 + }, + { + "epoch": 0.20938811855997966, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.3942203521728516e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.125194549560547, + "learning_rate": 6.973293768545994e-07, + "loss": 0.3752, + "mean_token_accuracy": 0.8736690282821655, + "num_tokens": 64000777.0, + "step": 1646 + }, + { + "epoch": 0.20951532883857016, + "ewc_loss": 0.005584716796875, + "ewc_loss_diag": 5.3942203521728516e-06, + "ewc_loss_parallel": 5.036592483520508e-06, + "grad_norm": 6.085143089294434, + "learning_rate": 6.977532852903773e-07, + "loss": 0.3998, + "mean_token_accuracy": 0.8639878034591675, + "num_tokens": 64046579.0, + "step": 1647 + }, + { + "epoch": 0.20964253911716066, + "ewc_loss": 0.005584716796875, + "ewc_loss_diag": 5.3942203521728516e-06, + "ewc_loss_parallel": 5.036592483520508e-06, + "grad_norm": 6.291959285736084, + "learning_rate": 6.981771937261551e-07, + "loss": 0.4548, + "mean_token_accuracy": 0.8492467403411865, + "num_tokens": 64082623.0, + "step": 1648 + }, + { + "epoch": 0.2097697493957512, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.3942203521728516e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.224257469177246, + "learning_rate": 6.986011021619331e-07, + "loss": 0.4115, + "mean_token_accuracy": 0.8606016635894775, + "num_tokens": 64123324.0, + "step": 1649 + }, + { + "epoch": 0.2098969596743417, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.3942203521728516e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.264763355255127, + "learning_rate": 6.990250105977109e-07, + "loss": 0.4195, + "mean_token_accuracy": 0.8615331649780273, + "num_tokens": 64161549.0, + "step": 1650 + }, + { + "epoch": 0.2100241699529322, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.3942203521728516e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.10119104385376, + "learning_rate": 6.994489190334886e-07, + "loss": 0.395, + "mean_token_accuracy": 0.8651626706123352, + "num_tokens": 64204136.0, + "step": 1651 + }, + { + "epoch": 0.21015138023152272, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.424022674560547e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.106308460235596, + "learning_rate": 6.998728274692666e-07, + "loss": 0.3609, + "mean_token_accuracy": 0.8776428699493408, + "num_tokens": 64245664.0, + "step": 1652 + }, + { + "epoch": 0.21027859051011322, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.424022674560547e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.238117218017578, + "learning_rate": 7.002967359050444e-07, + "loss": 0.3963, + "mean_token_accuracy": 0.8748961091041565, + "num_tokens": 64282569.0, + "step": 1653 + }, + { + "epoch": 0.21040580078870372, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.424022674560547e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.180049419403076, + "learning_rate": 7.007206443408224e-07, + "loss": 0.3542, + "mean_token_accuracy": 0.879608154296875, + "num_tokens": 64319869.0, + "step": 1654 + }, + { + "epoch": 0.21053301106729425, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.424022674560547e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.197930335998535, + "learning_rate": 7.011445527766002e-07, + "loss": 0.3962, + "mean_token_accuracy": 0.8665411472320557, + "num_tokens": 64358100.0, + "step": 1655 + }, + { + "epoch": 0.21066022134588475, + "ewc_loss": 0.005584716796875, + "ewc_loss_diag": 5.424022674560547e-06, + "ewc_loss_parallel": 5.036592483520508e-06, + "grad_norm": 6.155787944793701, + "learning_rate": 7.015684612123781e-07, + "loss": 0.3877, + "mean_token_accuracy": 0.8680003881454468, + "num_tokens": 64400740.0, + "step": 1656 + }, + { + "epoch": 0.21078743162447525, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.424022674560547e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.1604323387146, + "learning_rate": 7.01992369648156e-07, + "loss": 0.4152, + "mean_token_accuracy": 0.8605560064315796, + "num_tokens": 64445610.0, + "step": 1657 + }, + { + "epoch": 0.21091464190306577, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.424022674560547e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.428022861480713, + "learning_rate": 7.024162780839339e-07, + "loss": 0.4504, + "mean_token_accuracy": 0.8446197509765625, + "num_tokens": 64477588.0, + "step": 1658 + }, + { + "epoch": 0.21104185218165628, + "ewc_loss": 0.005706787109375, + "ewc_loss_diag": 5.424022674560547e-06, + "ewc_loss_parallel": 5.155801773071289e-06, + "grad_norm": 6.159511089324951, + "learning_rate": 7.028401865197116e-07, + "loss": 0.4307, + "mean_token_accuracy": 0.8530232906341553, + "num_tokens": 64521572.0, + "step": 1659 + }, + { + "epoch": 0.21116906246024678, + "ewc_loss": 0.005584716796875, + "ewc_loss_diag": 5.424022674560547e-06, + "ewc_loss_parallel": 5.036592483520508e-06, + "grad_norm": 6.063791275024414, + "learning_rate": 7.032640949554896e-07, + "loss": 0.3506, + "mean_token_accuracy": 0.877713680267334, + "num_tokens": 64566247.0, + "step": 1660 + }, + { + "epoch": 0.2112962727388373, + "ewc_loss": 0.005584716796875, + "ewc_loss_diag": 5.424022674560547e-06, + "ewc_loss_parallel": 5.036592483520508e-06, + "grad_norm": 6.3334550857543945, + "learning_rate": 7.036880033912674e-07, + "loss": 0.4525, + "mean_token_accuracy": 0.8487650156021118, + "num_tokens": 64606363.0, + "step": 1661 + }, + { + "epoch": 0.2114234830174278, + "ewc_loss": 0.005615234375, + "ewc_loss_diag": 5.453824996948242e-06, + "ewc_loss_parallel": 5.066394805908203e-06, + "grad_norm": 6.264466762542725, + "learning_rate": 7.041119118270454e-07, + "loss": 0.4051, + "mean_token_accuracy": 0.8628948330879211, + "num_tokens": 64643721.0, + "step": 1662 + }, + { + "epoch": 0.2115506932960183, + "ewc_loss": 0.005523681640625, + "ewc_loss_diag": 5.453824996948242e-06, + "ewc_loss_parallel": 4.976987838745117e-06, + "grad_norm": 6.193717002868652, + "learning_rate": 7.045358202628232e-07, + "loss": 0.4282, + "mean_token_accuracy": 0.8581827282905579, + "num_tokens": 64684096.0, + "step": 1663 + }, + { + "epoch": 0.21167790357460883, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.453824996948242e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.258746147155762, + "learning_rate": 7.049597286986011e-07, + "loss": 0.4243, + "mean_token_accuracy": 0.8583820462226868, + "num_tokens": 64724029.0, + "step": 1664 + }, + { + "epoch": 0.21180511385319933, + "ewc_loss": 0.005615234375, + "ewc_loss_diag": 5.4836273193359375e-06, + "ewc_loss_parallel": 5.066394805908203e-06, + "grad_norm": 6.358983516693115, + "learning_rate": 7.05383637134379e-07, + "loss": 0.3944, + "mean_token_accuracy": 0.8671046495437622, + "num_tokens": 64758840.0, + "step": 1665 + }, + { + "epoch": 0.21193232413178986, + "ewc_loss": 0.005615234375, + "ewc_loss_diag": 5.4836273193359375e-06, + "ewc_loss_parallel": 5.066394805908203e-06, + "grad_norm": 6.262275695800781, + "learning_rate": 7.058075455701568e-07, + "loss": 0.4183, + "mean_token_accuracy": 0.855576753616333, + "num_tokens": 64799961.0, + "step": 1666 + }, + { + "epoch": 0.21205953441038036, + "ewc_loss": 0.005615234375, + "ewc_loss_diag": 5.4836273193359375e-06, + "ewc_loss_parallel": 5.066394805908203e-06, + "grad_norm": 6.214540958404541, + "learning_rate": 7.062314540059346e-07, + "loss": 0.3719, + "mean_token_accuracy": 0.8752078413963318, + "num_tokens": 64838496.0, + "step": 1667 + }, + { + "epoch": 0.21218674468897086, + "ewc_loss": 0.005584716796875, + "ewc_loss_diag": 5.4836273193359375e-06, + "ewc_loss_parallel": 5.036592483520508e-06, + "grad_norm": 6.296034336090088, + "learning_rate": 7.066553624417126e-07, + "loss": 0.3949, + "mean_token_accuracy": 0.8675984144210815, + "num_tokens": 64873435.0, + "step": 1668 + }, + { + "epoch": 0.2123139549675614, + "ewc_loss": 0.005615234375, + "ewc_loss_diag": 5.4836273193359375e-06, + "ewc_loss_parallel": 5.066394805908203e-06, + "grad_norm": 6.197988986968994, + "learning_rate": 7.070792708774904e-07, + "loss": 0.4032, + "mean_token_accuracy": 0.8657006025314331, + "num_tokens": 64916314.0, + "step": 1669 + }, + { + "epoch": 0.2124411652461519, + "ewc_loss": 0.005584716796875, + "ewc_loss_diag": 5.513429641723633e-06, + "ewc_loss_parallel": 5.036592483520508e-06, + "grad_norm": 6.100146770477295, + "learning_rate": 7.075031793132684e-07, + "loss": 0.426, + "mean_token_accuracy": 0.8598597049713135, + "num_tokens": 64964460.0, + "step": 1670 + }, + { + "epoch": 0.2125683755247424, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.513429641723633e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.330128192901611, + "learning_rate": 7.079270877490462e-07, + "loss": 0.3856, + "mean_token_accuracy": 0.8695122003555298, + "num_tokens": 65000524.0, + "step": 1671 + }, + { + "epoch": 0.21269558580333292, + "ewc_loss": 0.005645751953125, + "ewc_loss_diag": 5.513429641723633e-06, + "ewc_loss_parallel": 5.0961971282958984e-06, + "grad_norm": 6.2774248123168945, + "learning_rate": 7.08350996184824e-07, + "loss": 0.4677, + "mean_token_accuracy": 0.8463637828826904, + "num_tokens": 65042076.0, + "step": 1672 + }, + { + "epoch": 0.21282279608192342, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.513429641723633e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.26336669921875, + "learning_rate": 7.08774904620602e-07, + "loss": 0.3779, + "mean_token_accuracy": 0.8754735589027405, + "num_tokens": 65080406.0, + "step": 1673 + }, + { + "epoch": 0.21295000636051392, + "ewc_loss": 0.005584716796875, + "ewc_loss_diag": 5.513429641723633e-06, + "ewc_loss_parallel": 5.036592483520508e-06, + "grad_norm": 6.242234706878662, + "learning_rate": 7.091988130563797e-07, + "loss": 0.4324, + "mean_token_accuracy": 0.8545389175415039, + "num_tokens": 65121239.0, + "step": 1674 + }, + { + "epoch": 0.21307721663910445, + "ewc_loss": 0.005584716796875, + "ewc_loss_diag": 5.513429641723633e-06, + "ewc_loss_parallel": 5.036592483520508e-06, + "grad_norm": 6.283698081970215, + "learning_rate": 7.096227214921576e-07, + "loss": 0.4263, + "mean_token_accuracy": 0.8571957349777222, + "num_tokens": 65160202.0, + "step": 1675 + }, + { + "epoch": 0.21320442691769495, + "ewc_loss": 0.005645751953125, + "ewc_loss_diag": 5.513429641723633e-06, + "ewc_loss_parallel": 5.0961971282958984e-06, + "grad_norm": 6.320886135101318, + "learning_rate": 7.100466299279355e-07, + "loss": 0.4089, + "mean_token_accuracy": 0.8636741638183594, + "num_tokens": 65195114.0, + "step": 1676 + }, + { + "epoch": 0.21333163719628545, + "ewc_loss": 0.005645751953125, + "ewc_loss_diag": 5.513429641723633e-06, + "ewc_loss_parallel": 5.0961971282958984e-06, + "grad_norm": 6.2703142166137695, + "learning_rate": 7.104705383637134e-07, + "loss": 0.4208, + "mean_token_accuracy": 0.8631996512413025, + "num_tokens": 65234587.0, + "step": 1677 + }, + { + "epoch": 0.21345884747487598, + "ewc_loss": 0.005584716796875, + "ewc_loss_diag": 5.543231964111328e-06, + "ewc_loss_parallel": 5.036592483520508e-06, + "grad_norm": 6.235283851623535, + "learning_rate": 7.108944467994913e-07, + "loss": 0.3948, + "mean_token_accuracy": 0.8676197528839111, + "num_tokens": 65274125.0, + "step": 1678 + }, + { + "epoch": 0.21358605775346648, + "ewc_loss": 0.005615234375, + "ewc_loss_diag": 5.543231964111328e-06, + "ewc_loss_parallel": 5.066394805908203e-06, + "grad_norm": 6.368505954742432, + "learning_rate": 7.113183552352692e-07, + "loss": 0.4409, + "mean_token_accuracy": 0.8521323800086975, + "num_tokens": 65313082.0, + "step": 1679 + }, + { + "epoch": 0.21371326803205698, + "ewc_loss": 0.005645751953125, + "ewc_loss_diag": 5.543231964111328e-06, + "ewc_loss_parallel": 5.0961971282958984e-06, + "grad_norm": 6.2458648681640625, + "learning_rate": 7.11742263671047e-07, + "loss": 0.3639, + "mean_token_accuracy": 0.8760741949081421, + "num_tokens": 65351766.0, + "step": 1680 + }, + { + "epoch": 0.2138404783106475, + "ewc_loss": 0.00555419921875, + "ewc_loss_diag": 5.543231964111328e-06, + "ewc_loss_parallel": 5.0067901611328125e-06, + "grad_norm": 6.498749256134033, + "learning_rate": 7.12166172106825e-07, + "loss": 0.4928, + "mean_token_accuracy": 0.8376657962799072, + "num_tokens": 65387127.0, + "step": 1681 + }, + { + "epoch": 0.213967688589238, + "ewc_loss": 0.005706787109375, + "ewc_loss_diag": 5.543231964111328e-06, + "ewc_loss_parallel": 5.155801773071289e-06, + "grad_norm": 6.306008815765381, + "learning_rate": 7.125900805426027e-07, + "loss": 0.3796, + "mean_token_accuracy": 0.8714004158973694, + "num_tokens": 65422163.0, + "step": 1682 + }, + { + "epoch": 0.2140948988678285, + "ewc_loss": 0.005615234375, + "ewc_loss_diag": 5.543231964111328e-06, + "ewc_loss_parallel": 5.066394805908203e-06, + "grad_norm": 6.310454845428467, + "learning_rate": 7.130139889783806e-07, + "loss": 0.4007, + "mean_token_accuracy": 0.8669217228889465, + "num_tokens": 65458681.0, + "step": 1683 + }, + { + "epoch": 0.21422210914641904, + "ewc_loss": 0.00567626953125, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.125999450683594e-06, + "grad_norm": 6.289297580718994, + "learning_rate": 7.134378974141585e-07, + "loss": 0.3911, + "mean_token_accuracy": 0.8669195175170898, + "num_tokens": 65497199.0, + "step": 1684 + }, + { + "epoch": 0.21434931942500954, + "ewc_loss": 0.005645751953125, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.0961971282958984e-06, + "grad_norm": 6.28583288192749, + "learning_rate": 7.138618058499364e-07, + "loss": 0.3906, + "mean_token_accuracy": 0.8687189221382141, + "num_tokens": 65533525.0, + "step": 1685 + }, + { + "epoch": 0.21447652970360004, + "ewc_loss": 0.00567626953125, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.125999450683594e-06, + "grad_norm": 6.147218227386475, + "learning_rate": 7.142857142857143e-07, + "loss": 0.3648, + "mean_token_accuracy": 0.8773880004882812, + "num_tokens": 65577683.0, + "step": 1686 + }, + { + "epoch": 0.21460373998219057, + "ewc_loss": 0.00567626953125, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.125999450683594e-06, + "grad_norm": 6.300646781921387, + "learning_rate": 7.147096227214922e-07, + "loss": 0.3921, + "mean_token_accuracy": 0.8686226606369019, + "num_tokens": 65619519.0, + "step": 1687 + }, + { + "epoch": 0.21473095026078107, + "ewc_loss": 0.0057373046875, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.185604095458984e-06, + "grad_norm": 6.296205043792725, + "learning_rate": 7.1513353115727e-07, + "loss": 0.3791, + "mean_token_accuracy": 0.869493842124939, + "num_tokens": 65656303.0, + "step": 1688 + }, + { + "epoch": 0.21485816053937157, + "ewc_loss": 0.00567626953125, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.125999450683594e-06, + "grad_norm": 6.310534954071045, + "learning_rate": 7.155574395930479e-07, + "loss": 0.4167, + "mean_token_accuracy": 0.8580622673034668, + "num_tokens": 65693113.0, + "step": 1689 + }, + { + "epoch": 0.2149853708179621, + "ewc_loss": 0.0057373046875, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.185604095458984e-06, + "grad_norm": 6.325847625732422, + "learning_rate": 7.159813480288257e-07, + "loss": 0.3961, + "mean_token_accuracy": 0.8626611232757568, + "num_tokens": 65731923.0, + "step": 1690 + }, + { + "epoch": 0.2151125810965526, + "ewc_loss": 0.0057373046875, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.185604095458984e-06, + "grad_norm": 6.325329303741455, + "learning_rate": 7.164052564646035e-07, + "loss": 0.3692, + "mean_token_accuracy": 0.8733054399490356, + "num_tokens": 65768242.0, + "step": 1691 + }, + { + "epoch": 0.21523979137514312, + "ewc_loss": 0.005767822265625, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.21540641784668e-06, + "grad_norm": 6.284122467041016, + "learning_rate": 7.168291649003815e-07, + "loss": 0.3958, + "mean_token_accuracy": 0.8717746734619141, + "num_tokens": 65809840.0, + "step": 1692 + }, + { + "epoch": 0.21536700165373363, + "ewc_loss": 0.005706787109375, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.155801773071289e-06, + "grad_norm": 6.323261260986328, + "learning_rate": 7.172530733361593e-07, + "loss": 0.4233, + "mean_token_accuracy": 0.858412504196167, + "num_tokens": 65848746.0, + "step": 1693 + }, + { + "epoch": 0.21549421193232413, + "ewc_loss": 0.0057373046875, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.185604095458984e-06, + "grad_norm": 6.1701507568359375, + "learning_rate": 7.176769817719373e-07, + "loss": 0.3464, + "mean_token_accuracy": 0.8861852288246155, + "num_tokens": 65888881.0, + "step": 1694 + }, + { + "epoch": 0.21562142221091465, + "ewc_loss": 0.005706787109375, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.155801773071289e-06, + "grad_norm": 6.3567118644714355, + "learning_rate": 7.181008902077151e-07, + "loss": 0.4027, + "mean_token_accuracy": 0.8663872480392456, + "num_tokens": 65926205.0, + "step": 1695 + }, + { + "epoch": 0.21574863248950515, + "ewc_loss": 0.00579833984375, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.245208740234375e-06, + "grad_norm": 6.3060173988342285, + "learning_rate": 7.18524798643493e-07, + "loss": 0.406, + "mean_token_accuracy": 0.8646191954612732, + "num_tokens": 65966778.0, + "step": 1696 + }, + { + "epoch": 0.21587584276809565, + "ewc_loss": 0.0057373046875, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.185604095458984e-06, + "grad_norm": 6.305298328399658, + "learning_rate": 7.189487070792708e-07, + "loss": 0.3958, + "mean_token_accuracy": 0.8698852062225342, + "num_tokens": 66004335.0, + "step": 1697 + }, + { + "epoch": 0.21600305304668618, + "ewc_loss": 0.005767822265625, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.21540641784668e-06, + "grad_norm": 6.251436710357666, + "learning_rate": 7.193726155150487e-07, + "loss": 0.3998, + "mean_token_accuracy": 0.8667607307434082, + "num_tokens": 66046894.0, + "step": 1698 + }, + { + "epoch": 0.21613026332527668, + "ewc_loss": 0.005706787109375, + "ewc_loss_diag": 5.5730342864990234e-06, + "ewc_loss_parallel": 5.155801773071289e-06, + "grad_norm": 6.292831897735596, + "learning_rate": 7.197965239508265e-07, + "loss": 0.4013, + "mean_token_accuracy": 0.8695882558822632, + "num_tokens": 66086468.0, + "step": 1699 + }, + { + "epoch": 0.21625747360386718, + "ewc_loss": 0.005767822265625, + "ewc_loss_diag": 5.602836608886719e-06, + "ewc_loss_parallel": 5.21540641784668e-06, + "grad_norm": 6.216721534729004, + "learning_rate": 7.202204323866045e-07, + "loss": 0.3957, + "mean_token_accuracy": 0.8691048622131348, + "num_tokens": 66128622.0, + "step": 1700 + }, + { + "epoch": 0.2163846838824577, + "ewc_loss": 0.005706787109375, + "ewc_loss_diag": 5.602836608886719e-06, + "ewc_loss_parallel": 5.155801773071289e-06, + "grad_norm": 6.246982097625732, + "learning_rate": 7.206443408223823e-07, + "loss": 0.4038, + "mean_token_accuracy": 0.8642517924308777, + "num_tokens": 66169121.0, + "step": 1701 + }, + { + "epoch": 0.2165118941610482, + "ewc_loss": 0.005706787109375, + "ewc_loss_diag": 5.602836608886719e-06, + "ewc_loss_parallel": 5.155801773071289e-06, + "grad_norm": 6.346123695373535, + "learning_rate": 7.210682492581603e-07, + "loss": 0.3884, + "mean_token_accuracy": 0.8677884340286255, + "num_tokens": 66201632.0, + "step": 1702 + }, + { + "epoch": 0.2166391044396387, + "ewc_loss": 0.0057373046875, + "ewc_loss_diag": 5.602836608886719e-06, + "ewc_loss_parallel": 5.185604095458984e-06, + "grad_norm": 6.334092140197754, + "learning_rate": 7.214921576939381e-07, + "loss": 0.4253, + "mean_token_accuracy": 0.8547285795211792, + "num_tokens": 66242675.0, + "step": 1703 + }, + { + "epoch": 0.21676631471822924, + "ewc_loss": 0.005767822265625, + "ewc_loss_diag": 5.602836608886719e-06, + "ewc_loss_parallel": 5.21540641784668e-06, + "grad_norm": 6.295527458190918, + "learning_rate": 7.219160661297159e-07, + "loss": 0.379, + "mean_token_accuracy": 0.8701796531677246, + "num_tokens": 66278916.0, + "step": 1704 + }, + { + "epoch": 0.21689352499681974, + "ewc_loss": 0.0057373046875, + "ewc_loss_diag": 5.662441253662109e-06, + "ewc_loss_parallel": 5.185604095458984e-06, + "grad_norm": 6.331966876983643, + "learning_rate": 7.223399745654938e-07, + "loss": 0.4155, + "mean_token_accuracy": 0.8626337051391602, + "num_tokens": 66318962.0, + "step": 1705 + }, + { + "epoch": 0.21702073527541024, + "ewc_loss": 0.00579833984375, + "ewc_loss_diag": 5.662441253662109e-06, + "ewc_loss_parallel": 5.21540641784668e-06, + "grad_norm": 6.322761058807373, + "learning_rate": 7.227638830012717e-07, + "loss": 0.4159, + "mean_token_accuracy": 0.8594547510147095, + "num_tokens": 66358537.0, + "step": 1706 + }, + { + "epoch": 0.21714794555400077, + "ewc_loss": 0.0057373046875, + "ewc_loss_diag": 5.692243576049805e-06, + "ewc_loss_parallel": 5.155801773071289e-06, + "grad_norm": 6.336374759674072, + "learning_rate": 7.231877914370495e-07, + "loss": 0.3675, + "mean_token_accuracy": 0.8774932622909546, + "num_tokens": 66395674.0, + "step": 1707 + }, + { + "epoch": 0.21727515583259127, + "ewc_loss": 0.005859375, + "ewc_loss_diag": 5.692243576049805e-06, + "ewc_loss_parallel": 5.27501106262207e-06, + "grad_norm": 6.497890949249268, + "learning_rate": 7.236116998728275e-07, + "loss": 0.4421, + "mean_token_accuracy": 0.8517183065414429, + "num_tokens": 66429157.0, + "step": 1708 + }, + { + "epoch": 0.21740236611118177, + "ewc_loss": 0.005828857421875, + "ewc_loss_diag": 5.692243576049805e-06, + "ewc_loss_parallel": 5.245208740234375e-06, + "grad_norm": 6.23553466796875, + "learning_rate": 7.240356083086053e-07, + "loss": 0.3299, + "mean_token_accuracy": 0.886106014251709, + "num_tokens": 66467701.0, + "step": 1709 + }, + { + "epoch": 0.2175295763897723, + "ewc_loss": 0.005767822265625, + "ewc_loss_diag": 5.692243576049805e-06, + "ewc_loss_parallel": 5.185604095458984e-06, + "grad_norm": 6.275259494781494, + "learning_rate": 7.244595167443833e-07, + "loss": 0.4181, + "mean_token_accuracy": 0.8613961935043335, + "num_tokens": 66512485.0, + "step": 1710 + }, + { + "epoch": 0.2176567866683628, + "ewc_loss": 0.005828857421875, + "ewc_loss_diag": 5.692243576049805e-06, + "ewc_loss_parallel": 5.245208740234375e-06, + "grad_norm": 6.305373668670654, + "learning_rate": 7.248834251801611e-07, + "loss": 0.384, + "mean_token_accuracy": 0.8731415271759033, + "num_tokens": 66555628.0, + "step": 1711 + }, + { + "epoch": 0.2177839969469533, + "ewc_loss": 0.005889892578125, + "ewc_loss_diag": 5.692243576049805e-06, + "ewc_loss_parallel": 5.304813385009766e-06, + "grad_norm": 6.27331018447876, + "learning_rate": 7.253073336159388e-07, + "loss": 0.3853, + "mean_token_accuracy": 0.8703649044036865, + "num_tokens": 66593832.0, + "step": 1712 + }, + { + "epoch": 0.21791120722554383, + "ewc_loss": 0.00579833984375, + "ewc_loss_diag": 5.7220458984375e-06, + "ewc_loss_parallel": 5.21540641784668e-06, + "grad_norm": 6.463079929351807, + "learning_rate": 7.257312420517168e-07, + "loss": 0.36, + "mean_token_accuracy": 0.8769866824150085, + "num_tokens": 66625676.0, + "step": 1713 + }, + { + "epoch": 0.21803841750413433, + "ewc_loss": 0.005859375, + "ewc_loss_diag": 5.7220458984375e-06, + "ewc_loss_parallel": 5.27501106262207e-06, + "grad_norm": 6.374102592468262, + "learning_rate": 7.261551504874946e-07, + "loss": 0.384, + "mean_token_accuracy": 0.8674707412719727, + "num_tokens": 66665077.0, + "step": 1714 + }, + { + "epoch": 0.21816562778272486, + "ewc_loss": 0.005859375, + "ewc_loss_diag": 5.7220458984375e-06, + "ewc_loss_parallel": 5.27501106262207e-06, + "grad_norm": 6.239567279815674, + "learning_rate": 7.265790589232725e-07, + "loss": 0.3798, + "mean_token_accuracy": 0.8718031644821167, + "num_tokens": 66702545.0, + "step": 1715 + }, + { + "epoch": 0.21829283806131536, + "ewc_loss": 0.00579833984375, + "ewc_loss_diag": 5.7220458984375e-06, + "ewc_loss_parallel": 5.21540641784668e-06, + "grad_norm": 6.2651472091674805, + "learning_rate": 7.270029673590504e-07, + "loss": 0.4137, + "mean_token_accuracy": 0.8616654872894287, + "num_tokens": 66746320.0, + "step": 1716 + }, + { + "epoch": 0.21842004833990586, + "ewc_loss": 0.00579833984375, + "ewc_loss_diag": 5.7220458984375e-06, + "ewc_loss_parallel": 5.21540641784668e-06, + "grad_norm": 6.412649631500244, + "learning_rate": 7.274268757948283e-07, + "loss": 0.455, + "mean_token_accuracy": 0.8509008288383484, + "num_tokens": 66786150.0, + "step": 1717 + }, + { + "epoch": 0.2185472586184964, + "ewc_loss": 0.005889892578125, + "ewc_loss_diag": 5.7220458984375e-06, + "ewc_loss_parallel": 5.304813385009766e-06, + "grad_norm": 6.388260841369629, + "learning_rate": 7.278507842306062e-07, + "loss": 0.3751, + "mean_token_accuracy": 0.8727119565010071, + "num_tokens": 66821340.0, + "step": 1718 + }, + { + "epoch": 0.2186744688970869, + "ewc_loss": 0.005889892578125, + "ewc_loss_diag": 5.751848220825195e-06, + "ewc_loss_parallel": 5.304813385009766e-06, + "grad_norm": 6.378894805908203, + "learning_rate": 7.282746926663841e-07, + "loss": 0.459, + "mean_token_accuracy": 0.850121796131134, + "num_tokens": 66860188.0, + "step": 1719 + }, + { + "epoch": 0.2188016791756774, + "ewc_loss": 0.005828857421875, + "ewc_loss_diag": 5.751848220825195e-06, + "ewc_loss_parallel": 5.245208740234375e-06, + "grad_norm": 6.355986595153809, + "learning_rate": 7.286986011021618e-07, + "loss": 0.3897, + "mean_token_accuracy": 0.8700530529022217, + "num_tokens": 66897354.0, + "step": 1720 + }, + { + "epoch": 0.21892888945426792, + "ewc_loss": 0.005828857421875, + "ewc_loss_diag": 5.751848220825195e-06, + "ewc_loss_parallel": 5.245208740234375e-06, + "grad_norm": 6.214761257171631, + "learning_rate": 7.291225095379398e-07, + "loss": 0.4068, + "mean_token_accuracy": 0.8674252033233643, + "num_tokens": 66943586.0, + "step": 1721 + }, + { + "epoch": 0.21905609973285842, + "ewc_loss": 0.005828857421875, + "ewc_loss_diag": 5.781650543212891e-06, + "ewc_loss_parallel": 5.245208740234375e-06, + "grad_norm": 6.248122692108154, + "learning_rate": 7.295464179737176e-07, + "loss": 0.3965, + "mean_token_accuracy": 0.8689143657684326, + "num_tokens": 66983271.0, + "step": 1722 + }, + { + "epoch": 0.21918331001144892, + "ewc_loss": 0.00592041015625, + "ewc_loss_diag": 5.781650543212891e-06, + "ewc_loss_parallel": 5.334615707397461e-06, + "grad_norm": 6.307678699493408, + "learning_rate": 7.299703264094955e-07, + "loss": 0.3919, + "mean_token_accuracy": 0.8666824102401733, + "num_tokens": 67020876.0, + "step": 1723 + }, + { + "epoch": 0.21931052029003945, + "ewc_loss": 0.005889892578125, + "ewc_loss_diag": 5.781650543212891e-06, + "ewc_loss_parallel": 5.304813385009766e-06, + "grad_norm": 6.304995059967041, + "learning_rate": 7.303942348452734e-07, + "loss": 0.4198, + "mean_token_accuracy": 0.860138475894928, + "num_tokens": 67062560.0, + "step": 1724 + }, + { + "epoch": 0.21943773056862995, + "ewc_loss": 0.005859375, + "ewc_loss_diag": 5.781650543212891e-06, + "ewc_loss_parallel": 5.27501106262207e-06, + "grad_norm": 6.265782356262207, + "learning_rate": 7.308181432810513e-07, + "loss": 0.3717, + "mean_token_accuracy": 0.8745718002319336, + "num_tokens": 67101713.0, + "step": 1725 + }, + { + "epoch": 0.21956494084722045, + "ewc_loss": 0.00592041015625, + "ewc_loss_diag": 5.781650543212891e-06, + "ewc_loss_parallel": 5.334615707397461e-06, + "grad_norm": 6.28082275390625, + "learning_rate": 7.312420517168292e-07, + "loss": 0.3905, + "mean_token_accuracy": 0.8679640293121338, + "num_tokens": 67141589.0, + "step": 1726 + }, + { + "epoch": 0.21969215112581097, + "ewc_loss": 0.00592041015625, + "ewc_loss_diag": 5.781650543212891e-06, + "ewc_loss_parallel": 5.334615707397461e-06, + "grad_norm": 6.359455108642578, + "learning_rate": 7.31665960152607e-07, + "loss": 0.4154, + "mean_token_accuracy": 0.857338547706604, + "num_tokens": 67183736.0, + "step": 1727 + }, + { + "epoch": 0.21981936140440148, + "ewc_loss": 0.005950927734375, + "ewc_loss_diag": 5.781650543212891e-06, + "ewc_loss_parallel": 5.364418029785156e-06, + "grad_norm": 6.3151679039001465, + "learning_rate": 7.320898685883848e-07, + "loss": 0.3947, + "mean_token_accuracy": 0.8690265417098999, + "num_tokens": 67222758.0, + "step": 1728 + }, + { + "epoch": 0.21994657168299198, + "ewc_loss": 0.005859375, + "ewc_loss_diag": 5.781650543212891e-06, + "ewc_loss_parallel": 5.27501106262207e-06, + "grad_norm": 6.409320831298828, + "learning_rate": 7.325137770241628e-07, + "loss": 0.4893, + "mean_token_accuracy": 0.8410302400588989, + "num_tokens": 67264352.0, + "step": 1729 + }, + { + "epoch": 0.2200737819615825, + "ewc_loss": 0.005950927734375, + "ewc_loss_diag": 5.781650543212891e-06, + "ewc_loss_parallel": 5.364418029785156e-06, + "grad_norm": 6.310605049133301, + "learning_rate": 7.329376854599406e-07, + "loss": 0.4188, + "mean_token_accuracy": 0.8603825569152832, + "num_tokens": 67306926.0, + "step": 1730 + }, + { + "epoch": 0.220200992240173, + "ewc_loss": 0.005889892578125, + "ewc_loss_diag": 5.781650543212891e-06, + "ewc_loss_parallel": 5.304813385009766e-06, + "grad_norm": 6.37901496887207, + "learning_rate": 7.333615938957184e-07, + "loss": 0.4038, + "mean_token_accuracy": 0.8658249378204346, + "num_tokens": 67345306.0, + "step": 1731 + }, + { + "epoch": 0.2203282025187635, + "ewc_loss": 0.005889892578125, + "ewc_loss_diag": 5.781650543212891e-06, + "ewc_loss_parallel": 5.304813385009766e-06, + "grad_norm": 6.363714218139648, + "learning_rate": 7.337855023314964e-07, + "loss": 0.3883, + "mean_token_accuracy": 0.8673848509788513, + "num_tokens": 67384939.0, + "step": 1732 + }, + { + "epoch": 0.22045541279735403, + "ewc_loss": 0.00592041015625, + "ewc_loss_diag": 5.781650543212891e-06, + "ewc_loss_parallel": 5.334615707397461e-06, + "grad_norm": 6.319650650024414, + "learning_rate": 7.342094107672742e-07, + "loss": 0.435, + "mean_token_accuracy": 0.8525841236114502, + "num_tokens": 67428223.0, + "step": 1733 + }, + { + "epoch": 0.22058262307594453, + "ewc_loss": 0.005950927734375, + "ewc_loss_diag": 5.811452865600586e-06, + "ewc_loss_parallel": 5.364418029785156e-06, + "grad_norm": 6.479025363922119, + "learning_rate": 7.346333192030522e-07, + "loss": 0.4003, + "mean_token_accuracy": 0.8681814074516296, + "num_tokens": 67465535.0, + "step": 1734 + }, + { + "epoch": 0.22070983335453503, + "ewc_loss": 0.0059814453125, + "ewc_loss_diag": 5.811452865600586e-06, + "ewc_loss_parallel": 5.3942203521728516e-06, + "grad_norm": 6.390556812286377, + "learning_rate": 7.350572276388299e-07, + "loss": 0.4403, + "mean_token_accuracy": 0.8512105941772461, + "num_tokens": 67508494.0, + "step": 1735 + }, + { + "epoch": 0.22083704363312556, + "ewc_loss": 0.00592041015625, + "ewc_loss_diag": 5.811452865600586e-06, + "ewc_loss_parallel": 5.334615707397461e-06, + "grad_norm": 6.4289984703063965, + "learning_rate": 7.354811360746078e-07, + "loss": 0.4117, + "mean_token_accuracy": 0.8618567585945129, + "num_tokens": 67546065.0, + "step": 1736 + }, + { + "epoch": 0.22096425391171606, + "ewc_loss": 0.005889892578125, + "ewc_loss_diag": 5.811452865600586e-06, + "ewc_loss_parallel": 5.304813385009766e-06, + "grad_norm": 6.435615539550781, + "learning_rate": 7.359050445103857e-07, + "loss": 0.4284, + "mean_token_accuracy": 0.8578352928161621, + "num_tokens": 67583207.0, + "step": 1737 + }, + { + "epoch": 0.22109146419030656, + "ewc_loss": 0.00592041015625, + "ewc_loss_diag": 5.811452865600586e-06, + "ewc_loss_parallel": 5.334615707397461e-06, + "grad_norm": 6.335570812225342, + "learning_rate": 7.363289529461636e-07, + "loss": 0.3979, + "mean_token_accuracy": 0.8679803013801575, + "num_tokens": 67622844.0, + "step": 1738 + }, + { + "epoch": 0.2212186744688971, + "ewc_loss": 0.00592041015625, + "ewc_loss_diag": 5.811452865600586e-06, + "ewc_loss_parallel": 5.334615707397461e-06, + "grad_norm": 6.28322696685791, + "learning_rate": 7.367528613819415e-07, + "loss": 0.3458, + "mean_token_accuracy": 0.8816287517547607, + "num_tokens": 67665111.0, + "step": 1739 + }, + { + "epoch": 0.2213458847474876, + "ewc_loss": 0.005859375, + "ewc_loss_diag": 5.811452865600586e-06, + "ewc_loss_parallel": 5.27501106262207e-06, + "grad_norm": 6.415442943572998, + "learning_rate": 7.371767698177194e-07, + "loss": 0.3757, + "mean_token_accuracy": 0.8703023791313171, + "num_tokens": 67700338.0, + "step": 1740 + }, + { + "epoch": 0.22147309502607812, + "ewc_loss": 0.005889892578125, + "ewc_loss_diag": 5.811452865600586e-06, + "ewc_loss_parallel": 5.304813385009766e-06, + "grad_norm": 6.403069019317627, + "learning_rate": 7.376006782534972e-07, + "loss": 0.372, + "mean_token_accuracy": 0.8751944899559021, + "num_tokens": 67734291.0, + "step": 1741 + }, + { + "epoch": 0.22160030530466862, + "ewc_loss": 0.00592041015625, + "ewc_loss_diag": 5.811452865600586e-06, + "ewc_loss_parallel": 5.334615707397461e-06, + "grad_norm": 6.378697872161865, + "learning_rate": 7.380245866892751e-07, + "loss": 0.3796, + "mean_token_accuracy": 0.8706440329551697, + "num_tokens": 67770113.0, + "step": 1742 + }, + { + "epoch": 0.22172751558325912, + "ewc_loss": 0.005889892578125, + "ewc_loss_diag": 5.811452865600586e-06, + "ewc_loss_parallel": 5.304813385009766e-06, + "grad_norm": 6.274598598480225, + "learning_rate": 7.384484951250529e-07, + "loss": 0.3885, + "mean_token_accuracy": 0.8695241808891296, + "num_tokens": 67814109.0, + "step": 1743 + }, + { + "epoch": 0.22185472586184965, + "ewc_loss": 0.005889892578125, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.304813385009766e-06, + "grad_norm": 6.364073753356934, + "learning_rate": 7.388724035608308e-07, + "loss": 0.3823, + "mean_token_accuracy": 0.8729591965675354, + "num_tokens": 67849736.0, + "step": 1744 + }, + { + "epoch": 0.22198193614044015, + "ewc_loss": 0.005950927734375, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.364418029785156e-06, + "grad_norm": 6.409918308258057, + "learning_rate": 7.392963119966087e-07, + "loss": 0.3714, + "mean_token_accuracy": 0.8733499646186829, + "num_tokens": 67881929.0, + "step": 1745 + }, + { + "epoch": 0.22210914641903065, + "ewc_loss": 0.00592041015625, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.334615707397461e-06, + "grad_norm": 6.488662242889404, + "learning_rate": 7.397202204323866e-07, + "loss": 0.4156, + "mean_token_accuracy": 0.8611682653427124, + "num_tokens": 67919766.0, + "step": 1746 + }, + { + "epoch": 0.22223635669762118, + "ewc_loss": 0.0059814453125, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.3942203521728516e-06, + "grad_norm": 6.315069675445557, + "learning_rate": 7.401441288681645e-07, + "loss": 0.3633, + "mean_token_accuracy": 0.874679684638977, + "num_tokens": 67960640.0, + "step": 1747 + }, + { + "epoch": 0.22236356697621168, + "ewc_loss": 0.005889892578125, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.304813385009766e-06, + "grad_norm": 6.291140079498291, + "learning_rate": 7.405680373039424e-07, + "loss": 0.3663, + "mean_token_accuracy": 0.8773535490036011, + "num_tokens": 68000463.0, + "step": 1748 + }, + { + "epoch": 0.22249077725480218, + "ewc_loss": 0.005950927734375, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.364418029785156e-06, + "grad_norm": 6.477694988250732, + "learning_rate": 7.409919457397202e-07, + "loss": 0.4155, + "mean_token_accuracy": 0.8615229725837708, + "num_tokens": 68033720.0, + "step": 1749 + }, + { + "epoch": 0.2226179875333927, + "ewc_loss": 0.006011962890625, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.424022674560547e-06, + "grad_norm": 6.455885410308838, + "learning_rate": 7.414158541754981e-07, + "loss": 0.4459, + "mean_token_accuracy": 0.8498668074607849, + "num_tokens": 68073725.0, + "step": 1750 + }, + { + "epoch": 0.2227451978119832, + "ewc_loss": 0.0059814453125, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.3942203521728516e-06, + "grad_norm": 6.603471279144287, + "learning_rate": 7.418397626112759e-07, + "loss": 0.4028, + "mean_token_accuracy": 0.8717148303985596, + "num_tokens": 68104073.0, + "step": 1751 + }, + { + "epoch": 0.2228724080905737, + "ewc_loss": 0.00604248046875, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.453824996948242e-06, + "grad_norm": 6.481344699859619, + "learning_rate": 7.422636710470537e-07, + "loss": 0.4021, + "mean_token_accuracy": 0.8654545545578003, + "num_tokens": 68146982.0, + "step": 1752 + }, + { + "epoch": 0.22299961836916424, + "ewc_loss": 0.0059814453125, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.3942203521728516e-06, + "grad_norm": 6.423767566680908, + "learning_rate": 7.426875794828317e-07, + "loss": 0.423, + "mean_token_accuracy": 0.8618590831756592, + "num_tokens": 68184390.0, + "step": 1753 + }, + { + "epoch": 0.22312682864775474, + "ewc_loss": 0.006011962890625, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.424022674560547e-06, + "grad_norm": 6.402594089508057, + "learning_rate": 7.431114879186095e-07, + "loss": 0.4404, + "mean_token_accuracy": 0.8543139696121216, + "num_tokens": 68224479.0, + "step": 1754 + }, + { + "epoch": 0.22325403892634524, + "ewc_loss": 0.006011962890625, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.424022674560547e-06, + "grad_norm": 6.479334354400635, + "learning_rate": 7.435353963543875e-07, + "loss": 0.3946, + "mean_token_accuracy": 0.868390679359436, + "num_tokens": 68267254.0, + "step": 1755 + }, + { + "epoch": 0.22338124920493577, + "ewc_loss": 0.006072998046875, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.4836273193359375e-06, + "grad_norm": 6.327544689178467, + "learning_rate": 7.439593047901653e-07, + "loss": 0.3631, + "mean_token_accuracy": 0.8748527765274048, + "num_tokens": 68306644.0, + "step": 1756 + }, + { + "epoch": 0.22350845948352627, + "ewc_loss": 0.0059814453125, + "ewc_loss_diag": 5.841255187988281e-06, + "ewc_loss_parallel": 5.3942203521728516e-06, + "grad_norm": 6.460899829864502, + "learning_rate": 7.443832132259431e-07, + "loss": 0.4138, + "mean_token_accuracy": 0.8591843247413635, + "num_tokens": 68343880.0, + "step": 1757 + }, + { + "epoch": 0.22363566976211677, + "ewc_loss": 0.00604248046875, + "ewc_loss_diag": 5.8710575103759766e-06, + "ewc_loss_parallel": 5.453824996948242e-06, + "grad_norm": 6.420841693878174, + "learning_rate": 7.44807121661721e-07, + "loss": 0.3964, + "mean_token_accuracy": 0.8647767305374146, + "num_tokens": 68382578.0, + "step": 1758 + }, + { + "epoch": 0.2237628800407073, + "ewc_loss": 0.006011962890625, + "ewc_loss_diag": 5.8710575103759766e-06, + "ewc_loss_parallel": 5.424022674560547e-06, + "grad_norm": 6.565048694610596, + "learning_rate": 7.452310300974989e-07, + "loss": 0.3688, + "mean_token_accuracy": 0.8709299564361572, + "num_tokens": 68417459.0, + "step": 1759 + }, + { + "epoch": 0.2238900903192978, + "ewc_loss": 0.0059814453125, + "ewc_loss_diag": 5.8710575103759766e-06, + "ewc_loss_parallel": 5.3942203521728516e-06, + "grad_norm": 6.364506244659424, + "learning_rate": 7.456549385332767e-07, + "loss": 0.4303, + "mean_token_accuracy": 0.8563557267189026, + "num_tokens": 68458533.0, + "step": 1760 + }, + { + "epoch": 0.2240173005978883, + "ewc_loss": 0.006011962890625, + "ewc_loss_diag": 5.8710575103759766e-06, + "ewc_loss_parallel": 5.424022674560547e-06, + "grad_norm": 6.301523685455322, + "learning_rate": 7.460788469690547e-07, + "loss": 0.3409, + "mean_token_accuracy": 0.8841415643692017, + "num_tokens": 68497448.0, + "step": 1761 + }, + { + "epoch": 0.22414451087647883, + "ewc_loss": 0.00604248046875, + "ewc_loss_diag": 5.8710575103759766e-06, + "ewc_loss_parallel": 5.453824996948242e-06, + "grad_norm": 6.410353660583496, + "learning_rate": 7.465027554048325e-07, + "loss": 0.4627, + "mean_token_accuracy": 0.8490885496139526, + "num_tokens": 68541889.0, + "step": 1762 + }, + { + "epoch": 0.22427172115506933, + "ewc_loss": 0.00604248046875, + "ewc_loss_diag": 5.8710575103759766e-06, + "ewc_loss_parallel": 5.453824996948242e-06, + "grad_norm": 6.43147087097168, + "learning_rate": 7.469266638406105e-07, + "loss": 0.462, + "mean_token_accuracy": 0.8474806547164917, + "num_tokens": 68586924.0, + "step": 1763 + }, + { + "epoch": 0.22439893143365983, + "ewc_loss": 0.00604248046875, + "ewc_loss_diag": 5.8710575103759766e-06, + "ewc_loss_parallel": 5.453824996948242e-06, + "grad_norm": 6.429632663726807, + "learning_rate": 7.473505722763883e-07, + "loss": 0.3859, + "mean_token_accuracy": 0.8715930581092834, + "num_tokens": 68623225.0, + "step": 1764 + }, + { + "epoch": 0.22452614171225035, + "ewc_loss": 0.00604248046875, + "ewc_loss_diag": 5.900859832763672e-06, + "ewc_loss_parallel": 5.453824996948242e-06, + "grad_norm": 6.488564968109131, + "learning_rate": 7.477744807121661e-07, + "loss": 0.4242, + "mean_token_accuracy": 0.8572795987129211, + "num_tokens": 68661684.0, + "step": 1765 + }, + { + "epoch": 0.22465335199084085, + "ewc_loss": 0.006103515625, + "ewc_loss_diag": 5.900859832763672e-06, + "ewc_loss_parallel": 5.513429641723633e-06, + "grad_norm": 6.523934841156006, + "learning_rate": 7.48198389147944e-07, + "loss": 0.4096, + "mean_token_accuracy": 0.8665159344673157, + "num_tokens": 68697847.0, + "step": 1766 + }, + { + "epoch": 0.22478056226943138, + "ewc_loss": 0.00604248046875, + "ewc_loss_diag": 5.8710575103759766e-06, + "ewc_loss_parallel": 5.453824996948242e-06, + "grad_norm": 6.54506778717041, + "learning_rate": 7.486222975837219e-07, + "loss": 0.3962, + "mean_token_accuracy": 0.8639760613441467, + "num_tokens": 68739653.0, + "step": 1767 + }, + { + "epoch": 0.22490777254802188, + "ewc_loss": 0.00604248046875, + "ewc_loss_diag": 5.900859832763672e-06, + "ewc_loss_parallel": 5.453824996948242e-06, + "grad_norm": 6.386998653411865, + "learning_rate": 7.490462060194997e-07, + "loss": 0.3573, + "mean_token_accuracy": 0.87629234790802, + "num_tokens": 68774770.0, + "step": 1768 + }, + { + "epoch": 0.22503498282661238, + "ewc_loss": 0.0059814453125, + "ewc_loss_diag": 5.900859832763672e-06, + "ewc_loss_parallel": 5.3942203521728516e-06, + "grad_norm": 6.569522857666016, + "learning_rate": 7.494701144552777e-07, + "loss": 0.4336, + "mean_token_accuracy": 0.854491651058197, + "num_tokens": 68809298.0, + "step": 1769 + }, + { + "epoch": 0.2251621931052029, + "ewc_loss": 0.006134033203125, + "ewc_loss_diag": 5.900859832763672e-06, + "ewc_loss_parallel": 5.543231964111328e-06, + "grad_norm": 6.4090895652771, + "learning_rate": 7.498940228910555e-07, + "loss": 0.413, + "mean_token_accuracy": 0.8600752353668213, + "num_tokens": 68850832.0, + "step": 1770 + }, + { + "epoch": 0.2252894033837934, + "ewc_loss": 0.00604248046875, + "ewc_loss_diag": 5.900859832763672e-06, + "ewc_loss_parallel": 5.453824996948242e-06, + "grad_norm": 6.516145706176758, + "learning_rate": 7.503179313268335e-07, + "loss": 0.4032, + "mean_token_accuracy": 0.8635010719299316, + "num_tokens": 68885796.0, + "step": 1771 + }, + { + "epoch": 0.2254166136623839, + "ewc_loss": 0.006103515625, + "ewc_loss_diag": 5.900859832763672e-06, + "ewc_loss_parallel": 5.513429641723633e-06, + "grad_norm": 6.44976282119751, + "learning_rate": 7.507418397626113e-07, + "loss": 0.4279, + "mean_token_accuracy": 0.8570414781570435, + "num_tokens": 68927132.0, + "step": 1772 + }, + { + "epoch": 0.22554382394097444, + "ewc_loss": 0.006072998046875, + "ewc_loss_diag": 5.900859832763672e-06, + "ewc_loss_parallel": 5.4836273193359375e-06, + "grad_norm": 6.413853645324707, + "learning_rate": 7.51165748198389e-07, + "loss": 0.4169, + "mean_token_accuracy": 0.8563293218612671, + "num_tokens": 68968501.0, + "step": 1773 + }, + { + "epoch": 0.22567103421956494, + "ewc_loss": 0.00604248046875, + "ewc_loss_diag": 5.900859832763672e-06, + "ewc_loss_parallel": 5.453824996948242e-06, + "grad_norm": 6.4080376625061035, + "learning_rate": 7.51589656634167e-07, + "loss": 0.3885, + "mean_token_accuracy": 0.8663520812988281, + "num_tokens": 69004962.0, + "step": 1774 + }, + { + "epoch": 0.22579824449815544, + "ewc_loss": 0.006103515625, + "ewc_loss_diag": 5.900859832763672e-06, + "ewc_loss_parallel": 5.513429641723633e-06, + "grad_norm": 6.433106899261475, + "learning_rate": 7.520135650699448e-07, + "loss": 0.4018, + "mean_token_accuracy": 0.8646952509880066, + "num_tokens": 69043834.0, + "step": 1775 + }, + { + "epoch": 0.22592545477674597, + "ewc_loss": 0.006134033203125, + "ewc_loss_diag": 5.900859832763672e-06, + "ewc_loss_parallel": 5.543231964111328e-06, + "grad_norm": 6.792252063751221, + "learning_rate": 7.524374735057227e-07, + "loss": 0.4123, + "mean_token_accuracy": 0.8639246821403503, + "num_tokens": 69080506.0, + "step": 1776 + }, + { + "epoch": 0.22605266505533647, + "ewc_loss": 0.00616455078125, + "ewc_loss_diag": 5.900859832763672e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.519698143005371, + "learning_rate": 7.528613819415006e-07, + "loss": 0.3972, + "mean_token_accuracy": 0.8689770698547363, + "num_tokens": 69115580.0, + "step": 1777 + }, + { + "epoch": 0.22617987533392697, + "ewc_loss": 0.006103515625, + "ewc_loss_diag": 5.9604644775390625e-06, + "ewc_loss_parallel": 5.4836273193359375e-06, + "grad_norm": 6.533524990081787, + "learning_rate": 7.532852903772785e-07, + "loss": 0.4128, + "mean_token_accuracy": 0.8605685234069824, + "num_tokens": 69152280.0, + "step": 1778 + }, + { + "epoch": 0.2263070856125175, + "ewc_loss": 0.006103515625, + "ewc_loss_diag": 5.9604644775390625e-06, + "ewc_loss_parallel": 5.513429641723633e-06, + "grad_norm": 6.5642242431640625, + "learning_rate": 7.537091988130564e-07, + "loss": 0.4307, + "mean_token_accuracy": 0.8579306602478027, + "num_tokens": 69187577.0, + "step": 1779 + }, + { + "epoch": 0.226434295891108, + "ewc_loss": 0.006103515625, + "ewc_loss_diag": 5.9604644775390625e-06, + "ewc_loss_parallel": 5.513429641723633e-06, + "grad_norm": 6.49827766418457, + "learning_rate": 7.541331072488342e-07, + "loss": 0.4116, + "mean_token_accuracy": 0.862696647644043, + "num_tokens": 69224105.0, + "step": 1780 + }, + { + "epoch": 0.2265615061696985, + "ewc_loss": 0.00616455078125, + "ewc_loss_diag": 5.9604644775390625e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.411589622497559, + "learning_rate": 7.54557015684612e-07, + "loss": 0.3967, + "mean_token_accuracy": 0.8648422956466675, + "num_tokens": 69264734.0, + "step": 1781 + }, + { + "epoch": 0.22668871644828903, + "ewc_loss": 0.006103515625, + "ewc_loss_diag": 5.9604644775390625e-06, + "ewc_loss_parallel": 5.513429641723633e-06, + "grad_norm": 6.395570755004883, + "learning_rate": 7.5498092412039e-07, + "loss": 0.3871, + "mean_token_accuracy": 0.8698720932006836, + "num_tokens": 69305522.0, + "step": 1782 + }, + { + "epoch": 0.22681592672687953, + "ewc_loss": 0.00616455078125, + "ewc_loss_diag": 5.9604644775390625e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.3933424949646, + "learning_rate": 7.554048325561678e-07, + "loss": 0.3553, + "mean_token_accuracy": 0.8776422142982483, + "num_tokens": 69345149.0, + "step": 1783 + }, + { + "epoch": 0.22694313700547003, + "ewc_loss": 0.00616455078125, + "ewc_loss_diag": 5.9604644775390625e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.501384258270264, + "learning_rate": 7.558287409919457e-07, + "loss": 0.4339, + "mean_token_accuracy": 0.8574832677841187, + "num_tokens": 69383969.0, + "step": 1784 + }, + { + "epoch": 0.22707034728406056, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 5.9604644775390625e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.604353904724121, + "learning_rate": 7.562526494277236e-07, + "loss": 0.4279, + "mean_token_accuracy": 0.8554365038871765, + "num_tokens": 69420901.0, + "step": 1785 + }, + { + "epoch": 0.22719755756265106, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 5.9604644775390625e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.401125907897949, + "learning_rate": 7.566765578635015e-07, + "loss": 0.391, + "mean_token_accuracy": 0.8714988231658936, + "num_tokens": 69462406.0, + "step": 1786 + }, + { + "epoch": 0.22732476784124156, + "ewc_loss": 0.00616455078125, + "ewc_loss_diag": 5.9604644775390625e-06, + "ewc_loss_parallel": 5.543231964111328e-06, + "grad_norm": 6.448819637298584, + "learning_rate": 7.571004662992794e-07, + "loss": 0.4245, + "mean_token_accuracy": 0.8620443344116211, + "num_tokens": 69501079.0, + "step": 1787 + }, + { + "epoch": 0.2274519781198321, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.490705490112305, + "learning_rate": 7.575243747350572e-07, + "loss": 0.4513, + "mean_token_accuracy": 0.8530713319778442, + "num_tokens": 69542731.0, + "step": 1788 + }, + { + "epoch": 0.2275791883984226, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.417104244232178, + "learning_rate": 7.57948283170835e-07, + "loss": 0.4115, + "mean_token_accuracy": 0.864781379699707, + "num_tokens": 69583741.0, + "step": 1789 + }, + { + "epoch": 0.2277063986770131, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.596858978271484, + "learning_rate": 7.58372191606613e-07, + "loss": 0.4031, + "mean_token_accuracy": 0.8677650094032288, + "num_tokens": 69618603.0, + "step": 1790 + }, + { + "epoch": 0.22783360895560362, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.486363887786865, + "learning_rate": 7.587961000423908e-07, + "loss": 0.4014, + "mean_token_accuracy": 0.8640545606613159, + "num_tokens": 69657248.0, + "step": 1791 + }, + { + "epoch": 0.22796081923419412, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.562165260314941, + "learning_rate": 7.592200084781686e-07, + "loss": 0.4468, + "mean_token_accuracy": 0.8541616201400757, + "num_tokens": 69692562.0, + "step": 1792 + }, + { + "epoch": 0.22808802951278465, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.4959187507629395, + "learning_rate": 7.596439169139466e-07, + "loss": 0.4003, + "mean_token_accuracy": 0.8662421703338623, + "num_tokens": 69728430.0, + "step": 1793 + }, + { + "epoch": 0.22821523979137515, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.4455366134643555, + "learning_rate": 7.600678253497244e-07, + "loss": 0.408, + "mean_token_accuracy": 0.8656243085861206, + "num_tokens": 69769212.0, + "step": 1794 + }, + { + "epoch": 0.22834245006996565, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.430332660675049, + "learning_rate": 7.604917337855023e-07, + "loss": 0.3793, + "mean_token_accuracy": 0.8719680905342102, + "num_tokens": 69806856.0, + "step": 1795 + }, + { + "epoch": 0.22846966034855618, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.511874675750732, + "learning_rate": 7.609156422212801e-07, + "loss": 0.3635, + "mean_token_accuracy": 0.8750337958335876, + "num_tokens": 69844883.0, + "step": 1796 + }, + { + "epoch": 0.22859687062714668, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.484302520751953, + "learning_rate": 7.61339550657058e-07, + "loss": 0.3799, + "mean_token_accuracy": 0.8710217475891113, + "num_tokens": 69884360.0, + "step": 1797 + }, + { + "epoch": 0.22872408090573718, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.644623279571533, + "learning_rate": 7.617634590928359e-07, + "loss": 0.3845, + "mean_token_accuracy": 0.872637152671814, + "num_tokens": 69918602.0, + "step": 1798 + }, + { + "epoch": 0.2288512911843277, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.485826015472412, + "learning_rate": 7.621873675286138e-07, + "loss": 0.418, + "mean_token_accuracy": 0.8598278760910034, + "num_tokens": 69961049.0, + "step": 1799 + }, + { + "epoch": 0.2289785014629182, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.537254810333252, + "learning_rate": 7.626112759643916e-07, + "loss": 0.417, + "mean_token_accuracy": 0.8605690598487854, + "num_tokens": 69999203.0, + "step": 1800 + }, + { + "epoch": 0.2291057117415087, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 5.990266799926758e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.556743144989014, + "learning_rate": 7.630351844001696e-07, + "loss": 0.3924, + "mean_token_accuracy": 0.8687412738800049, + "num_tokens": 70034561.0, + "step": 1801 + }, + { + "epoch": 0.22923292202009923, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 6.0498714447021484e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.459514141082764, + "learning_rate": 7.634590928359474e-07, + "loss": 0.3798, + "mean_token_accuracy": 0.8734538555145264, + "num_tokens": 70073376.0, + "step": 1802 + }, + { + "epoch": 0.22936013229868973, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.0498714447021484e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.433907985687256, + "learning_rate": 7.638830012717253e-07, + "loss": 0.4206, + "mean_token_accuracy": 0.8566967248916626, + "num_tokens": 70114133.0, + "step": 1803 + }, + { + "epoch": 0.22948734257728023, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.0498714447021484e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.447808265686035, + "learning_rate": 7.643069097075031e-07, + "loss": 0.3866, + "mean_token_accuracy": 0.869391143321991, + "num_tokens": 70155020.0, + "step": 1804 + }, + { + "epoch": 0.22961455285587076, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.0498714447021484e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.4958882331848145, + "learning_rate": 7.64730818143281e-07, + "loss": 0.3662, + "mean_token_accuracy": 0.8739117980003357, + "num_tokens": 70196631.0, + "step": 1805 + }, + { + "epoch": 0.22974176313446126, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 6.0498714447021484e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.491243362426758, + "learning_rate": 7.651547265790589e-07, + "loss": 0.428, + "mean_token_accuracy": 0.85761958360672, + "num_tokens": 70237520.0, + "step": 1806 + }, + { + "epoch": 0.22986897341305176, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 6.0498714447021484e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.633078575134277, + "learning_rate": 7.655786350148368e-07, + "loss": 0.4311, + "mean_token_accuracy": 0.8566383123397827, + "num_tokens": 70277987.0, + "step": 1807 + }, + { + "epoch": 0.2299961836916423, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 6.0498714447021484e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.393885612487793, + "learning_rate": 7.660025434506146e-07, + "loss": 0.3976, + "mean_token_accuracy": 0.8670169115066528, + "num_tokens": 70324856.0, + "step": 1808 + }, + { + "epoch": 0.2301233939702328, + "ewc_loss": 0.006134033203125, + "ewc_loss_diag": 6.0498714447021484e-06, + "ewc_loss_parallel": 5.513429641723633e-06, + "grad_norm": 6.477935314178467, + "learning_rate": 7.664264518863926e-07, + "loss": 0.3895, + "mean_token_accuracy": 0.8696556091308594, + "num_tokens": 70364710.0, + "step": 1809 + }, + { + "epoch": 0.2302506042488233, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.0498714447021484e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.567889213562012, + "learning_rate": 7.668503603221704e-07, + "loss": 0.446, + "mean_token_accuracy": 0.8502670526504517, + "num_tokens": 70406462.0, + "step": 1810 + }, + { + "epoch": 0.23037781452741382, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.0498714447021484e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.443786144256592, + "learning_rate": 7.672742687579483e-07, + "loss": 0.3585, + "mean_token_accuracy": 0.8769895434379578, + "num_tokens": 70445323.0, + "step": 1811 + }, + { + "epoch": 0.23050502480600432, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.0498714447021484e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.628953456878662, + "learning_rate": 7.676981771937261e-07, + "loss": 0.3535, + "mean_token_accuracy": 0.87633216381073, + "num_tokens": 70476127.0, + "step": 1812 + }, + { + "epoch": 0.23063223508459482, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 6.079673767089844e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.490397930145264, + "learning_rate": 7.681220856295039e-07, + "loss": 0.3798, + "mean_token_accuracy": 0.8721317648887634, + "num_tokens": 70517100.0, + "step": 1813 + }, + { + "epoch": 0.23075944536318535, + "ewc_loss": 0.00616455078125, + "ewc_loss_diag": 6.079673767089844e-06, + "ewc_loss_parallel": 5.543231964111328e-06, + "grad_norm": 6.624815464019775, + "learning_rate": 7.685459940652819e-07, + "loss": 0.4049, + "mean_token_accuracy": 0.8623970746994019, + "num_tokens": 70552477.0, + "step": 1814 + }, + { + "epoch": 0.23088665564177585, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.079673767089844e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.500863075256348, + "learning_rate": 7.689699025010597e-07, + "loss": 0.3968, + "mean_token_accuracy": 0.8647763133049011, + "num_tokens": 70595372.0, + "step": 1815 + }, + { + "epoch": 0.23101386592036638, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.079673767089844e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.63328742980957, + "learning_rate": 7.693938109368376e-07, + "loss": 0.4141, + "mean_token_accuracy": 0.8607170581817627, + "num_tokens": 70634117.0, + "step": 1816 + }, + { + "epoch": 0.23114107619895688, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 6.079673767089844e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.586994647979736, + "learning_rate": 7.698177193726155e-07, + "loss": 0.3961, + "mean_token_accuracy": 0.8643826842308044, + "num_tokens": 70672637.0, + "step": 1817 + }, + { + "epoch": 0.23126828647754738, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.079673767089844e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.543002605438232, + "learning_rate": 7.702416278083933e-07, + "loss": 0.415, + "mean_token_accuracy": 0.8612866997718811, + "num_tokens": 70707586.0, + "step": 1818 + }, + { + "epoch": 0.2313954967561379, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.079673767089844e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.449738502502441, + "learning_rate": 7.706655362441712e-07, + "loss": 0.399, + "mean_token_accuracy": 0.8649217486381531, + "num_tokens": 70750991.0, + "step": 1819 + }, + { + "epoch": 0.2315227070347284, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 6.079673767089844e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.581831455230713, + "learning_rate": 7.710894446799491e-07, + "loss": 0.3735, + "mean_token_accuracy": 0.8724441528320312, + "num_tokens": 70784625.0, + "step": 1820 + }, + { + "epoch": 0.2316499173133189, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 6.079673767089844e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.55730676651001, + "learning_rate": 7.715133531157269e-07, + "loss": 0.3763, + "mean_token_accuracy": 0.8768019080162048, + "num_tokens": 70826148.0, + "step": 1821 + }, + { + "epoch": 0.23177712759190944, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.139278411865234e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.482253074645996, + "learning_rate": 7.719372615515049e-07, + "loss": 0.4325, + "mean_token_accuracy": 0.8555188179016113, + "num_tokens": 70870810.0, + "step": 1822 + }, + { + "epoch": 0.23190433787049994, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.139278411865234e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.5034708976745605, + "learning_rate": 7.723611699872827e-07, + "loss": 0.3883, + "mean_token_accuracy": 0.8683183193206787, + "num_tokens": 70915101.0, + "step": 1823 + }, + { + "epoch": 0.23203154814909044, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 6.139278411865234e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.648190021514893, + "learning_rate": 7.727850784230606e-07, + "loss": 0.4163, + "mean_token_accuracy": 0.85401850938797, + "num_tokens": 70954561.0, + "step": 1824 + }, + { + "epoch": 0.23215875842768097, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 6.139278411865234e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.547781944274902, + "learning_rate": 7.732089868588385e-07, + "loss": 0.4102, + "mean_token_accuracy": 0.8645113110542297, + "num_tokens": 70993563.0, + "step": 1825 + }, + { + "epoch": 0.23228596870627147, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 6.139278411865234e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.405584812164307, + "learning_rate": 7.736328952946163e-07, + "loss": 0.3533, + "mean_token_accuracy": 0.8789623975753784, + "num_tokens": 71037912.0, + "step": 1826 + }, + { + "epoch": 0.23241317898486197, + "ewc_loss": 0.00616455078125, + "ewc_loss_diag": 6.139278411865234e-06, + "ewc_loss_parallel": 5.543231964111328e-06, + "grad_norm": 6.637455463409424, + "learning_rate": 7.740568037303942e-07, + "loss": 0.3697, + "mean_token_accuracy": 0.8707593083381653, + "num_tokens": 71071921.0, + "step": 1827 + }, + { + "epoch": 0.2325403892634525, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 6.139278411865234e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.500644683837891, + "learning_rate": 7.744807121661721e-07, + "loss": 0.3488, + "mean_token_accuracy": 0.8827091455459595, + "num_tokens": 71110086.0, + "step": 1828 + }, + { + "epoch": 0.232667599542043, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.16908073425293e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.496595859527588, + "learning_rate": 7.749046206019499e-07, + "loss": 0.4018, + "mean_token_accuracy": 0.8656251430511475, + "num_tokens": 71151493.0, + "step": 1829 + }, + { + "epoch": 0.2327948098206335, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.16908073425293e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.520114898681641, + "learning_rate": 7.753285290377279e-07, + "loss": 0.3438, + "mean_token_accuracy": 0.8799409866333008, + "num_tokens": 71188549.0, + "step": 1830 + }, + { + "epoch": 0.23292202009922403, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.16908073425293e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.622659206390381, + "learning_rate": 7.757524374735057e-07, + "loss": 0.4431, + "mean_token_accuracy": 0.850750207901001, + "num_tokens": 71227288.0, + "step": 1831 + }, + { + "epoch": 0.23304923037781453, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.16908073425293e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.482158660888672, + "learning_rate": 7.761763459092836e-07, + "loss": 0.3574, + "mean_token_accuracy": 0.8815381526947021, + "num_tokens": 71265281.0, + "step": 1832 + }, + { + "epoch": 0.23317644065640503, + "ewc_loss": 0.00616455078125, + "ewc_loss_diag": 6.16908073425293e-06, + "ewc_loss_parallel": 5.543231964111328e-06, + "grad_norm": 6.631593227386475, + "learning_rate": 7.766002543450614e-07, + "loss": 0.3825, + "mean_token_accuracy": 0.8706133365631104, + "num_tokens": 71304243.0, + "step": 1833 + }, + { + "epoch": 0.23330365093499555, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 6.16908073425293e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.524791240692139, + "learning_rate": 7.770241627808392e-07, + "loss": 0.3519, + "mean_token_accuracy": 0.8809309601783752, + "num_tokens": 71342391.0, + "step": 1834 + }, + { + "epoch": 0.23343086121358606, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.16908073425293e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.518024444580078, + "learning_rate": 7.774480712166172e-07, + "loss": 0.4139, + "mean_token_accuracy": 0.8608264923095703, + "num_tokens": 71384325.0, + "step": 1835 + }, + { + "epoch": 0.23355807149217656, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 6.16908073425293e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.5395402908325195, + "learning_rate": 7.77871979652395e-07, + "loss": 0.4125, + "mean_token_accuracy": 0.8607637882232666, + "num_tokens": 71424423.0, + "step": 1836 + }, + { + "epoch": 0.23368528177076708, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 6.16908073425293e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.625799655914307, + "learning_rate": 7.782958880881729e-07, + "loss": 0.396, + "mean_token_accuracy": 0.8664445877075195, + "num_tokens": 71459172.0, + "step": 1837 + }, + { + "epoch": 0.23381249204935758, + "ewc_loss": 0.0062255859375, + "ewc_loss_diag": 6.16908073425293e-06, + "ewc_loss_parallel": 5.602836608886719e-06, + "grad_norm": 6.530845642089844, + "learning_rate": 7.787197965239508e-07, + "loss": 0.3626, + "mean_token_accuracy": 0.8743302226066589, + "num_tokens": 71496635.0, + "step": 1838 + }, + { + "epoch": 0.23393970232794808, + "ewc_loss": 0.006195068359375, + "ewc_loss_diag": 6.16908073425293e-06, + "ewc_loss_parallel": 5.5730342864990234e-06, + "grad_norm": 6.84526252746582, + "learning_rate": 7.791437049597287e-07, + "loss": 0.4719, + "mean_token_accuracy": 0.84404456615448, + "num_tokens": 71527268.0, + "step": 1839 + }, + { + "epoch": 0.2340669126065386, + "ewc_loss": 0.006317138671875, + "ewc_loss_diag": 6.16908073425293e-06, + "ewc_loss_parallel": 5.692243576049805e-06, + "grad_norm": 6.572174072265625, + "learning_rate": 7.795676133955065e-07, + "loss": 0.405, + "mean_token_accuracy": 0.8647201061248779, + "num_tokens": 71566411.0, + "step": 1840 + }, + { + "epoch": 0.2341941228851291, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 6.16908073425293e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.531970024108887, + "learning_rate": 7.799915218312844e-07, + "loss": 0.4066, + "mean_token_accuracy": 0.8632310628890991, + "num_tokens": 71600926.0, + "step": 1841 + }, + { + "epoch": 0.23432133316371964, + "ewc_loss": 0.006317138671875, + "ewc_loss_diag": 6.198883056640625e-06, + "ewc_loss_parallel": 5.692243576049805e-06, + "grad_norm": 6.493643283843994, + "learning_rate": 7.804154302670622e-07, + "loss": 0.3629, + "mean_token_accuracy": 0.8784511089324951, + "num_tokens": 71640953.0, + "step": 1842 + }, + { + "epoch": 0.23444854344231014, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 6.198883056640625e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.507047176361084, + "learning_rate": 7.808393387028402e-07, + "loss": 0.3976, + "mean_token_accuracy": 0.8643423318862915, + "num_tokens": 71682023.0, + "step": 1843 + }, + { + "epoch": 0.23457575372090064, + "ewc_loss": 0.00628662109375, + "ewc_loss_diag": 6.198883056640625e-06, + "ewc_loss_parallel": 5.662441253662109e-06, + "grad_norm": 6.618675231933594, + "learning_rate": 7.81263247138618e-07, + "loss": 0.4327, + "mean_token_accuracy": 0.8541176319122314, + "num_tokens": 71720407.0, + "step": 1844 + }, + { + "epoch": 0.23470296399949117, + "ewc_loss": 0.006256103515625, + "ewc_loss_diag": 6.198883056640625e-06, + "ewc_loss_parallel": 5.632638931274414e-06, + "grad_norm": 6.4993672370910645, + "learning_rate": 7.816871555743959e-07, + "loss": 0.4029, + "mean_token_accuracy": 0.8668879270553589, + "num_tokens": 71758389.0, + "step": 1845 + }, + { + "epoch": 0.23483017427808167, + "ewc_loss": 0.00628662109375, + "ewc_loss_diag": 6.198883056640625e-06, + "ewc_loss_parallel": 5.662441253662109e-06, + "grad_norm": 6.461213111877441, + "learning_rate": 7.821110640101738e-07, + "loss": 0.4112, + "mean_token_accuracy": 0.8610974550247192, + "num_tokens": 71803370.0, + "step": 1846 + }, + { + "epoch": 0.23495738455667217, + "ewc_loss": 0.00628662109375, + "ewc_loss_diag": 6.22868537902832e-06, + "ewc_loss_parallel": 5.662441253662109e-06, + "grad_norm": 6.487809181213379, + "learning_rate": 7.825349724459517e-07, + "loss": 0.3925, + "mean_token_accuracy": 0.8701953887939453, + "num_tokens": 71843994.0, + "step": 1847 + }, + { + "epoch": 0.2350845948352627, + "ewc_loss": 0.00628662109375, + "ewc_loss_diag": 6.22868537902832e-06, + "ewc_loss_parallel": 5.662441253662109e-06, + "grad_norm": 6.601606369018555, + "learning_rate": 7.829588808817294e-07, + "loss": 0.3952, + "mean_token_accuracy": 0.8674957752227783, + "num_tokens": 71886055.0, + "step": 1848 + }, + { + "epoch": 0.2352118051138532, + "ewc_loss": 0.00634765625, + "ewc_loss_diag": 6.258487701416016e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.831803798675537, + "learning_rate": 7.833827893175074e-07, + "loss": 0.4323, + "mean_token_accuracy": 0.8523635864257812, + "num_tokens": 71916955.0, + "step": 1849 + }, + { + "epoch": 0.2353390153924437, + "ewc_loss": 0.006317138671875, + "ewc_loss_diag": 6.22868537902832e-06, + "ewc_loss_parallel": 5.692243576049805e-06, + "grad_norm": 6.634726047515869, + "learning_rate": 7.838066977532852e-07, + "loss": 0.3904, + "mean_token_accuracy": 0.8692389726638794, + "num_tokens": 71955775.0, + "step": 1850 + }, + { + "epoch": 0.23546622567103423, + "ewc_loss": 0.00634765625, + "ewc_loss_diag": 6.258487701416016e-06, + "ewc_loss_parallel": 5.692243576049805e-06, + "grad_norm": 6.528904438018799, + "learning_rate": 7.842306061890632e-07, + "loss": 0.4308, + "mean_token_accuracy": 0.8581713438034058, + "num_tokens": 71996768.0, + "step": 1851 + }, + { + "epoch": 0.23559343594962473, + "ewc_loss": 0.00628662109375, + "ewc_loss_diag": 6.258487701416016e-06, + "ewc_loss_parallel": 5.662441253662109e-06, + "grad_norm": 6.622517108917236, + "learning_rate": 7.84654514624841e-07, + "loss": 0.4103, + "mean_token_accuracy": 0.863502025604248, + "num_tokens": 72034367.0, + "step": 1852 + }, + { + "epoch": 0.23572064622821523, + "ewc_loss": 0.00634765625, + "ewc_loss_diag": 6.258487701416016e-06, + "ewc_loss_parallel": 5.692243576049805e-06, + "grad_norm": 6.61928653717041, + "learning_rate": 7.850784230606188e-07, + "loss": 0.432, + "mean_token_accuracy": 0.8601608872413635, + "num_tokens": 72071669.0, + "step": 1853 + }, + { + "epoch": 0.23584785650680576, + "ewc_loss": 0.00634765625, + "ewc_loss_diag": 6.258487701416016e-06, + "ewc_loss_parallel": 5.692243576049805e-06, + "grad_norm": 6.441192626953125, + "learning_rate": 7.855023314963968e-07, + "loss": 0.4238, + "mean_token_accuracy": 0.860629677772522, + "num_tokens": 72117423.0, + "step": 1854 + }, + { + "epoch": 0.23597506678539626, + "ewc_loss": 0.00628662109375, + "ewc_loss_diag": 6.258487701416016e-06, + "ewc_loss_parallel": 5.662441253662109e-06, + "grad_norm": 6.584507942199707, + "learning_rate": 7.859262399321746e-07, + "loss": 0.4166, + "mean_token_accuracy": 0.858519434928894, + "num_tokens": 72159840.0, + "step": 1855 + }, + { + "epoch": 0.23610227706398676, + "ewc_loss": 0.00634765625, + "ewc_loss_diag": 6.258487701416016e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.46100378036499, + "learning_rate": 7.863501483679524e-07, + "loss": 0.3625, + "mean_token_accuracy": 0.876491129398346, + "num_tokens": 72201540.0, + "step": 1856 + }, + { + "epoch": 0.2362294873425773, + "ewc_loss": 0.00634765625, + "ewc_loss_diag": 6.258487701416016e-06, + "ewc_loss_parallel": 5.692243576049805e-06, + "grad_norm": 6.54313325881958, + "learning_rate": 7.867740568037303e-07, + "loss": 0.377, + "mean_token_accuracy": 0.8691385984420776, + "num_tokens": 72239948.0, + "step": 1857 + }, + { + "epoch": 0.2363566976211678, + "ewc_loss": 0.00634765625, + "ewc_loss_diag": 6.258487701416016e-06, + "ewc_loss_parallel": 5.692243576049805e-06, + "grad_norm": 6.673386573791504, + "learning_rate": 7.871979652395082e-07, + "loss": 0.4001, + "mean_token_accuracy": 0.8658114671707153, + "num_tokens": 72274811.0, + "step": 1858 + }, + { + "epoch": 0.2364839078997583, + "ewc_loss": 0.00634765625, + "ewc_loss_diag": 6.258487701416016e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.456371307373047, + "learning_rate": 7.876218736752861e-07, + "loss": 0.3735, + "mean_token_accuracy": 0.8762548565864563, + "num_tokens": 72316040.0, + "step": 1859 + }, + { + "epoch": 0.23661111817834882, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.288290023803711e-06, + "ewc_loss_parallel": 5.7220458984375e-06, + "grad_norm": 6.634702205657959, + "learning_rate": 7.88045782111064e-07, + "loss": 0.4261, + "mean_token_accuracy": 0.8590856194496155, + "num_tokens": 72357464.0, + "step": 1860 + }, + { + "epoch": 0.23673832845693932, + "ewc_loss": 0.00640869140625, + "ewc_loss_diag": 6.288290023803711e-06, + "ewc_loss_parallel": 5.781650543212891e-06, + "grad_norm": 6.486454486846924, + "learning_rate": 7.884696905468418e-07, + "loss": 0.3663, + "mean_token_accuracy": 0.8773002028465271, + "num_tokens": 72400368.0, + "step": 1861 + }, + { + "epoch": 0.23686553873552982, + "ewc_loss": 0.00634765625, + "ewc_loss_diag": 6.288290023803711e-06, + "ewc_loss_parallel": 5.692243576049805e-06, + "grad_norm": 6.571753978729248, + "learning_rate": 7.888935989826198e-07, + "loss": 0.4109, + "mean_token_accuracy": 0.8612557649612427, + "num_tokens": 72440309.0, + "step": 1862 + }, + { + "epoch": 0.23699274901412035, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.288290023803711e-06, + "ewc_loss_parallel": 5.7220458984375e-06, + "grad_norm": 6.610426425933838, + "learning_rate": 7.893175074183976e-07, + "loss": 0.4025, + "mean_token_accuracy": 0.8632113933563232, + "num_tokens": 72479688.0, + "step": 1863 + }, + { + "epoch": 0.23711995929271085, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.288290023803711e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.501398086547852, + "learning_rate": 7.897414158541754e-07, + "loss": 0.3937, + "mean_token_accuracy": 0.8665414452552795, + "num_tokens": 72520377.0, + "step": 1864 + }, + { + "epoch": 0.23724716957130135, + "ewc_loss": 0.00634765625, + "ewc_loss_diag": 6.318092346191406e-06, + "ewc_loss_parallel": 5.692243576049805e-06, + "grad_norm": 6.624228477478027, + "learning_rate": 7.901653242899533e-07, + "loss": 0.4231, + "mean_token_accuracy": 0.8618302345275879, + "num_tokens": 72558431.0, + "step": 1865 + }, + { + "epoch": 0.23737437984989188, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.318092346191406e-06, + "ewc_loss_parallel": 5.7220458984375e-06, + "grad_norm": 6.662827014923096, + "learning_rate": 7.905892327257312e-07, + "loss": 0.4043, + "mean_token_accuracy": 0.8636813163757324, + "num_tokens": 72595729.0, + "step": 1866 + }, + { + "epoch": 0.23750159012848238, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.318092346191406e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.5762248039245605, + "learning_rate": 7.910131411615091e-07, + "loss": 0.3974, + "mean_token_accuracy": 0.8694329857826233, + "num_tokens": 72633965.0, + "step": 1867 + }, + { + "epoch": 0.2376288004070729, + "ewc_loss": 0.006317138671875, + "ewc_loss_diag": 6.318092346191406e-06, + "ewc_loss_parallel": 5.662441253662109e-06, + "grad_norm": 6.552317142486572, + "learning_rate": 7.91437049597287e-07, + "loss": 0.3875, + "mean_token_accuracy": 0.8705294132232666, + "num_tokens": 72673967.0, + "step": 1868 + }, + { + "epoch": 0.2377560106856634, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.318092346191406e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.611135482788086, + "learning_rate": 7.918609580330648e-07, + "loss": 0.4454, + "mean_token_accuracy": 0.8535864353179932, + "num_tokens": 72717658.0, + "step": 1869 + }, + { + "epoch": 0.2378832209642539, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.318092346191406e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.621760368347168, + "learning_rate": 7.922848664688428e-07, + "loss": 0.4214, + "mean_token_accuracy": 0.8598263263702393, + "num_tokens": 72757389.0, + "step": 1870 + }, + { + "epoch": 0.23801043124284443, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.318092346191406e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.573599815368652, + "learning_rate": 7.927087749046205e-07, + "loss": 0.3536, + "mean_token_accuracy": 0.8802379369735718, + "num_tokens": 72794823.0, + "step": 1871 + }, + { + "epoch": 0.23813764152143493, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.318092346191406e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.647446155548096, + "learning_rate": 7.931326833403983e-07, + "loss": 0.3862, + "mean_token_accuracy": 0.8688136339187622, + "num_tokens": 72828744.0, + "step": 1872 + }, + { + "epoch": 0.23826485180002543, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.318092346191406e-06, + "ewc_loss_parallel": 5.7220458984375e-06, + "grad_norm": 6.571037769317627, + "learning_rate": 7.935565917761763e-07, + "loss": 0.382, + "mean_token_accuracy": 0.8716037273406982, + "num_tokens": 72874090.0, + "step": 1873 + }, + { + "epoch": 0.23839206207861596, + "ewc_loss": 0.00634765625, + "ewc_loss_diag": 6.318092346191406e-06, + "ewc_loss_parallel": 5.692243576049805e-06, + "grad_norm": 6.685146331787109, + "learning_rate": 7.939805002119541e-07, + "loss": 0.4117, + "mean_token_accuracy": 0.8605250120162964, + "num_tokens": 72909909.0, + "step": 1874 + }, + { + "epoch": 0.23851927235720646, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.318092346191406e-06, + "ewc_loss_parallel": 5.7220458984375e-06, + "grad_norm": 6.672237873077393, + "learning_rate": 7.944044086477321e-07, + "loss": 0.4169, + "mean_token_accuracy": 0.8608970046043396, + "num_tokens": 72950163.0, + "step": 1875 + }, + { + "epoch": 0.23864648263579696, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.318092346191406e-06, + "ewc_loss_parallel": 5.7220458984375e-06, + "grad_norm": 6.591289043426514, + "learning_rate": 7.948283170835099e-07, + "loss": 0.3665, + "mean_token_accuracy": 0.8743035793304443, + "num_tokens": 72987394.0, + "step": 1876 + }, + { + "epoch": 0.2387736929143875, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.377696990966797e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.440837860107422, + "learning_rate": 7.952522255192878e-07, + "loss": 0.3428, + "mean_token_accuracy": 0.881840705871582, + "num_tokens": 73030253.0, + "step": 1877 + }, + { + "epoch": 0.238900903192978, + "ewc_loss": 0.00634765625, + "ewc_loss_diag": 6.377696990966797e-06, + "ewc_loss_parallel": 5.692243576049805e-06, + "grad_norm": 6.616383075714111, + "learning_rate": 7.956761339550657e-07, + "loss": 0.3546, + "mean_token_accuracy": 0.879179060459137, + "num_tokens": 73069706.0, + "step": 1878 + }, + { + "epoch": 0.2390281134715685, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.377696990966797e-06, + "ewc_loss_parallel": 5.7220458984375e-06, + "grad_norm": 6.676611423492432, + "learning_rate": 7.961000423908435e-07, + "loss": 0.4001, + "mean_token_accuracy": 0.8707424402236938, + "num_tokens": 73110097.0, + "step": 1879 + }, + { + "epoch": 0.23915532375015902, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.407499313354492e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.642672538757324, + "learning_rate": 7.965239508266214e-07, + "loss": 0.389, + "mean_token_accuracy": 0.8688993453979492, + "num_tokens": 73148518.0, + "step": 1880 + }, + { + "epoch": 0.23928253402874952, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.407499313354492e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.575906276702881, + "learning_rate": 7.969478592623993e-07, + "loss": 0.3574, + "mean_token_accuracy": 0.8758917450904846, + "num_tokens": 73187257.0, + "step": 1881 + }, + { + "epoch": 0.23940974430734002, + "ewc_loss": 0.00634765625, + "ewc_loss_diag": 6.4373016357421875e-06, + "ewc_loss_parallel": 5.692243576049805e-06, + "grad_norm": 6.594869613647461, + "learning_rate": 7.973717676981771e-07, + "loss": 0.3879, + "mean_token_accuracy": 0.8691682815551758, + "num_tokens": 73226680.0, + "step": 1882 + }, + { + "epoch": 0.23953695458593055, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.4373016357421875e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.596930980682373, + "learning_rate": 7.977956761339551e-07, + "loss": 0.3554, + "mean_token_accuracy": 0.8784970045089722, + "num_tokens": 73263766.0, + "step": 1883 + }, + { + "epoch": 0.23966416486452105, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.4373016357421875e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.632097244262695, + "learning_rate": 7.982195845697329e-07, + "loss": 0.4675, + "mean_token_accuracy": 0.8440237045288086, + "num_tokens": 73305125.0, + "step": 1884 + }, + { + "epoch": 0.23979137514311155, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.467103958129883e-06, + "ewc_loss_parallel": 5.7220458984375e-06, + "grad_norm": 6.642877101898193, + "learning_rate": 7.986434930055108e-07, + "loss": 0.366, + "mean_token_accuracy": 0.8748534321784973, + "num_tokens": 73341245.0, + "step": 1885 + }, + { + "epoch": 0.23991858542170208, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.467103958129883e-06, + "ewc_loss_parallel": 5.7220458984375e-06, + "grad_norm": 7.907421588897705, + "learning_rate": 7.990674014412886e-07, + "loss": 0.4013, + "mean_token_accuracy": 0.8663660287857056, + "num_tokens": 73386166.0, + "step": 1886 + }, + { + "epoch": 0.24004579570029258, + "ewc_loss": 0.0067138671875, + "ewc_loss_diag": 6.467103958129883e-06, + "ewc_loss_parallel": 6.079673767089844e-06, + "grad_norm": 6.689626693725586, + "learning_rate": 7.994913098770665e-07, + "loss": 0.3936, + "mean_token_accuracy": 0.8673033714294434, + "num_tokens": 73426340.0, + "step": 1887 + }, + { + "epoch": 0.24017300597888308, + "ewc_loss": 0.006317138671875, + "ewc_loss_diag": 6.467103958129883e-06, + "ewc_loss_parallel": 5.662441253662109e-06, + "grad_norm": 6.5470075607299805, + "learning_rate": 7.999152183128444e-07, + "loss": 0.4225, + "mean_token_accuracy": 0.8583730459213257, + "num_tokens": 73463766.0, + "step": 1888 + }, + { + "epoch": 0.2403002162574736, + "ewc_loss": 0.006378173828125, + "ewc_loss_diag": 6.496906280517578e-06, + "ewc_loss_parallel": 5.751848220825195e-06, + "grad_norm": 6.663452625274658, + "learning_rate": 8.003391267486223e-07, + "loss": 0.3728, + "mean_token_accuracy": 0.8730841279029846, + "num_tokens": 73505079.0, + "step": 1889 + }, + { + "epoch": 0.2404274265360641, + "ewc_loss": 0.00640869140625, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.781650543212891e-06, + "grad_norm": 6.740760326385498, + "learning_rate": 8.007630351844001e-07, + "loss": 0.3782, + "mean_token_accuracy": 0.8730250597000122, + "num_tokens": 73537668.0, + "step": 1890 + }, + { + "epoch": 0.24055463681465464, + "ewc_loss": 0.00653076171875, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.900859832763672e-06, + "grad_norm": 6.585613250732422, + "learning_rate": 8.011869436201781e-07, + "loss": 0.3755, + "mean_token_accuracy": 0.8731220960617065, + "num_tokens": 73573470.0, + "step": 1891 + }, + { + "epoch": 0.24068184709324514, + "ewc_loss": 0.00640869140625, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.7220458984375e-06, + "grad_norm": 6.54466438293457, + "learning_rate": 8.016108520559559e-07, + "loss": 0.3958, + "mean_token_accuracy": 0.866547703742981, + "num_tokens": 73618257.0, + "step": 1892 + }, + { + "epoch": 0.24080905737183564, + "ewc_loss": 0.00640869140625, + "ewc_loss_diag": 6.5267086029052734e-06, + "ewc_loss_parallel": 5.781650543212891e-06, + "grad_norm": 6.729666233062744, + "learning_rate": 8.020347604917338e-07, + "loss": 0.367, + "mean_token_accuracy": 0.8746193647384644, + "num_tokens": 73653772.0, + "step": 1893 + }, + { + "epoch": 0.24093626765042617, + "ewc_loss": 0.0064697265625, + "ewc_loss_diag": 6.5267086029052734e-06, + "ewc_loss_parallel": 5.841255187988281e-06, + "grad_norm": 6.6556477546691895, + "learning_rate": 8.024586689275116e-07, + "loss": 0.3927, + "mean_token_accuracy": 0.8688453435897827, + "num_tokens": 73693884.0, + "step": 1894 + }, + { + "epoch": 0.24106347792901667, + "ewc_loss": 0.006439208984375, + "ewc_loss_diag": 6.5267086029052734e-06, + "ewc_loss_parallel": 5.811452865600586e-06, + "grad_norm": 6.637680530548096, + "learning_rate": 8.028825773632894e-07, + "loss": 0.3896, + "mean_token_accuracy": 0.870978832244873, + "num_tokens": 73731514.0, + "step": 1895 + }, + { + "epoch": 0.24119068820760717, + "ewc_loss": 0.006500244140625, + "ewc_loss_diag": 6.5267086029052734e-06, + "ewc_loss_parallel": 5.8710575103759766e-06, + "grad_norm": 6.667625904083252, + "learning_rate": 8.033064857990674e-07, + "loss": 0.3312, + "mean_token_accuracy": 0.8862754106521606, + "num_tokens": 73768229.0, + "step": 1896 + }, + { + "epoch": 0.2413178984861977, + "ewc_loss": 0.0064697265625, + "ewc_loss_diag": 6.5267086029052734e-06, + "ewc_loss_parallel": 5.841255187988281e-06, + "grad_norm": 6.609668254852295, + "learning_rate": 8.037303942348452e-07, + "loss": 0.3806, + "mean_token_accuracy": 0.8715438842773438, + "num_tokens": 73805684.0, + "step": 1897 + }, + { + "epoch": 0.2414451087647882, + "ewc_loss": 0.006439208984375, + "ewc_loss_diag": 6.5267086029052734e-06, + "ewc_loss_parallel": 5.811452865600586e-06, + "grad_norm": 6.632684707641602, + "learning_rate": 8.041543026706231e-07, + "loss": 0.3743, + "mean_token_accuracy": 0.8735024929046631, + "num_tokens": 73845022.0, + "step": 1898 + }, + { + "epoch": 0.2415723190433787, + "ewc_loss": 0.006439208984375, + "ewc_loss_diag": 6.5267086029052734e-06, + "ewc_loss_parallel": 5.811452865600586e-06, + "grad_norm": 6.674437999725342, + "learning_rate": 8.04578211106401e-07, + "loss": 0.3874, + "mean_token_accuracy": 0.869623064994812, + "num_tokens": 73882171.0, + "step": 1899 + }, + { + "epoch": 0.24169952932196923, + "ewc_loss": 0.00653076171875, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.8710575103759766e-06, + "grad_norm": 6.652509689331055, + "learning_rate": 8.050021195421789e-07, + "loss": 0.4194, + "mean_token_accuracy": 0.8594394326210022, + "num_tokens": 73922129.0, + "step": 1900 + }, + { + "epoch": 0.24182673960055973, + "ewc_loss": 0.0064697265625, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.841255187988281e-06, + "grad_norm": 6.7546706199646, + "learning_rate": 8.054260279779567e-07, + "loss": 0.3802, + "mean_token_accuracy": 0.8719325661659241, + "num_tokens": 73959332.0, + "step": 1901 + }, + { + "epoch": 0.24195394987915023, + "ewc_loss": 0.0064697265625, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.841255187988281e-06, + "grad_norm": 6.66086483001709, + "learning_rate": 8.058499364137346e-07, + "loss": 0.418, + "mean_token_accuracy": 0.859449565410614, + "num_tokens": 74000261.0, + "step": 1902 + }, + { + "epoch": 0.24208116015774075, + "ewc_loss": 0.00653076171875, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.8710575103759766e-06, + "grad_norm": 6.570493221282959, + "learning_rate": 8.062738448495124e-07, + "loss": 0.3489, + "mean_token_accuracy": 0.8804872632026672, + "num_tokens": 74042635.0, + "step": 1903 + }, + { + "epoch": 0.24220837043633126, + "ewc_loss": 0.00653076171875, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.8710575103759766e-06, + "grad_norm": 6.609396457672119, + "learning_rate": 8.066977532852904e-07, + "loss": 0.3416, + "mean_token_accuracy": 0.8860334157943726, + "num_tokens": 74082633.0, + "step": 1904 + }, + { + "epoch": 0.24233558071492176, + "ewc_loss": 0.0064697265625, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.811452865600586e-06, + "grad_norm": 6.790255069732666, + "learning_rate": 8.071216617210682e-07, + "loss": 0.4111, + "mean_token_accuracy": 0.8635635375976562, + "num_tokens": 74115690.0, + "step": 1905 + }, + { + "epoch": 0.24246279099351228, + "ewc_loss": 0.0064697265625, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.811452865600586e-06, + "grad_norm": 6.593344688415527, + "learning_rate": 8.075455701568461e-07, + "loss": 0.3843, + "mean_token_accuracy": 0.8713432550430298, + "num_tokens": 74154783.0, + "step": 1906 + }, + { + "epoch": 0.24259000127210278, + "ewc_loss": 0.00640869140625, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.781650543212891e-06, + "grad_norm": 6.54823637008667, + "learning_rate": 8.07969478592624e-07, + "loss": 0.3761, + "mean_token_accuracy": 0.875527560710907, + "num_tokens": 74197403.0, + "step": 1907 + }, + { + "epoch": 0.24271721155069328, + "ewc_loss": 0.0064697265625, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.841255187988281e-06, + "grad_norm": 6.699301242828369, + "learning_rate": 8.083933870284019e-07, + "loss": 0.3503, + "mean_token_accuracy": 0.8838797807693481, + "num_tokens": 74229111.0, + "step": 1908 + }, + { + "epoch": 0.2428444218292838, + "ewc_loss": 0.00653076171875, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.900859832763672e-06, + "grad_norm": 6.701574802398682, + "learning_rate": 8.088172954641796e-07, + "loss": 0.3766, + "mean_token_accuracy": 0.8717827796936035, + "num_tokens": 74266252.0, + "step": 1909 + }, + { + "epoch": 0.2429716321078743, + "ewc_loss": 0.0064697265625, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.841255187988281e-06, + "grad_norm": 6.863373279571533, + "learning_rate": 8.092412038999576e-07, + "loss": 0.4197, + "mean_token_accuracy": 0.8576071858406067, + "num_tokens": 74297795.0, + "step": 1910 + }, + { + "epoch": 0.24309884238646481, + "ewc_loss": 0.006591796875, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.930662155151367e-06, + "grad_norm": 6.810490131378174, + "learning_rate": 8.096651123357354e-07, + "loss": 0.432, + "mean_token_accuracy": 0.8553308248519897, + "num_tokens": 74332883.0, + "step": 1911 + }, + { + "epoch": 0.24322605266505534, + "ewc_loss": 0.0064697265625, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.811452865600586e-06, + "grad_norm": 6.684571743011475, + "learning_rate": 8.100890207715134e-07, + "loss": 0.4266, + "mean_token_accuracy": 0.8589619994163513, + "num_tokens": 74369769.0, + "step": 1912 + }, + { + "epoch": 0.24335326294364584, + "ewc_loss": 0.00653076171875, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.900859832763672e-06, + "grad_norm": 6.742810249328613, + "learning_rate": 8.105129292072912e-07, + "loss": 0.3859, + "mean_token_accuracy": 0.8703447580337524, + "num_tokens": 74409583.0, + "step": 1913 + }, + { + "epoch": 0.24348047322223634, + "ewc_loss": 0.00653076171875, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.8710575103759766e-06, + "grad_norm": 6.742048263549805, + "learning_rate": 8.10936837643069e-07, + "loss": 0.421, + "mean_token_accuracy": 0.8608646988868713, + "num_tokens": 74444531.0, + "step": 1914 + }, + { + "epoch": 0.24360768350082687, + "ewc_loss": 0.006591796875, + "ewc_loss_diag": 6.586313247680664e-06, + "ewc_loss_parallel": 5.930662155151367e-06, + "grad_norm": 6.694882869720459, + "learning_rate": 8.11360746078847e-07, + "loss": 0.3821, + "mean_token_accuracy": 0.8702567219734192, + "num_tokens": 74482367.0, + "step": 1915 + }, + { + "epoch": 0.24373489377941737, + "ewc_loss": 0.00653076171875, + "ewc_loss_diag": 6.556510925292969e-06, + "ewc_loss_parallel": 5.900859832763672e-06, + "grad_norm": 6.536222457885742, + "learning_rate": 8.117846545146248e-07, + "loss": 0.3381, + "mean_token_accuracy": 0.8814934492111206, + "num_tokens": 74524344.0, + "step": 1916 + }, + { + "epoch": 0.2438621040580079, + "ewc_loss": 0.006591796875, + "ewc_loss_diag": 6.586313247680664e-06, + "ewc_loss_parallel": 5.930662155151367e-06, + "grad_norm": 6.770202159881592, + "learning_rate": 8.122085629504026e-07, + "loss": 0.421, + "mean_token_accuracy": 0.8549420833587646, + "num_tokens": 74564939.0, + "step": 1917 + }, + { + "epoch": 0.2439893143365984, + "ewc_loss": 0.006622314453125, + "ewc_loss_diag": 6.586313247680664e-06, + "ewc_loss_parallel": 5.9604644775390625e-06, + "grad_norm": 6.7123589515686035, + "learning_rate": 8.126324713861805e-07, + "loss": 0.3971, + "mean_token_accuracy": 0.8680916428565979, + "num_tokens": 74607999.0, + "step": 1918 + }, + { + "epoch": 0.2441165246151889, + "ewc_loss": 0.00665283203125, + "ewc_loss_diag": 6.586313247680664e-06, + "ewc_loss_parallel": 5.990266799926758e-06, + "grad_norm": 6.699489116668701, + "learning_rate": 8.130563798219584e-07, + "loss": 0.3777, + "mean_token_accuracy": 0.8752071857452393, + "num_tokens": 74648726.0, + "step": 1919 + }, + { + "epoch": 0.24424373489377943, + "ewc_loss": 0.006591796875, + "ewc_loss_diag": 6.586313247680664e-06, + "ewc_loss_parallel": 5.930662155151367e-06, + "grad_norm": 6.667773246765137, + "learning_rate": 8.134802882577363e-07, + "loss": 0.4037, + "mean_token_accuracy": 0.8681272268295288, + "num_tokens": 74689596.0, + "step": 1920 + }, + { + "epoch": 0.24437094517236993, + "ewc_loss": 0.00665283203125, + "ewc_loss_diag": 6.645917892456055e-06, + "ewc_loss_parallel": 5.990266799926758e-06, + "grad_norm": 6.573232650756836, + "learning_rate": 8.139041966935142e-07, + "loss": 0.3984, + "mean_token_accuracy": 0.8685575723648071, + "num_tokens": 74737829.0, + "step": 1921 + }, + { + "epoch": 0.24449815545096043, + "ewc_loss": 0.006622314453125, + "ewc_loss_diag": 6.645917892456055e-06, + "ewc_loss_parallel": 5.9604644775390625e-06, + "grad_norm": 6.835758209228516, + "learning_rate": 8.14328105129292e-07, + "loss": 0.4559, + "mean_token_accuracy": 0.8557384014129639, + "num_tokens": 74775412.0, + "step": 1922 + }, + { + "epoch": 0.24462536572955096, + "ewc_loss": 0.006683349609375, + "ewc_loss_diag": 6.645917892456055e-06, + "ewc_loss_parallel": 6.020069122314453e-06, + "grad_norm": 6.803566932678223, + "learning_rate": 8.1475201356507e-07, + "loss": 0.4111, + "mean_token_accuracy": 0.8639788031578064, + "num_tokens": 74811660.0, + "step": 1923 + }, + { + "epoch": 0.24475257600814146, + "ewc_loss": 0.006622314453125, + "ewc_loss_diag": 6.645917892456055e-06, + "ewc_loss_parallel": 5.9604644775390625e-06, + "grad_norm": 6.678756237030029, + "learning_rate": 8.151759220008477e-07, + "loss": 0.3825, + "mean_token_accuracy": 0.8689385652542114, + "num_tokens": 74855050.0, + "step": 1924 + }, + { + "epoch": 0.24487978628673196, + "ewc_loss": 0.00665283203125, + "ewc_loss_diag": 6.645917892456055e-06, + "ewc_loss_parallel": 5.990266799926758e-06, + "grad_norm": 6.833189010620117, + "learning_rate": 8.155998304366256e-07, + "loss": 0.4022, + "mean_token_accuracy": 0.8683784008026123, + "num_tokens": 74888862.0, + "step": 1925 + }, + { + "epoch": 0.2450069965653225, + "ewc_loss": 0.006622314453125, + "ewc_loss_diag": 6.645917892456055e-06, + "ewc_loss_parallel": 5.9604644775390625e-06, + "grad_norm": 6.865089416503906, + "learning_rate": 8.160237388724035e-07, + "loss": 0.4256, + "mean_token_accuracy": 0.8572566509246826, + "num_tokens": 74923096.0, + "step": 1926 + }, + { + "epoch": 0.245134206843913, + "ewc_loss": 0.006561279296875, + "ewc_loss_diag": 6.645917892456055e-06, + "ewc_loss_parallel": 5.900859832763672e-06, + "grad_norm": 6.643838405609131, + "learning_rate": 8.164476473081814e-07, + "loss": 0.3994, + "mean_token_accuracy": 0.8664265275001526, + "num_tokens": 74961746.0, + "step": 1927 + }, + { + "epoch": 0.2452614171225035, + "ewc_loss": 0.00665283203125, + "ewc_loss_diag": 6.645917892456055e-06, + "ewc_loss_parallel": 5.990266799926758e-06, + "grad_norm": 6.589580535888672, + "learning_rate": 8.168715557439593e-07, + "loss": 0.3531, + "mean_token_accuracy": 0.8813301920890808, + "num_tokens": 75003688.0, + "step": 1928 + }, + { + "epoch": 0.24538862740109402, + "ewc_loss": 0.006622314453125, + "ewc_loss_diag": 6.645917892456055e-06, + "ewc_loss_parallel": 5.9604644775390625e-06, + "grad_norm": 6.710680961608887, + "learning_rate": 8.172954641797372e-07, + "loss": 0.3729, + "mean_token_accuracy": 0.8747743368148804, + "num_tokens": 75044620.0, + "step": 1929 + }, + { + "epoch": 0.24551583767968452, + "ewc_loss": 0.006683349609375, + "ewc_loss_diag": 6.645917892456055e-06, + "ewc_loss_parallel": 6.020069122314453e-06, + "grad_norm": 6.687868118286133, + "learning_rate": 8.17719372615515e-07, + "loss": 0.4633, + "mean_token_accuracy": 0.8459817171096802, + "num_tokens": 75093007.0, + "step": 1930 + }, + { + "epoch": 0.24564304795827502, + "ewc_loss": 0.006622314453125, + "ewc_loss_diag": 6.645917892456055e-06, + "ewc_loss_parallel": 5.9604644775390625e-06, + "grad_norm": 6.745787620544434, + "learning_rate": 8.18143281051293e-07, + "loss": 0.436, + "mean_token_accuracy": 0.8551106452941895, + "num_tokens": 75133509.0, + "step": 1931 + }, + { + "epoch": 0.24577025823686555, + "ewc_loss": 0.006622314453125, + "ewc_loss_diag": 6.645917892456055e-06, + "ewc_loss_parallel": 5.9604644775390625e-06, + "grad_norm": 6.737939357757568, + "learning_rate": 8.185671894870707e-07, + "loss": 0.4531, + "mean_token_accuracy": 0.8507301807403564, + "num_tokens": 75173296.0, + "step": 1932 + }, + { + "epoch": 0.24589746851545605, + "ewc_loss": 0.00665283203125, + "ewc_loss_diag": 6.645917892456055e-06, + "ewc_loss_parallel": 5.990266799926758e-06, + "grad_norm": 6.811801433563232, + "learning_rate": 8.189910979228485e-07, + "loss": 0.3675, + "mean_token_accuracy": 0.8748144507408142, + "num_tokens": 75205781.0, + "step": 1933 + }, + { + "epoch": 0.24602467879404655, + "ewc_loss": 0.00665283203125, + "ewc_loss_diag": 6.705522537231445e-06, + "ewc_loss_parallel": 5.990266799926758e-06, + "grad_norm": 6.82023286819458, + "learning_rate": 8.194150063586265e-07, + "loss": 0.45, + "mean_token_accuracy": 0.8527441620826721, + "num_tokens": 75243690.0, + "step": 1934 + }, + { + "epoch": 0.24615188907263708, + "ewc_loss": 0.00665283203125, + "ewc_loss_diag": 6.705522537231445e-06, + "ewc_loss_parallel": 5.990266799926758e-06, + "grad_norm": 6.732302665710449, + "learning_rate": 8.198389147944043e-07, + "loss": 0.4126, + "mean_token_accuracy": 0.8587644696235657, + "num_tokens": 75282259.0, + "step": 1935 + }, + { + "epoch": 0.24627909935122758, + "ewc_loss": 0.00665283203125, + "ewc_loss_diag": 6.705522537231445e-06, + "ewc_loss_parallel": 5.990266799926758e-06, + "grad_norm": 6.794791221618652, + "learning_rate": 8.202628232301823e-07, + "loss": 0.4099, + "mean_token_accuracy": 0.8644284605979919, + "num_tokens": 75318735.0, + "step": 1936 + }, + { + "epoch": 0.24640630962981808, + "ewc_loss": 0.006683349609375, + "ewc_loss_diag": 6.705522537231445e-06, + "ewc_loss_parallel": 6.020069122314453e-06, + "grad_norm": 6.604631423950195, + "learning_rate": 8.206867316659601e-07, + "loss": 0.4305, + "mean_token_accuracy": 0.8539160490036011, + "num_tokens": 75366616.0, + "step": 1937 + }, + { + "epoch": 0.2465335199084086, + "ewc_loss": 0.006683349609375, + "ewc_loss_diag": 6.705522537231445e-06, + "ewc_loss_parallel": 6.020069122314453e-06, + "grad_norm": 6.636267185211182, + "learning_rate": 8.21110640101738e-07, + "loss": 0.3292, + "mean_token_accuracy": 0.8846604824066162, + "num_tokens": 75407516.0, + "step": 1938 + }, + { + "epoch": 0.2466607301869991, + "ewc_loss": 0.006622314453125, + "ewc_loss_diag": 6.705522537231445e-06, + "ewc_loss_parallel": 5.9604644775390625e-06, + "grad_norm": 6.738718509674072, + "learning_rate": 8.215345485375159e-07, + "loss": 0.4418, + "mean_token_accuracy": 0.8523452877998352, + "num_tokens": 75452279.0, + "step": 1939 + }, + { + "epoch": 0.2467879404655896, + "ewc_loss": 0.0067138671875, + "ewc_loss_diag": 6.705522537231445e-06, + "ewc_loss_parallel": 6.0498714447021484e-06, + "grad_norm": 6.737045764923096, + "learning_rate": 8.219584569732937e-07, + "loss": 0.4045, + "mean_token_accuracy": 0.8618646860122681, + "num_tokens": 75493443.0, + "step": 1940 + }, + { + "epoch": 0.24691515074418013, + "ewc_loss": 0.006683349609375, + "ewc_loss_diag": 6.705522537231445e-06, + "ewc_loss_parallel": 6.020069122314453e-06, + "grad_norm": 6.735186576843262, + "learning_rate": 8.223823654090715e-07, + "loss": 0.347, + "mean_token_accuracy": 0.8818578124046326, + "num_tokens": 75530988.0, + "step": 1941 + }, + { + "epoch": 0.24704236102277063, + "ewc_loss": 0.006683349609375, + "ewc_loss_diag": 6.705522537231445e-06, + "ewc_loss_parallel": 6.020069122314453e-06, + "grad_norm": 6.718809127807617, + "learning_rate": 8.228062738448495e-07, + "loss": 0.3823, + "mean_token_accuracy": 0.868482232093811, + "num_tokens": 75572898.0, + "step": 1942 + }, + { + "epoch": 0.24716957130136116, + "ewc_loss": 0.006683349609375, + "ewc_loss_diag": 6.705522537231445e-06, + "ewc_loss_parallel": 6.020069122314453e-06, + "grad_norm": 6.764143466949463, + "learning_rate": 8.232301822806273e-07, + "loss": 0.4105, + "mean_token_accuracy": 0.8628132343292236, + "num_tokens": 75613168.0, + "step": 1943 + }, + { + "epoch": 0.24729678157995166, + "ewc_loss": 0.0067138671875, + "ewc_loss_diag": 6.735324859619141e-06, + "ewc_loss_parallel": 6.0498714447021484e-06, + "grad_norm": 6.772488594055176, + "learning_rate": 8.236540907164053e-07, + "loss": 0.392, + "mean_token_accuracy": 0.8696484565734863, + "num_tokens": 75650715.0, + "step": 1944 + }, + { + "epoch": 0.24742399185854216, + "ewc_loss": 0.00665283203125, + "ewc_loss_diag": 6.794929504394531e-06, + "ewc_loss_parallel": 5.990266799926758e-06, + "grad_norm": 6.77156400680542, + "learning_rate": 8.240779991521831e-07, + "loss": 0.3829, + "mean_token_accuracy": 0.8707807064056396, + "num_tokens": 75690804.0, + "step": 1945 + }, + { + "epoch": 0.2475512021371327, + "ewc_loss": 0.0067138671875, + "ewc_loss_diag": 6.794929504394531e-06, + "ewc_loss_parallel": 6.0498714447021484e-06, + "grad_norm": 6.734562873840332, + "learning_rate": 8.24501907587961e-07, + "loss": 0.4128, + "mean_token_accuracy": 0.860905647277832, + "num_tokens": 75735092.0, + "step": 1946 + }, + { + "epoch": 0.2476784124157232, + "ewc_loss": 0.00665283203125, + "ewc_loss_diag": 6.794929504394531e-06, + "ewc_loss_parallel": 5.990266799926758e-06, + "grad_norm": 6.946446418762207, + "learning_rate": 8.249258160237388e-07, + "loss": 0.4236, + "mean_token_accuracy": 0.8639587759971619, + "num_tokens": 75770506.0, + "step": 1947 + }, + { + "epoch": 0.2478056226943137, + "ewc_loss": 0.006683349609375, + "ewc_loss_diag": 6.794929504394531e-06, + "ewc_loss_parallel": 6.020069122314453e-06, + "grad_norm": 6.8355793952941895, + "learning_rate": 8.253497244595167e-07, + "loss": 0.3947, + "mean_token_accuracy": 0.8684121370315552, + "num_tokens": 75809504.0, + "step": 1948 + }, + { + "epoch": 0.24793283297290422, + "ewc_loss": 0.00665283203125, + "ewc_loss_diag": 6.794929504394531e-06, + "ewc_loss_parallel": 5.990266799926758e-06, + "grad_norm": 6.720508098602295, + "learning_rate": 8.257736328952945e-07, + "loss": 0.4355, + "mean_token_accuracy": 0.8542047739028931, + "num_tokens": 75853012.0, + "step": 1949 + }, + { + "epoch": 0.24806004325149472, + "ewc_loss": 0.006622314453125, + "ewc_loss_diag": 6.794929504394531e-06, + "ewc_loss_parallel": 5.9604644775390625e-06, + "grad_norm": 6.790255546569824, + "learning_rate": 8.261975413310725e-07, + "loss": 0.4127, + "mean_token_accuracy": 0.86378014087677, + "num_tokens": 75890985.0, + "step": 1950 + }, + { + "epoch": 0.24818725353008522, + "ewc_loss": 0.006683349609375, + "ewc_loss_diag": 6.794929504394531e-06, + "ewc_loss_parallel": 6.020069122314453e-06, + "grad_norm": 6.636184215545654, + "learning_rate": 8.266214497668503e-07, + "loss": 0.3654, + "mean_token_accuracy": 0.8779835104942322, + "num_tokens": 75933912.0, + "step": 1951 + }, + { + "epoch": 0.24831446380867575, + "ewc_loss": 0.0067138671875, + "ewc_loss_diag": 6.854534149169922e-06, + "ewc_loss_parallel": 6.020069122314453e-06, + "grad_norm": 6.849499702453613, + "learning_rate": 8.270453582026283e-07, + "loss": 0.4365, + "mean_token_accuracy": 0.8527287244796753, + "num_tokens": 75972735.0, + "step": 1952 + }, + { + "epoch": 0.24844167408726625, + "ewc_loss": 0.0067138671875, + "ewc_loss_diag": 6.854534149169922e-06, + "ewc_loss_parallel": 6.0498714447021484e-06, + "grad_norm": 6.845413684844971, + "learning_rate": 8.274692666384061e-07, + "loss": 0.3912, + "mean_token_accuracy": 0.8711954951286316, + "num_tokens": 76006714.0, + "step": 1953 + }, + { + "epoch": 0.24856888436585675, + "ewc_loss": 0.0067138671875, + "ewc_loss_diag": 6.854534149169922e-06, + "ewc_loss_parallel": 6.0498714447021484e-06, + "grad_norm": 6.839261531829834, + "learning_rate": 8.27893175074184e-07, + "loss": 0.3975, + "mean_token_accuracy": 0.865515947341919, + "num_tokens": 76040789.0, + "step": 1954 + }, + { + "epoch": 0.24869609464444728, + "ewc_loss": 0.0067138671875, + "ewc_loss_diag": 6.854534149169922e-06, + "ewc_loss_parallel": 6.0498714447021484e-06, + "grad_norm": 6.765071392059326, + "learning_rate": 8.283170835099618e-07, + "loss": 0.381, + "mean_token_accuracy": 0.8737573027610779, + "num_tokens": 76078033.0, + "step": 1955 + }, + { + "epoch": 0.24882330492303778, + "ewc_loss": 0.00665283203125, + "ewc_loss_diag": 6.854534149169922e-06, + "ewc_loss_parallel": 5.9604644775390625e-06, + "grad_norm": 6.708216667175293, + "learning_rate": 8.287409919457396e-07, + "loss": 0.3681, + "mean_token_accuracy": 0.8744558691978455, + "num_tokens": 76117618.0, + "step": 1956 + }, + { + "epoch": 0.24895051520162828, + "ewc_loss": 0.00665283203125, + "ewc_loss_diag": 6.854534149169922e-06, + "ewc_loss_parallel": 5.990266799926758e-06, + "grad_norm": 6.7596659660339355, + "learning_rate": 8.291649003815175e-07, + "loss": 0.4138, + "mean_token_accuracy": 0.8623589277267456, + "num_tokens": 76155954.0, + "step": 1957 + }, + { + "epoch": 0.2490777254802188, + "ewc_loss": 0.0067138671875, + "ewc_loss_diag": 6.854534149169922e-06, + "ewc_loss_parallel": 6.0498714447021484e-06, + "grad_norm": 6.717151641845703, + "learning_rate": 8.295888088172954e-07, + "loss": 0.4154, + "mean_token_accuracy": 0.8620488047599792, + "num_tokens": 76196905.0, + "step": 1958 + }, + { + "epoch": 0.2492049357588093, + "ewc_loss": 0.0067138671875, + "ewc_loss_diag": 6.854534149169922e-06, + "ewc_loss_parallel": 6.020069122314453e-06, + "grad_norm": 6.735530376434326, + "learning_rate": 8.300127172530733e-07, + "loss": 0.3354, + "mean_token_accuracy": 0.8838320970535278, + "num_tokens": 76234441.0, + "step": 1959 + }, + { + "epoch": 0.2493321460373998, + "ewc_loss": 0.006805419921875, + "ewc_loss_diag": 6.9141387939453125e-06, + "ewc_loss_parallel": 6.109476089477539e-06, + "grad_norm": 6.845365524291992, + "learning_rate": 8.304366256888512e-07, + "loss": 0.3771, + "mean_token_accuracy": 0.8725415468215942, + "num_tokens": 76270770.0, + "step": 1960 + }, + { + "epoch": 0.24945935631599034, + "ewc_loss": 0.006744384765625, + "ewc_loss_diag": 6.9141387939453125e-06, + "ewc_loss_parallel": 6.0498714447021484e-06, + "grad_norm": 6.8280816078186035, + "learning_rate": 8.308605341246291e-07, + "loss": 0.3838, + "mean_token_accuracy": 0.871944785118103, + "num_tokens": 76307758.0, + "step": 1961 + }, + { + "epoch": 0.24958656659458084, + "ewc_loss": 0.006683349609375, + "ewc_loss_diag": 6.9141387939453125e-06, + "ewc_loss_parallel": 5.990266799926758e-06, + "grad_norm": 6.776634216308594, + "learning_rate": 8.312844425604068e-07, + "loss": 0.3861, + "mean_token_accuracy": 0.8739794492721558, + "num_tokens": 76343920.0, + "step": 1962 + }, + { + "epoch": 0.24971377687317134, + "ewc_loss": 0.00677490234375, + "ewc_loss_diag": 6.9141387939453125e-06, + "ewc_loss_parallel": 6.079673767089844e-06, + "grad_norm": 6.749993801116943, + "learning_rate": 8.317083509961848e-07, + "loss": 0.4077, + "mean_token_accuracy": 0.8644511699676514, + "num_tokens": 76386335.0, + "step": 1963 + }, + { + "epoch": 0.24984098715176187, + "ewc_loss": 0.0067138671875, + "ewc_loss_diag": 6.9141387939453125e-06, + "ewc_loss_parallel": 6.020069122314453e-06, + "grad_norm": 6.726916313171387, + "learning_rate": 8.321322594319626e-07, + "loss": 0.3653, + "mean_token_accuracy": 0.8742915391921997, + "num_tokens": 76426890.0, + "step": 1964 + }, + { + "epoch": 0.24996819743035237, + "ewc_loss": 0.0067138671875, + "ewc_loss_diag": 6.9141387939453125e-06, + "ewc_loss_parallel": 6.020069122314453e-06, + "grad_norm": 6.7679924964904785, + "learning_rate": 8.325561678677405e-07, + "loss": 0.4138, + "mean_token_accuracy": 0.8620553612709045, + "num_tokens": 76469480.0, + "step": 1965 + }, + { + "epoch": 0.2500954077089429, + "ewc_loss": 0.006744384765625, + "ewc_loss_diag": 6.9141387939453125e-06, + "ewc_loss_parallel": 6.0498714447021484e-06, + "grad_norm": 6.7667741775512695, + "learning_rate": 8.329800763035184e-07, + "loss": 0.4266, + "mean_token_accuracy": 0.8593320846557617, + "num_tokens": 76510816.0, + "step": 1966 + }, + { + "epoch": 0.25022261798753337, + "ewc_loss": 0.006744384765625, + "ewc_loss_diag": 6.9141387939453125e-06, + "ewc_loss_parallel": 6.0498714447021484e-06, + "grad_norm": 6.80720853805542, + "learning_rate": 8.334039847392963e-07, + "loss": 0.3818, + "mean_token_accuracy": 0.8697172403335571, + "num_tokens": 76544917.0, + "step": 1967 + }, + { + "epoch": 0.2503498282661239, + "ewc_loss": 0.006744384765625, + "ewc_loss_diag": 6.943941116333008e-06, + "ewc_loss_parallel": 6.0498714447021484e-06, + "grad_norm": 6.950234413146973, + "learning_rate": 8.338278931750742e-07, + "loss": 0.3717, + "mean_token_accuracy": 0.8744217157363892, + "num_tokens": 76578054.0, + "step": 1968 + }, + { + "epoch": 0.2504770385447144, + "ewc_loss": 0.00677490234375, + "ewc_loss_diag": 6.943941116333008e-06, + "ewc_loss_parallel": 6.079673767089844e-06, + "grad_norm": 6.821166515350342, + "learning_rate": 8.342518016108521e-07, + "loss": 0.3905, + "mean_token_accuracy": 0.864607572555542, + "num_tokens": 76613357.0, + "step": 1969 + }, + { + "epoch": 0.2506042488233049, + "ewc_loss": 0.006744384765625, + "ewc_loss_diag": 6.943941116333008e-06, + "ewc_loss_parallel": 6.0498714447021484e-06, + "grad_norm": 6.892116069793701, + "learning_rate": 8.346757100466298e-07, + "loss": 0.3872, + "mean_token_accuracy": 0.8709229230880737, + "num_tokens": 76647288.0, + "step": 1970 + }, + { + "epoch": 0.2507314591018954, + "ewc_loss": 0.00677490234375, + "ewc_loss_diag": 6.943941116333008e-06, + "ewc_loss_parallel": 6.079673767089844e-06, + "grad_norm": 6.873899459838867, + "learning_rate": 8.350996184824078e-07, + "loss": 0.4246, + "mean_token_accuracy": 0.8547849059104919, + "num_tokens": 76684734.0, + "step": 1971 + }, + { + "epoch": 0.25085866938048595, + "ewc_loss": 0.00677490234375, + "ewc_loss_diag": 6.943941116333008e-06, + "ewc_loss_parallel": 6.079673767089844e-06, + "grad_norm": 6.794357776641846, + "learning_rate": 8.355235269181856e-07, + "loss": 0.3978, + "mean_token_accuracy": 0.8641730546951294, + "num_tokens": 76721472.0, + "step": 1972 + }, + { + "epoch": 0.2509858796590764, + "ewc_loss": 0.006744384765625, + "ewc_loss_diag": 6.943941116333008e-06, + "ewc_loss_parallel": 6.0498714447021484e-06, + "grad_norm": 6.73670768737793, + "learning_rate": 8.359474353539635e-07, + "loss": 0.3689, + "mean_token_accuracy": 0.8753647804260254, + "num_tokens": 76762607.0, + "step": 1973 + }, + { + "epoch": 0.25111308993766696, + "ewc_loss": 0.00677490234375, + "ewc_loss_diag": 6.943941116333008e-06, + "ewc_loss_parallel": 6.079673767089844e-06, + "grad_norm": 6.714682102203369, + "learning_rate": 8.363713437897414e-07, + "loss": 0.4537, + "mean_token_accuracy": 0.8498340249061584, + "num_tokens": 76812105.0, + "step": 1974 + }, + { + "epoch": 0.2512403002162575, + "ewc_loss": 0.00677490234375, + "ewc_loss_diag": 6.943941116333008e-06, + "ewc_loss_parallel": 6.079673767089844e-06, + "grad_norm": 6.853599548339844, + "learning_rate": 8.367952522255193e-07, + "loss": 0.398, + "mean_token_accuracy": 0.8641560077667236, + "num_tokens": 76852598.0, + "step": 1975 + }, + { + "epoch": 0.25136751049484796, + "ewc_loss": 0.00677490234375, + "ewc_loss_diag": 7.0035457611083984e-06, + "ewc_loss_parallel": 6.079673767089844e-06, + "grad_norm": 6.8155837059021, + "learning_rate": 8.372191606612972e-07, + "loss": 0.3899, + "mean_token_accuracy": 0.8694400191307068, + "num_tokens": 76891395.0, + "step": 1976 + }, + { + "epoch": 0.2514947207734385, + "ewc_loss": 0.006805419921875, + "ewc_loss_diag": 7.0035457611083984e-06, + "ewc_loss_parallel": 6.109476089477539e-06, + "grad_norm": 6.893911838531494, + "learning_rate": 8.376430690970749e-07, + "loss": 0.3687, + "mean_token_accuracy": 0.8753435611724854, + "num_tokens": 76922281.0, + "step": 1977 + }, + { + "epoch": 0.251621931052029, + "ewc_loss": 0.006805419921875, + "ewc_loss_diag": 7.0035457611083984e-06, + "ewc_loss_parallel": 6.109476089477539e-06, + "grad_norm": 6.707235336303711, + "learning_rate": 8.380669775328528e-07, + "loss": 0.3649, + "mean_token_accuracy": 0.8767647743225098, + "num_tokens": 76962580.0, + "step": 1978 + }, + { + "epoch": 0.25174914133061954, + "ewc_loss": 0.00677490234375, + "ewc_loss_diag": 7.0035457611083984e-06, + "ewc_loss_parallel": 6.079673767089844e-06, + "grad_norm": 6.771359443664551, + "learning_rate": 8.384908859686307e-07, + "loss": 0.3611, + "mean_token_accuracy": 0.8780300617218018, + "num_tokens": 77004453.0, + "step": 1979 + }, + { + "epoch": 0.25187635160921, + "ewc_loss": 0.00677490234375, + "ewc_loss_diag": 7.0035457611083984e-06, + "ewc_loss_parallel": 6.079673767089844e-06, + "grad_norm": 6.790439605712891, + "learning_rate": 8.389147944044086e-07, + "loss": 0.384, + "mean_token_accuracy": 0.8681888580322266, + "num_tokens": 77049172.0, + "step": 1980 + }, + { + "epoch": 0.25200356188780054, + "ewc_loss": 0.00677490234375, + "ewc_loss_diag": 7.0035457611083984e-06, + "ewc_loss_parallel": 6.079673767089844e-06, + "grad_norm": 6.890732765197754, + "learning_rate": 8.393387028401864e-07, + "loss": 0.4168, + "mean_token_accuracy": 0.8580267429351807, + "num_tokens": 77087153.0, + "step": 1981 + }, + { + "epoch": 0.25213077216639107, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.0035457611083984e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.961048126220703, + "learning_rate": 8.397626112759644e-07, + "loss": 0.4367, + "mean_token_accuracy": 0.8554213047027588, + "num_tokens": 77120507.0, + "step": 1982 + }, + { + "epoch": 0.25225798244498154, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.0035457611083984e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.97731351852417, + "learning_rate": 8.401865197117422e-07, + "loss": 0.3916, + "mean_token_accuracy": 0.8684583902359009, + "num_tokens": 77151061.0, + "step": 1983 + }, + { + "epoch": 0.25238519272357207, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.0035457611083984e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.7695770263671875, + "learning_rate": 8.406104281475202e-07, + "loss": 0.3644, + "mean_token_accuracy": 0.8759102821350098, + "num_tokens": 77191635.0, + "step": 1984 + }, + { + "epoch": 0.2525124030021626, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.063150405883789e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.71283483505249, + "learning_rate": 8.410343365832979e-07, + "loss": 0.3819, + "mean_token_accuracy": 0.870668888092041, + "num_tokens": 77233518.0, + "step": 1985 + }, + { + "epoch": 0.2526396132807531, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.063150405883789e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.773528575897217, + "learning_rate": 8.414582450190758e-07, + "loss": 0.3822, + "mean_token_accuracy": 0.8713170886039734, + "num_tokens": 77278340.0, + "step": 1986 + }, + { + "epoch": 0.2527668235593436, + "ewc_loss": 0.00689697265625, + "ewc_loss_diag": 7.063150405883789e-06, + "ewc_loss_parallel": 6.198883056640625e-06, + "grad_norm": 6.95112419128418, + "learning_rate": 8.418821534548537e-07, + "loss": 0.4639, + "mean_token_accuracy": 0.8492848873138428, + "num_tokens": 77318889.0, + "step": 1987 + }, + { + "epoch": 0.25289403383793413, + "ewc_loss": 0.006866455078125, + "ewc_loss_diag": 7.063150405883789e-06, + "ewc_loss_parallel": 6.16908073425293e-06, + "grad_norm": 6.827263832092285, + "learning_rate": 8.423060618906316e-07, + "loss": 0.3906, + "mean_token_accuracy": 0.867729663848877, + "num_tokens": 77359451.0, + "step": 1988 + }, + { + "epoch": 0.2530212441165246, + "ewc_loss": 0.006805419921875, + "ewc_loss_diag": 7.063150405883789e-06, + "ewc_loss_parallel": 6.109476089477539e-06, + "grad_norm": 6.734288215637207, + "learning_rate": 8.427299703264095e-07, + "loss": 0.4054, + "mean_token_accuracy": 0.8638496398925781, + "num_tokens": 77401979.0, + "step": 1989 + }, + { + "epoch": 0.25314845439511513, + "ewc_loss": 0.006866455078125, + "ewc_loss_diag": 7.092952728271484e-06, + "ewc_loss_parallel": 6.16908073425293e-06, + "grad_norm": 6.763587951660156, + "learning_rate": 8.431538787621874e-07, + "loss": 0.4073, + "mean_token_accuracy": 0.8643242120742798, + "num_tokens": 77445568.0, + "step": 1990 + }, + { + "epoch": 0.25327566467370566, + "ewc_loss": 0.006866455078125, + "ewc_loss_diag": 7.092952728271484e-06, + "ewc_loss_parallel": 6.16908073425293e-06, + "grad_norm": 6.809014320373535, + "learning_rate": 8.435777871979652e-07, + "loss": 0.3696, + "mean_token_accuracy": 0.8741141557693481, + "num_tokens": 77486600.0, + "step": 1991 + }, + { + "epoch": 0.25340287495229613, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.092952728271484e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.850742816925049, + "learning_rate": 8.440016956337432e-07, + "loss": 0.3946, + "mean_token_accuracy": 0.8694427013397217, + "num_tokens": 77525745.0, + "step": 1992 + }, + { + "epoch": 0.25353008523088666, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.092952728271484e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.85006856918335, + "learning_rate": 8.444256040695209e-07, + "loss": 0.4043, + "mean_token_accuracy": 0.8663696050643921, + "num_tokens": 77567963.0, + "step": 1993 + }, + { + "epoch": 0.2536572955094772, + "ewc_loss": 0.006866455078125, + "ewc_loss_diag": 7.092952728271484e-06, + "ewc_loss_parallel": 6.16908073425293e-06, + "grad_norm": 6.81576681137085, + "learning_rate": 8.448495125052988e-07, + "loss": 0.4008, + "mean_token_accuracy": 0.8658722639083862, + "num_tokens": 77606720.0, + "step": 1994 + }, + { + "epoch": 0.25378450578806766, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.092952728271484e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.827599048614502, + "learning_rate": 8.452734209410767e-07, + "loss": 0.3363, + "mean_token_accuracy": 0.8870180249214172, + "num_tokens": 77641794.0, + "step": 1995 + }, + { + "epoch": 0.2539117160666582, + "ewc_loss": 0.006805419921875, + "ewc_loss_diag": 7.092952728271484e-06, + "ewc_loss_parallel": 6.109476089477539e-06, + "grad_norm": 6.8755011558532715, + "learning_rate": 8.456973293768545e-07, + "loss": 0.3706, + "mean_token_accuracy": 0.8724390268325806, + "num_tokens": 77677174.0, + "step": 1996 + }, + { + "epoch": 0.2540389263452487, + "ewc_loss": 0.006866455078125, + "ewc_loss_diag": 7.092952728271484e-06, + "ewc_loss_parallel": 6.16908073425293e-06, + "grad_norm": 6.835495471954346, + "learning_rate": 8.461212378126325e-07, + "loss": 0.423, + "mean_token_accuracy": 0.8580954074859619, + "num_tokens": 77717211.0, + "step": 1997 + }, + { + "epoch": 0.2541661366238392, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.12275505065918e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.956977367401123, + "learning_rate": 8.465451462484103e-07, + "loss": 0.4391, + "mean_token_accuracy": 0.8572062849998474, + "num_tokens": 77754187.0, + "step": 1998 + }, + { + "epoch": 0.2542933469024297, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.12275505065918e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.882411956787109, + "learning_rate": 8.469690546841882e-07, + "loss": 0.3904, + "mean_token_accuracy": 0.8683310747146606, + "num_tokens": 77792865.0, + "step": 1999 + }, + { + "epoch": 0.25442055718102025, + "ewc_loss": 0.006805419921875, + "ewc_loss_diag": 7.12275505065918e-06, + "ewc_loss_parallel": 6.109476089477539e-06, + "grad_norm": 6.867037773132324, + "learning_rate": 8.47392963119966e-07, + "loss": 0.3693, + "mean_token_accuracy": 0.8740857243537903, + "num_tokens": 77828720.0, + "step": 2000 + }, + { + "epoch": 0.2545477674596107, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.12275505065918e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.875713348388672, + "learning_rate": 8.478168715557439e-07, + "loss": 0.3978, + "mean_token_accuracy": 0.8687962293624878, + "num_tokens": 77866003.0, + "step": 2001 + }, + { + "epoch": 0.25467497773820125, + "ewc_loss": 0.00677490234375, + "ewc_loss_diag": 7.12275505065918e-06, + "ewc_loss_parallel": 6.079673767089844e-06, + "grad_norm": 6.856675148010254, + "learning_rate": 8.482407799915217e-07, + "loss": 0.4026, + "mean_token_accuracy": 0.8636792302131653, + "num_tokens": 77905528.0, + "step": 2002 + }, + { + "epoch": 0.2548021880167918, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.12275505065918e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.847962856292725, + "learning_rate": 8.486646884272997e-07, + "loss": 0.3923, + "mean_token_accuracy": 0.8674588799476624, + "num_tokens": 77944493.0, + "step": 2003 + }, + { + "epoch": 0.25492939829538225, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.152557373046875e-06, + "ewc_loss_parallel": 6.109476089477539e-06, + "grad_norm": 6.958096981048584, + "learning_rate": 8.490885968630775e-07, + "loss": 0.4104, + "mean_token_accuracy": 0.8642992973327637, + "num_tokens": 77980856.0, + "step": 2004 + }, + { + "epoch": 0.2550566085739728, + "ewc_loss": 0.00689697265625, + "ewc_loss_diag": 7.152557373046875e-06, + "ewc_loss_parallel": 6.16908073425293e-06, + "grad_norm": 6.798332214355469, + "learning_rate": 8.495125052988555e-07, + "loss": 0.4055, + "mean_token_accuracy": 0.8664757013320923, + "num_tokens": 78026257.0, + "step": 2005 + }, + { + "epoch": 0.2551838188525633, + "ewc_loss": 0.00689697265625, + "ewc_loss_diag": 7.152557373046875e-06, + "ewc_loss_parallel": 6.16908073425293e-06, + "grad_norm": 6.766721248626709, + "learning_rate": 8.499364137346333e-07, + "loss": 0.3895, + "mean_token_accuracy": 0.8705719709396362, + "num_tokens": 78068956.0, + "step": 2006 + }, + { + "epoch": 0.2553110291311538, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.152557373046875e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.912528991699219, + "learning_rate": 8.503603221704112e-07, + "loss": 0.4179, + "mean_token_accuracy": 0.8593742847442627, + "num_tokens": 78105267.0, + "step": 2007 + }, + { + "epoch": 0.2554382394097443, + "ewc_loss": 0.00689697265625, + "ewc_loss_diag": 7.152557373046875e-06, + "ewc_loss_parallel": 6.198883056640625e-06, + "grad_norm": 6.805705547332764, + "learning_rate": 8.50784230606189e-07, + "loss": 0.3908, + "mean_token_accuracy": 0.8681331872940063, + "num_tokens": 78144203.0, + "step": 2008 + }, + { + "epoch": 0.25556544968833483, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.152557373046875e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.8864312171936035, + "learning_rate": 8.512081390419669e-07, + "loss": 0.4194, + "mean_token_accuracy": 0.8676828742027283, + "num_tokens": 78186087.0, + "step": 2009 + }, + { + "epoch": 0.2556926599669253, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.18235969543457e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.9824113845825195, + "learning_rate": 8.516320474777447e-07, + "loss": 0.4051, + "mean_token_accuracy": 0.8645821809768677, + "num_tokens": 78221141.0, + "step": 2010 + }, + { + "epoch": 0.25581987024551583, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.18235969543457e-06, + "ewc_loss_parallel": 6.109476089477539e-06, + "grad_norm": 6.837393760681152, + "learning_rate": 8.520559559135227e-07, + "loss": 0.3625, + "mean_token_accuracy": 0.8779889345169067, + "num_tokens": 78258658.0, + "step": 2011 + }, + { + "epoch": 0.25594708052410636, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.18235969543457e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.9140472412109375, + "learning_rate": 8.524798643493005e-07, + "loss": 0.4024, + "mean_token_accuracy": 0.8680693507194519, + "num_tokens": 78298194.0, + "step": 2012 + }, + { + "epoch": 0.25607429080269684, + "ewc_loss": 0.00689697265625, + "ewc_loss_diag": 7.18235969543457e-06, + "ewc_loss_parallel": 6.198883056640625e-06, + "grad_norm": 6.85304594039917, + "learning_rate": 8.529037727850785e-07, + "loss": 0.3826, + "mean_token_accuracy": 0.8690997362136841, + "num_tokens": 78335883.0, + "step": 2013 + }, + { + "epoch": 0.25620150108128736, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.18235969543457e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.855217456817627, + "learning_rate": 8.533276812208563e-07, + "loss": 0.3471, + "mean_token_accuracy": 0.8796986937522888, + "num_tokens": 78371228.0, + "step": 2014 + }, + { + "epoch": 0.2563287113598779, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.18235969543457e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.808842182159424, + "learning_rate": 8.53751589656634e-07, + "loss": 0.3825, + "mean_token_accuracy": 0.8712071180343628, + "num_tokens": 78414537.0, + "step": 2015 + }, + { + "epoch": 0.25645592163846836, + "ewc_loss": 0.00689697265625, + "ewc_loss_diag": 7.18235969543457e-06, + "ewc_loss_parallel": 6.16908073425293e-06, + "grad_norm": 7.069624423980713, + "learning_rate": 8.54175498092412e-07, + "loss": 0.387, + "mean_token_accuracy": 0.8690041899681091, + "num_tokens": 78448319.0, + "step": 2016 + }, + { + "epoch": 0.2565831319170589, + "ewc_loss": 0.00689697265625, + "ewc_loss_diag": 7.18235969543457e-06, + "ewc_loss_parallel": 6.198883056640625e-06, + "grad_norm": 6.950966835021973, + "learning_rate": 8.545994065281898e-07, + "loss": 0.364, + "mean_token_accuracy": 0.8771865963935852, + "num_tokens": 78480872.0, + "step": 2017 + }, + { + "epoch": 0.2567103421956494, + "ewc_loss": 0.0069580078125, + "ewc_loss_diag": 7.18235969543457e-06, + "ewc_loss_parallel": 6.22868537902832e-06, + "grad_norm": 6.874212265014648, + "learning_rate": 8.550233149639677e-07, + "loss": 0.4426, + "mean_token_accuracy": 0.8525397181510925, + "num_tokens": 78521432.0, + "step": 2018 + }, + { + "epoch": 0.2568375524742399, + "ewc_loss": 0.0068359375, + "ewc_loss_diag": 7.18235969543457e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 6.838528156280518, + "learning_rate": 8.554472233997456e-07, + "loss": 0.4203, + "mean_token_accuracy": 0.8587790727615356, + "num_tokens": 78561620.0, + "step": 2019 + }, + { + "epoch": 0.2569647627528304, + "ewc_loss": 0.0069580078125, + "ewc_loss_diag": 7.18235969543457e-06, + "ewc_loss_parallel": 6.258487701416016e-06, + "grad_norm": 7.060171127319336, + "learning_rate": 8.558711318355235e-07, + "loss": 0.4489, + "mean_token_accuracy": 0.8519030809402466, + "num_tokens": 78599179.0, + "step": 2020 + }, + { + "epoch": 0.25709197303142095, + "ewc_loss": 0.007049560546875, + "ewc_loss_diag": 7.212162017822266e-06, + "ewc_loss_parallel": 6.318092346191406e-06, + "grad_norm": 6.815989017486572, + "learning_rate": 8.562950402713014e-07, + "loss": 0.3398, + "mean_token_accuracy": 0.8866835236549377, + "num_tokens": 78634233.0, + "step": 2021 + }, + { + "epoch": 0.2572191833100114, + "ewc_loss": 0.006927490234375, + "ewc_loss_diag": 7.212162017822266e-06, + "ewc_loss_parallel": 6.198883056640625e-06, + "grad_norm": 6.960564613342285, + "learning_rate": 8.567189487070793e-07, + "loss": 0.44, + "mean_token_accuracy": 0.8572226762771606, + "num_tokens": 78671483.0, + "step": 2022 + }, + { + "epoch": 0.25734639358860195, + "ewc_loss": 0.007080078125, + "ewc_loss_diag": 7.212162017822266e-06, + "ewc_loss_parallel": 6.3478946685791016e-06, + "grad_norm": 6.820615768432617, + "learning_rate": 8.57142857142857e-07, + "loss": 0.4339, + "mean_token_accuracy": 0.8552875518798828, + "num_tokens": 78716638.0, + "step": 2023 + }, + { + "epoch": 0.2574736038671925, + "ewc_loss": 0.00689697265625, + "ewc_loss_diag": 7.241964340209961e-06, + "ewc_loss_parallel": 6.16908073425293e-06, + "grad_norm": 6.872918605804443, + "learning_rate": 8.57566765578635e-07, + "loss": 0.3731, + "mean_token_accuracy": 0.8786052465438843, + "num_tokens": 78758912.0, + "step": 2024 + }, + { + "epoch": 0.25760081414578295, + "ewc_loss": 0.0069580078125, + "ewc_loss_diag": 7.241964340209961e-06, + "ewc_loss_parallel": 6.22868537902832e-06, + "grad_norm": 6.841848850250244, + "learning_rate": 8.579906740144128e-07, + "loss": 0.3838, + "mean_token_accuracy": 0.8729652762413025, + "num_tokens": 78803323.0, + "step": 2025 + }, + { + "epoch": 0.2577280244243735, + "ewc_loss": 0.0069580078125, + "ewc_loss_diag": 7.241964340209961e-06, + "ewc_loss_parallel": 6.22868537902832e-06, + "grad_norm": 6.922694206237793, + "learning_rate": 8.584145824501907e-07, + "loss": 0.4285, + "mean_token_accuracy": 0.854153037071228, + "num_tokens": 78842904.0, + "step": 2026 + }, + { + "epoch": 0.257855234702964, + "ewc_loss": 0.007049560546875, + "ewc_loss_diag": 7.241964340209961e-06, + "ewc_loss_parallel": 6.318092346191406e-06, + "grad_norm": 6.888250827789307, + "learning_rate": 8.588384908859686e-07, + "loss": 0.4192, + "mean_token_accuracy": 0.8642803430557251, + "num_tokens": 78882307.0, + "step": 2027 + }, + { + "epoch": 0.25798244498155454, + "ewc_loss": 0.0069580078125, + "ewc_loss_diag": 7.241964340209961e-06, + "ewc_loss_parallel": 6.22868537902832e-06, + "grad_norm": 6.899381637573242, + "learning_rate": 8.592623993217465e-07, + "loss": 0.4057, + "mean_token_accuracy": 0.8668309450149536, + "num_tokens": 78921642.0, + "step": 2028 + }, + { + "epoch": 0.258109655260145, + "ewc_loss": 0.006988525390625, + "ewc_loss_diag": 7.241964340209961e-06, + "ewc_loss_parallel": 6.258487701416016e-06, + "grad_norm": 7.006178379058838, + "learning_rate": 8.596863077575244e-07, + "loss": 0.4329, + "mean_token_accuracy": 0.8587713241577148, + "num_tokens": 78960559.0, + "step": 2029 + }, + { + "epoch": 0.25823686553873554, + "ewc_loss": 0.006927490234375, + "ewc_loss_diag": 7.241964340209961e-06, + "ewc_loss_parallel": 6.198883056640625e-06, + "grad_norm": 6.909628391265869, + "learning_rate": 8.601102161933023e-07, + "loss": 0.4162, + "mean_token_accuracy": 0.8603165149688721, + "num_tokens": 78999443.0, + "step": 2030 + }, + { + "epoch": 0.25836407581732607, + "ewc_loss": 0.0069580078125, + "ewc_loss_diag": 7.241964340209961e-06, + "ewc_loss_parallel": 6.22868537902832e-06, + "grad_norm": 6.828019618988037, + "learning_rate": 8.6053412462908e-07, + "loss": 0.3922, + "mean_token_accuracy": 0.8668580055236816, + "num_tokens": 79039570.0, + "step": 2031 + }, + { + "epoch": 0.25849128609591654, + "ewc_loss": 0.006866455078125, + "ewc_loss_diag": 7.241964340209961e-06, + "ewc_loss_parallel": 6.139278411865234e-06, + "grad_norm": 7.065168380737305, + "learning_rate": 8.60958033064858e-07, + "loss": 0.3797, + "mean_token_accuracy": 0.8749858140945435, + "num_tokens": 79072396.0, + "step": 2032 + }, + { + "epoch": 0.25861849637450707, + "ewc_loss": 0.007049560546875, + "ewc_loss_diag": 7.271766662597656e-06, + "ewc_loss_parallel": 6.318092346191406e-06, + "grad_norm": 6.950237274169922, + "learning_rate": 8.613819415006358e-07, + "loss": 0.4371, + "mean_token_accuracy": 0.8596362471580505, + "num_tokens": 79112013.0, + "step": 2033 + }, + { + "epoch": 0.2587457066530976, + "ewc_loss": 0.006927490234375, + "ewc_loss_diag": 7.271766662597656e-06, + "ewc_loss_parallel": 6.198883056640625e-06, + "grad_norm": 6.841294288635254, + "learning_rate": 8.618058499364137e-07, + "loss": 0.4127, + "mean_token_accuracy": 0.8637139797210693, + "num_tokens": 79154903.0, + "step": 2034 + }, + { + "epoch": 0.25887291693168807, + "ewc_loss": 0.006927490234375, + "ewc_loss_diag": 7.271766662597656e-06, + "ewc_loss_parallel": 6.198883056640625e-06, + "grad_norm": 6.837650775909424, + "learning_rate": 8.622297583721916e-07, + "loss": 0.384, + "mean_token_accuracy": 0.8744106292724609, + "num_tokens": 79200427.0, + "step": 2035 + }, + { + "epoch": 0.2590001272102786, + "ewc_loss": 0.0069580078125, + "ewc_loss_diag": 7.271766662597656e-06, + "ewc_loss_parallel": 6.22868537902832e-06, + "grad_norm": 6.954838275909424, + "learning_rate": 8.626536668079695e-07, + "loss": 0.4061, + "mean_token_accuracy": 0.862230658531189, + "num_tokens": 79235804.0, + "step": 2036 + }, + { + "epoch": 0.2591273374888691, + "ewc_loss": 0.007049560546875, + "ewc_loss_diag": 7.271766662597656e-06, + "ewc_loss_parallel": 6.318092346191406e-06, + "grad_norm": 6.9591450691223145, + "learning_rate": 8.630775752437474e-07, + "loss": 0.3824, + "mean_token_accuracy": 0.8701042532920837, + "num_tokens": 79269647.0, + "step": 2037 + }, + { + "epoch": 0.2592545477674596, + "ewc_loss": 0.007049560546875, + "ewc_loss_diag": 7.3015689849853516e-06, + "ewc_loss_parallel": 6.318092346191406e-06, + "grad_norm": 7.016931056976318, + "learning_rate": 8.635014836795251e-07, + "loss": 0.4284, + "mean_token_accuracy": 0.857756495475769, + "num_tokens": 79306061.0, + "step": 2038 + }, + { + "epoch": 0.2593817580460501, + "ewc_loss": 0.006988525390625, + "ewc_loss_diag": 7.3015689849853516e-06, + "ewc_loss_parallel": 6.258487701416016e-06, + "grad_norm": 6.996081829071045, + "learning_rate": 8.63925392115303e-07, + "loss": 0.3642, + "mean_token_accuracy": 0.8746124505996704, + "num_tokens": 79340450.0, + "step": 2039 + }, + { + "epoch": 0.25950896832464065, + "ewc_loss": 0.007049560546875, + "ewc_loss_diag": 7.3015689849853516e-06, + "ewc_loss_parallel": 6.318092346191406e-06, + "grad_norm": 6.935080528259277, + "learning_rate": 8.643493005510809e-07, + "loss": 0.3719, + "mean_token_accuracy": 0.8742583990097046, + "num_tokens": 79375499.0, + "step": 2040 + }, + { + "epoch": 0.2596361786032311, + "ewc_loss": 0.006988525390625, + "ewc_loss_diag": 7.3015689849853516e-06, + "ewc_loss_parallel": 6.258487701416016e-06, + "grad_norm": 6.839419364929199, + "learning_rate": 8.647732089868588e-07, + "loss": 0.4349, + "mean_token_accuracy": 0.8560250997543335, + "num_tokens": 79418420.0, + "step": 2041 + }, + { + "epoch": 0.25976338888182166, + "ewc_loss": 0.0069580078125, + "ewc_loss_diag": 7.3015689849853516e-06, + "ewc_loss_parallel": 6.22868537902832e-06, + "grad_norm": 6.926244258880615, + "learning_rate": 8.651971174226366e-07, + "loss": 0.3738, + "mean_token_accuracy": 0.8726694583892822, + "num_tokens": 79453302.0, + "step": 2042 + }, + { + "epoch": 0.2598905991604122, + "ewc_loss": 0.007080078125, + "ewc_loss_diag": 7.3015689849853516e-06, + "ewc_loss_parallel": 6.3478946685791016e-06, + "grad_norm": 6.986608982086182, + "learning_rate": 8.656210258584146e-07, + "loss": 0.4277, + "mean_token_accuracy": 0.8575275540351868, + "num_tokens": 79491433.0, + "step": 2043 + }, + { + "epoch": 0.26001780943900266, + "ewc_loss": 0.007110595703125, + "ewc_loss_diag": 7.3015689849853516e-06, + "ewc_loss_parallel": 6.377696990966797e-06, + "grad_norm": 6.967102527618408, + "learning_rate": 8.660449342941924e-07, + "loss": 0.4173, + "mean_token_accuracy": 0.8598426580429077, + "num_tokens": 79531121.0, + "step": 2044 + }, + { + "epoch": 0.2601450197175932, + "ewc_loss": 0.007080078125, + "ewc_loss_diag": 7.3015689849853516e-06, + "ewc_loss_parallel": 6.3478946685791016e-06, + "grad_norm": 6.903610706329346, + "learning_rate": 8.664688427299704e-07, + "loss": 0.4072, + "mean_token_accuracy": 0.8633414506912231, + "num_tokens": 79569617.0, + "step": 2045 + }, + { + "epoch": 0.2602722299961837, + "ewc_loss": 0.007080078125, + "ewc_loss_diag": 7.331371307373047e-06, + "ewc_loss_parallel": 6.3478946685791016e-06, + "grad_norm": 7.053361415863037, + "learning_rate": 8.668927511657481e-07, + "loss": 0.3808, + "mean_token_accuracy": 0.870959997177124, + "num_tokens": 79600648.0, + "step": 2046 + }, + { + "epoch": 0.2603994402747742, + "ewc_loss": 0.0072021484375, + "ewc_loss_diag": 7.361173629760742e-06, + "ewc_loss_parallel": 6.467103958129883e-06, + "grad_norm": 6.998647212982178, + "learning_rate": 8.67316659601526e-07, + "loss": 0.4263, + "mean_token_accuracy": 0.858906090259552, + "num_tokens": 79634317.0, + "step": 2047 + }, + { + "epoch": 0.2605266505533647, + "ewc_loss": 0.007110595703125, + "ewc_loss_diag": 7.361173629760742e-06, + "ewc_loss_parallel": 6.377696990966797e-06, + "grad_norm": 6.852221488952637, + "learning_rate": 8.677405680373039e-07, + "loss": 0.3737, + "mean_token_accuracy": 0.8761477470397949, + "num_tokens": 79676186.0, + "step": 2048 + }, + { + "epoch": 0.26065386083195524, + "ewc_loss": 0.007110595703125, + "ewc_loss_diag": 7.361173629760742e-06, + "ewc_loss_parallel": 6.377696990966797e-06, + "grad_norm": 7.09488582611084, + "learning_rate": 8.681644764730818e-07, + "loss": 0.4085, + "mean_token_accuracy": 0.8591964244842529, + "num_tokens": 79709773.0, + "step": 2049 + }, + { + "epoch": 0.2607810711105457, + "ewc_loss": 0.00732421875, + "ewc_loss_diag": 7.361173629760742e-06, + "ewc_loss_parallel": 6.586313247680664e-06, + "grad_norm": 6.913060665130615, + "learning_rate": 8.685883849088596e-07, + "loss": 0.353, + "mean_token_accuracy": 0.8817905783653259, + "num_tokens": 79748879.0, + "step": 2050 + }, + { + "epoch": 0.26090828138913624, + "ewc_loss": 0.007110595703125, + "ewc_loss_diag": 7.3909759521484375e-06, + "ewc_loss_parallel": 6.377696990966797e-06, + "grad_norm": 6.864279270172119, + "learning_rate": 8.690122933446376e-07, + "loss": 0.408, + "mean_token_accuracy": 0.8660805225372314, + "num_tokens": 79791554.0, + "step": 2051 + }, + { + "epoch": 0.26103549166772677, + "ewc_loss": 0.0072021484375, + "ewc_loss_diag": 7.3909759521484375e-06, + "ewc_loss_parallel": 6.467103958129883e-06, + "grad_norm": 6.914780616760254, + "learning_rate": 8.694362017804154e-07, + "loss": 0.3737, + "mean_token_accuracy": 0.8729091882705688, + "num_tokens": 79831974.0, + "step": 2052 + }, + { + "epoch": 0.26116270194631724, + "ewc_loss": 0.00714111328125, + "ewc_loss_diag": 7.3909759521484375e-06, + "ewc_loss_parallel": 6.407499313354492e-06, + "grad_norm": 6.880730628967285, + "learning_rate": 8.698601102161933e-07, + "loss": 0.4045, + "mean_token_accuracy": 0.8622775077819824, + "num_tokens": 79875488.0, + "step": 2053 + }, + { + "epoch": 0.26128991222490777, + "ewc_loss": 0.007171630859375, + "ewc_loss_diag": 7.3909759521484375e-06, + "ewc_loss_parallel": 6.4373016357421875e-06, + "grad_norm": 6.908905506134033, + "learning_rate": 8.702840186519711e-07, + "loss": 0.3649, + "mean_token_accuracy": 0.879913330078125, + "num_tokens": 79919056.0, + "step": 2054 + }, + { + "epoch": 0.2614171225034983, + "ewc_loss": 0.007171630859375, + "ewc_loss_diag": 7.3909759521484375e-06, + "ewc_loss_parallel": 6.4373016357421875e-06, + "grad_norm": 6.914716720581055, + "learning_rate": 8.70707927087749e-07, + "loss": 0.3649, + "mean_token_accuracy": 0.878075361251831, + "num_tokens": 79960155.0, + "step": 2055 + }, + { + "epoch": 0.2615443327820888, + "ewc_loss": 0.007232666015625, + "ewc_loss_diag": 7.3909759521484375e-06, + "ewc_loss_parallel": 6.496906280517578e-06, + "grad_norm": 7.0906548500061035, + "learning_rate": 8.711318355235269e-07, + "loss": 0.4421, + "mean_token_accuracy": 0.8524621725082397, + "num_tokens": 79996038.0, + "step": 2056 + }, + { + "epoch": 0.2616715430606793, + "ewc_loss": 0.007232666015625, + "ewc_loss_diag": 7.420778274536133e-06, + "ewc_loss_parallel": 6.496906280517578e-06, + "grad_norm": 6.9992451667785645, + "learning_rate": 8.715557439593047e-07, + "loss": 0.3679, + "mean_token_accuracy": 0.8743911981582642, + "num_tokens": 80030448.0, + "step": 2057 + }, + { + "epoch": 0.26179875333926983, + "ewc_loss": 0.00714111328125, + "ewc_loss_diag": 7.420778274536133e-06, + "ewc_loss_parallel": 6.407499313354492e-06, + "grad_norm": 6.951138496398926, + "learning_rate": 8.719796523950826e-07, + "loss": 0.4316, + "mean_token_accuracy": 0.8532420992851257, + "num_tokens": 80069685.0, + "step": 2058 + }, + { + "epoch": 0.2619259636178603, + "ewc_loss": 0.007171630859375, + "ewc_loss_diag": 7.420778274536133e-06, + "ewc_loss_parallel": 6.4373016357421875e-06, + "grad_norm": 6.9379119873046875, + "learning_rate": 8.724035608308605e-07, + "loss": 0.3443, + "mean_token_accuracy": 0.8827028274536133, + "num_tokens": 80102417.0, + "step": 2059 + }, + { + "epoch": 0.26205317389645083, + "ewc_loss": 0.00714111328125, + "ewc_loss_diag": 7.420778274536133e-06, + "ewc_loss_parallel": 6.407499313354492e-06, + "grad_norm": 6.929072856903076, + "learning_rate": 8.728274692666384e-07, + "loss": 0.3917, + "mean_token_accuracy": 0.8681685924530029, + "num_tokens": 80144886.0, + "step": 2060 + }, + { + "epoch": 0.26218038417504136, + "ewc_loss": 0.00726318359375, + "ewc_loss_diag": 7.420778274536133e-06, + "ewc_loss_parallel": 6.5267086029052734e-06, + "grad_norm": 7.0768208503723145, + "learning_rate": 8.732513777024162e-07, + "loss": 0.4098, + "mean_token_accuracy": 0.8640366196632385, + "num_tokens": 80179847.0, + "step": 2061 + }, + { + "epoch": 0.26230759445363183, + "ewc_loss": 0.007232666015625, + "ewc_loss_diag": 7.420778274536133e-06, + "ewc_loss_parallel": 6.496906280517578e-06, + "grad_norm": 6.996383190155029, + "learning_rate": 8.736752861381941e-07, + "loss": 0.438, + "mean_token_accuracy": 0.8540530204772949, + "num_tokens": 80223907.0, + "step": 2062 + }, + { + "epoch": 0.26243480473222236, + "ewc_loss": 0.007110595703125, + "ewc_loss_diag": 7.420778274536133e-06, + "ewc_loss_parallel": 6.377696990966797e-06, + "grad_norm": 6.984488487243652, + "learning_rate": 8.740991945739719e-07, + "loss": 0.3989, + "mean_token_accuracy": 0.8652193546295166, + "num_tokens": 80262910.0, + "step": 2063 + }, + { + "epoch": 0.2625620150108129, + "ewc_loss": 0.00714111328125, + "ewc_loss_diag": 7.420778274536133e-06, + "ewc_loss_parallel": 6.407499313354492e-06, + "grad_norm": 6.933890342712402, + "learning_rate": 8.745231030097499e-07, + "loss": 0.3814, + "mean_token_accuracy": 0.8717963099479675, + "num_tokens": 80307088.0, + "step": 2064 + }, + { + "epoch": 0.26268922528940336, + "ewc_loss": 0.00714111328125, + "ewc_loss_diag": 7.420778274536133e-06, + "ewc_loss_parallel": 6.407499313354492e-06, + "grad_norm": 6.928664684295654, + "learning_rate": 8.749470114455277e-07, + "loss": 0.3686, + "mean_token_accuracy": 0.8746167421340942, + "num_tokens": 80344105.0, + "step": 2065 + }, + { + "epoch": 0.2628164355679939, + "ewc_loss": 0.007171630859375, + "ewc_loss_diag": 7.420778274536133e-06, + "ewc_loss_parallel": 6.4373016357421875e-06, + "grad_norm": 6.970037460327148, + "learning_rate": 8.753709198813056e-07, + "loss": 0.4586, + "mean_token_accuracy": 0.8490111827850342, + "num_tokens": 80388937.0, + "step": 2066 + }, + { + "epoch": 0.2629436458465844, + "ewc_loss": 0.007171630859375, + "ewc_loss_diag": 7.420778274536133e-06, + "ewc_loss_parallel": 6.4373016357421875e-06, + "grad_norm": 6.8247857093811035, + "learning_rate": 8.757948283170835e-07, + "loss": 0.3547, + "mean_token_accuracy": 0.8817360401153564, + "num_tokens": 80432838.0, + "step": 2067 + }, + { + "epoch": 0.2630708561251749, + "ewc_loss": 0.007110595703125, + "ewc_loss_diag": 7.420778274536133e-06, + "ewc_loss_parallel": 6.377696990966797e-06, + "grad_norm": 6.958400249481201, + "learning_rate": 8.762187367528613e-07, + "loss": 0.3708, + "mean_token_accuracy": 0.8743520379066467, + "num_tokens": 80470884.0, + "step": 2068 + }, + { + "epoch": 0.2631980664037654, + "ewc_loss": 0.007171630859375, + "ewc_loss_diag": 7.420778274536133e-06, + "ewc_loss_parallel": 6.4373016357421875e-06, + "grad_norm": 6.913578033447266, + "learning_rate": 8.766426451886392e-07, + "loss": 0.4055, + "mean_token_accuracy": 0.8642663955688477, + "num_tokens": 80516212.0, + "step": 2069 + }, + { + "epoch": 0.26332527668235595, + "ewc_loss": 0.00714111328125, + "ewc_loss_diag": 7.4803829193115234e-06, + "ewc_loss_parallel": 6.407499313354492e-06, + "grad_norm": 7.023342132568359, + "learning_rate": 8.770665536244171e-07, + "loss": 0.4321, + "mean_token_accuracy": 0.8604913949966431, + "num_tokens": 80557777.0, + "step": 2070 + }, + { + "epoch": 0.2634524869609464, + "ewc_loss": 0.00726318359375, + "ewc_loss_diag": 7.4803829193115234e-06, + "ewc_loss_parallel": 6.496906280517578e-06, + "grad_norm": 7.0173659324646, + "learning_rate": 8.774904620601949e-07, + "loss": 0.3759, + "mean_token_accuracy": 0.872890830039978, + "num_tokens": 80592037.0, + "step": 2071 + }, + { + "epoch": 0.26357969723953695, + "ewc_loss": 0.00714111328125, + "ewc_loss_diag": 7.4803829193115234e-06, + "ewc_loss_parallel": 6.407499313354492e-06, + "grad_norm": 6.978922367095947, + "learning_rate": 8.779143704959729e-07, + "loss": 0.3699, + "mean_token_accuracy": 0.8753441572189331, + "num_tokens": 80631913.0, + "step": 2072 + }, + { + "epoch": 0.2637069075181275, + "ewc_loss": 0.00714111328125, + "ewc_loss_diag": 7.4803829193115234e-06, + "ewc_loss_parallel": 6.377696990966797e-06, + "grad_norm": 7.064454555511475, + "learning_rate": 8.783382789317507e-07, + "loss": 0.4079, + "mean_token_accuracy": 0.8652530908584595, + "num_tokens": 80668281.0, + "step": 2073 + }, + { + "epoch": 0.26383411779671795, + "ewc_loss": 0.0072021484375, + "ewc_loss_diag": 7.539987564086914e-06, + "ewc_loss_parallel": 6.4373016357421875e-06, + "grad_norm": 7.170063018798828, + "learning_rate": 8.787621873675286e-07, + "loss": 0.4068, + "mean_token_accuracy": 0.8611236810684204, + "num_tokens": 80700842.0, + "step": 2074 + }, + { + "epoch": 0.2639613280753085, + "ewc_loss": 0.007232666015625, + "ewc_loss_diag": 7.539987564086914e-06, + "ewc_loss_parallel": 6.467103958129883e-06, + "grad_norm": 7.152008533477783, + "learning_rate": 8.791860958033065e-07, + "loss": 0.3782, + "mean_token_accuracy": 0.8726669549942017, + "num_tokens": 80736479.0, + "step": 2075 + }, + { + "epoch": 0.264088538353899, + "ewc_loss": 0.007171630859375, + "ewc_loss_diag": 7.539987564086914e-06, + "ewc_loss_parallel": 6.407499313354492e-06, + "grad_norm": 6.880457878112793, + "learning_rate": 8.796100042390842e-07, + "loss": 0.3595, + "mean_token_accuracy": 0.8802364468574524, + "num_tokens": 80777273.0, + "step": 2076 + }, + { + "epoch": 0.2642157486324895, + "ewc_loss": 0.007110595703125, + "ewc_loss_diag": 7.539987564086914e-06, + "ewc_loss_parallel": 6.3478946685791016e-06, + "grad_norm": 7.02128267288208, + "learning_rate": 8.800339126748622e-07, + "loss": 0.3642, + "mean_token_accuracy": 0.8764264583587646, + "num_tokens": 80811517.0, + "step": 2077 + }, + { + "epoch": 0.26434295891108, + "ewc_loss": 0.00726318359375, + "ewc_loss_diag": 7.539987564086914e-06, + "ewc_loss_parallel": 6.496906280517578e-06, + "grad_norm": 7.049727916717529, + "learning_rate": 8.8045782111064e-07, + "loss": 0.4511, + "mean_token_accuracy": 0.8494156002998352, + "num_tokens": 80849103.0, + "step": 2078 + }, + { + "epoch": 0.26447016918967053, + "ewc_loss": 0.0072021484375, + "ewc_loss_diag": 7.569789886474609e-06, + "ewc_loss_parallel": 6.4373016357421875e-06, + "grad_norm": 7.0281476974487305, + "learning_rate": 8.808817295464179e-07, + "loss": 0.3931, + "mean_token_accuracy": 0.8684579730033875, + "num_tokens": 80886374.0, + "step": 2079 + }, + { + "epoch": 0.26459737946826106, + "ewc_loss": 0.007293701171875, + "ewc_loss_diag": 7.569789886474609e-06, + "ewc_loss_parallel": 6.5267086029052734e-06, + "grad_norm": 7.05360221862793, + "learning_rate": 8.813056379821958e-07, + "loss": 0.3751, + "mean_token_accuracy": 0.8728205561637878, + "num_tokens": 80920651.0, + "step": 2080 + }, + { + "epoch": 0.26472458974685154, + "ewc_loss": 0.007232666015625, + "ewc_loss_diag": 7.569789886474609e-06, + "ewc_loss_parallel": 6.467103958129883e-06, + "grad_norm": 7.042134761810303, + "learning_rate": 8.817295464179737e-07, + "loss": 0.391, + "mean_token_accuracy": 0.8687629699707031, + "num_tokens": 80956333.0, + "step": 2081 + }, + { + "epoch": 0.26485180002544206, + "ewc_loss": 0.007293701171875, + "ewc_loss_diag": 7.569789886474609e-06, + "ewc_loss_parallel": 6.5267086029052734e-06, + "grad_norm": 7.0244927406311035, + "learning_rate": 8.821534548537515e-07, + "loss": 0.4574, + "mean_token_accuracy": 0.8432403802871704, + "num_tokens": 80998666.0, + "step": 2082 + }, + { + "epoch": 0.2649790103040326, + "ewc_loss": 0.00732421875, + "ewc_loss_diag": 7.599592208862305e-06, + "ewc_loss_parallel": 6.556510925292969e-06, + "grad_norm": 6.989797592163086, + "learning_rate": 8.825773632895295e-07, + "loss": 0.3711, + "mean_token_accuracy": 0.874118447303772, + "num_tokens": 81040301.0, + "step": 2083 + }, + { + "epoch": 0.26510622058262306, + "ewc_loss": 0.00726318359375, + "ewc_loss_diag": 7.599592208862305e-06, + "ewc_loss_parallel": 6.496906280517578e-06, + "grad_norm": 6.990845203399658, + "learning_rate": 8.830012717253072e-07, + "loss": 0.3783, + "mean_token_accuracy": 0.8745245933532715, + "num_tokens": 81078916.0, + "step": 2084 + }, + { + "epoch": 0.2652334308612136, + "ewc_loss": 0.007354736328125, + "ewc_loss_diag": 7.599592208862305e-06, + "ewc_loss_parallel": 6.586313247680664e-06, + "grad_norm": 6.929470062255859, + "learning_rate": 8.834251801610852e-07, + "loss": 0.4002, + "mean_token_accuracy": 0.8666149377822876, + "num_tokens": 81125393.0, + "step": 2085 + }, + { + "epoch": 0.2653606411398041, + "ewc_loss": 0.00726318359375, + "ewc_loss_diag": 7.599592208862305e-06, + "ewc_loss_parallel": 6.496906280517578e-06, + "grad_norm": 7.000329494476318, + "learning_rate": 8.83849088596863e-07, + "loss": 0.3678, + "mean_token_accuracy": 0.8780975341796875, + "num_tokens": 81163774.0, + "step": 2086 + }, + { + "epoch": 0.2654878514183946, + "ewc_loss": 0.00738525390625, + "ewc_loss_diag": 7.599592208862305e-06, + "ewc_loss_parallel": 6.616115570068359e-06, + "grad_norm": 7.054783821105957, + "learning_rate": 8.842729970326409e-07, + "loss": 0.3673, + "mean_token_accuracy": 0.877498984336853, + "num_tokens": 81201225.0, + "step": 2087 + }, + { + "epoch": 0.2656150616969851, + "ewc_loss": 0.00738525390625, + "ewc_loss_diag": 7.599592208862305e-06, + "ewc_loss_parallel": 6.616115570068359e-06, + "grad_norm": 6.999528408050537, + "learning_rate": 8.846969054684188e-07, + "loss": 0.4424, + "mean_token_accuracy": 0.8540138006210327, + "num_tokens": 81243880.0, + "step": 2088 + }, + { + "epoch": 0.26574227197557565, + "ewc_loss": 0.00726318359375, + "ewc_loss_diag": 7.62939453125e-06, + "ewc_loss_parallel": 6.496906280517578e-06, + "grad_norm": 7.041396617889404, + "learning_rate": 8.851208139041967e-07, + "loss": 0.3583, + "mean_token_accuracy": 0.878527820110321, + "num_tokens": 81283701.0, + "step": 2089 + }, + { + "epoch": 0.2658694822541661, + "ewc_loss": 0.00726318359375, + "ewc_loss_diag": 7.62939453125e-06, + "ewc_loss_parallel": 6.496906280517578e-06, + "grad_norm": 7.065866470336914, + "learning_rate": 8.855447223399745e-07, + "loss": 0.3953, + "mean_token_accuracy": 0.8655280470848083, + "num_tokens": 81320395.0, + "step": 2090 + }, + { + "epoch": 0.26599669253275665, + "ewc_loss": 0.00726318359375, + "ewc_loss_diag": 7.62939453125e-06, + "ewc_loss_parallel": 6.496906280517578e-06, + "grad_norm": 7.050579071044922, + "learning_rate": 8.859686307757524e-07, + "loss": 0.435, + "mean_token_accuracy": 0.8548669815063477, + "num_tokens": 81359905.0, + "step": 2091 + }, + { + "epoch": 0.2661239028113472, + "ewc_loss": 0.00738525390625, + "ewc_loss_diag": 7.62939453125e-06, + "ewc_loss_parallel": 6.616115570068359e-06, + "grad_norm": 7.133986473083496, + "learning_rate": 8.863925392115302e-07, + "loss": 0.3782, + "mean_token_accuracy": 0.8729895353317261, + "num_tokens": 81395765.0, + "step": 2092 + }, + { + "epoch": 0.26625111308993765, + "ewc_loss": 0.00732421875, + "ewc_loss_diag": 7.62939453125e-06, + "ewc_loss_parallel": 6.556510925292969e-06, + "grad_norm": 6.937082290649414, + "learning_rate": 8.868164476473082e-07, + "loss": 0.3521, + "mean_token_accuracy": 0.8820400238037109, + "num_tokens": 81435043.0, + "step": 2093 + }, + { + "epoch": 0.2663783233685282, + "ewc_loss": 0.0072021484375, + "ewc_loss_diag": 7.62939453125e-06, + "ewc_loss_parallel": 6.4373016357421875e-06, + "grad_norm": 6.9993577003479, + "learning_rate": 8.87240356083086e-07, + "loss": 0.3778, + "mean_token_accuracy": 0.8756444454193115, + "num_tokens": 81475587.0, + "step": 2094 + }, + { + "epoch": 0.2665055336471187, + "ewc_loss": 0.007293701171875, + "ewc_loss_diag": 7.62939453125e-06, + "ewc_loss_parallel": 6.5267086029052734e-06, + "grad_norm": 7.0596137046813965, + "learning_rate": 8.876642645188639e-07, + "loss": 0.4222, + "mean_token_accuracy": 0.8586729764938354, + "num_tokens": 81515196.0, + "step": 2095 + }, + { + "epoch": 0.2666327439257092, + "ewc_loss": 0.00732421875, + "ewc_loss_diag": 7.62939453125e-06, + "ewc_loss_parallel": 6.556510925292969e-06, + "grad_norm": 7.05102014541626, + "learning_rate": 8.880881729546418e-07, + "loss": 0.3534, + "mean_token_accuracy": 0.8794556260108948, + "num_tokens": 81551672.0, + "step": 2096 + }, + { + "epoch": 0.2667599542042997, + "ewc_loss": 0.00732421875, + "ewc_loss_diag": 7.62939453125e-06, + "ewc_loss_parallel": 6.556510925292969e-06, + "grad_norm": 6.933681011199951, + "learning_rate": 8.885120813904197e-07, + "loss": 0.3666, + "mean_token_accuracy": 0.8771207332611084, + "num_tokens": 81595596.0, + "step": 2097 + }, + { + "epoch": 0.26688716448289024, + "ewc_loss": 0.00726318359375, + "ewc_loss_diag": 7.62939453125e-06, + "ewc_loss_parallel": 6.496906280517578e-06, + "grad_norm": 7.180620193481445, + "learning_rate": 8.889359898261976e-07, + "loss": 0.4453, + "mean_token_accuracy": 0.8517931699752808, + "num_tokens": 81630819.0, + "step": 2098 + }, + { + "epoch": 0.2670143747614807, + "ewc_loss": 0.00732421875, + "ewc_loss_diag": 7.68899917602539e-06, + "ewc_loss_parallel": 6.556510925292969e-06, + "grad_norm": 7.000391960144043, + "learning_rate": 8.893598982619753e-07, + "loss": 0.3785, + "mean_token_accuracy": 0.8741821646690369, + "num_tokens": 81670971.0, + "step": 2099 + }, + { + "epoch": 0.26714158504007124, + "ewc_loss": 0.007232666015625, + "ewc_loss_diag": 7.68899917602539e-06, + "ewc_loss_parallel": 6.467103958129883e-06, + "grad_norm": 6.878366470336914, + "learning_rate": 8.897838066977532e-07, + "loss": 0.3758, + "mean_token_accuracy": 0.8735917806625366, + "num_tokens": 81719491.0, + "step": 2100 + }, + { + "epoch": 0.26726879531866177, + "ewc_loss": 0.007232666015625, + "ewc_loss_diag": 7.68899917602539e-06, + "ewc_loss_parallel": 6.467103958129883e-06, + "grad_norm": 7.013457775115967, + "learning_rate": 8.902077151335311e-07, + "loss": 0.4028, + "mean_token_accuracy": 0.8641015887260437, + "num_tokens": 81760211.0, + "step": 2101 + }, + { + "epoch": 0.26739600559725224, + "ewc_loss": 0.007354736328125, + "ewc_loss_diag": 7.68899917602539e-06, + "ewc_loss_parallel": 6.586313247680664e-06, + "grad_norm": 7.096949577331543, + "learning_rate": 8.90631623569309e-07, + "loss": 0.4591, + "mean_token_accuracy": 0.8453649282455444, + "num_tokens": 81799283.0, + "step": 2102 + }, + { + "epoch": 0.26752321587584277, + "ewc_loss": 0.007293701171875, + "ewc_loss_diag": 7.68899917602539e-06, + "ewc_loss_parallel": 6.5267086029052734e-06, + "grad_norm": 7.107995986938477, + "learning_rate": 8.910555320050868e-07, + "loss": 0.3947, + "mean_token_accuracy": 0.8679260015487671, + "num_tokens": 81836510.0, + "step": 2103 + }, + { + "epoch": 0.2676504261544333, + "ewc_loss": 0.00738525390625, + "ewc_loss_diag": 7.68899917602539e-06, + "ewc_loss_parallel": 6.616115570068359e-06, + "grad_norm": 7.0565972328186035, + "learning_rate": 8.914794404408648e-07, + "loss": 0.4011, + "mean_token_accuracy": 0.8670862913131714, + "num_tokens": 81877304.0, + "step": 2104 + }, + { + "epoch": 0.26777763643302377, + "ewc_loss": 0.00726318359375, + "ewc_loss_diag": 7.68899917602539e-06, + "ewc_loss_parallel": 6.496906280517578e-06, + "grad_norm": 6.9290008544921875, + "learning_rate": 8.919033488766426e-07, + "loss": 0.3351, + "mean_token_accuracy": 0.8857419490814209, + "num_tokens": 81917737.0, + "step": 2105 + }, + { + "epoch": 0.2679048467116143, + "ewc_loss": 0.00732421875, + "ewc_loss_diag": 7.68899917602539e-06, + "ewc_loss_parallel": 6.556510925292969e-06, + "grad_norm": 7.160818099975586, + "learning_rate": 8.923272573124204e-07, + "loss": 0.4278, + "mean_token_accuracy": 0.8560689687728882, + "num_tokens": 81952250.0, + "step": 2106 + }, + { + "epoch": 0.2680320569902048, + "ewc_loss": 0.0074462890625, + "ewc_loss_diag": 7.68899917602539e-06, + "ewc_loss_parallel": 6.67572021484375e-06, + "grad_norm": 7.068387031555176, + "learning_rate": 8.927511657481983e-07, + "loss": 0.424, + "mean_token_accuracy": 0.857951283454895, + "num_tokens": 81996559.0, + "step": 2107 + }, + { + "epoch": 0.2681592672687953, + "ewc_loss": 0.00726318359375, + "ewc_loss_diag": 7.748603820800781e-06, + "ewc_loss_parallel": 6.496906280517578e-06, + "grad_norm": 7.0395894050598145, + "learning_rate": 8.931750741839762e-07, + "loss": 0.358, + "mean_token_accuracy": 0.8808403015136719, + "num_tokens": 82029021.0, + "step": 2108 + }, + { + "epoch": 0.2682864775473858, + "ewc_loss": 0.00732421875, + "ewc_loss_diag": 7.748603820800781e-06, + "ewc_loss_parallel": 6.556510925292969e-06, + "grad_norm": 7.075557231903076, + "learning_rate": 8.935989826197541e-07, + "loss": 0.3805, + "mean_token_accuracy": 0.874340295791626, + "num_tokens": 82067382.0, + "step": 2109 + }, + { + "epoch": 0.26841368782597635, + "ewc_loss": 0.007293701171875, + "ewc_loss_diag": 7.748603820800781e-06, + "ewc_loss_parallel": 6.5267086029052734e-06, + "grad_norm": 7.082600116729736, + "learning_rate": 8.94022891055532e-07, + "loss": 0.3659, + "mean_token_accuracy": 0.8804093599319458, + "num_tokens": 82104212.0, + "step": 2110 + }, + { + "epoch": 0.2685408981045668, + "ewc_loss": 0.007415771484375, + "ewc_loss_diag": 7.748603820800781e-06, + "ewc_loss_parallel": 6.645917892456055e-06, + "grad_norm": 7.037900924682617, + "learning_rate": 8.944467994913098e-07, + "loss": 0.3902, + "mean_token_accuracy": 0.8691239356994629, + "num_tokens": 82145537.0, + "step": 2111 + }, + { + "epoch": 0.26866810838315736, + "ewc_loss": 0.00738525390625, + "ewc_loss_diag": 7.748603820800781e-06, + "ewc_loss_parallel": 6.616115570068359e-06, + "grad_norm": 6.9746575355529785, + "learning_rate": 8.948707079270878e-07, + "loss": 0.337, + "mean_token_accuracy": 0.8855511546134949, + "num_tokens": 82184465.0, + "step": 2112 + }, + { + "epoch": 0.2687953186617479, + "ewc_loss": 0.007354736328125, + "ewc_loss_diag": 7.748603820800781e-06, + "ewc_loss_parallel": 6.586313247680664e-06, + "grad_norm": 7.1123881340026855, + "learning_rate": 8.952946163628656e-07, + "loss": 0.3839, + "mean_token_accuracy": 0.8700827360153198, + "num_tokens": 82217427.0, + "step": 2113 + }, + { + "epoch": 0.26892252894033836, + "ewc_loss": 0.007568359375, + "ewc_loss_diag": 7.748603820800781e-06, + "ewc_loss_parallel": 6.794929504394531e-06, + "grad_norm": 6.981068134307861, + "learning_rate": 8.957185247986434e-07, + "loss": 0.3413, + "mean_token_accuracy": 0.8838039636611938, + "num_tokens": 82254805.0, + "step": 2114 + }, + { + "epoch": 0.2690497392189289, + "ewc_loss": 0.007354736328125, + "ewc_loss_diag": 7.748603820800781e-06, + "ewc_loss_parallel": 6.586313247680664e-06, + "grad_norm": 7.063745498657227, + "learning_rate": 8.961424332344213e-07, + "loss": 0.3837, + "mean_token_accuracy": 0.8738715648651123, + "num_tokens": 82297658.0, + "step": 2115 + }, + { + "epoch": 0.2691769494975194, + "ewc_loss": 0.0074462890625, + "ewc_loss_diag": 7.748603820800781e-06, + "ewc_loss_parallel": 6.67572021484375e-06, + "grad_norm": 7.03511381149292, + "learning_rate": 8.965663416701992e-07, + "loss": 0.3893, + "mean_token_accuracy": 0.868587076663971, + "num_tokens": 82336834.0, + "step": 2116 + }, + { + "epoch": 0.2693041597761099, + "ewc_loss": 0.007415771484375, + "ewc_loss_diag": 7.748603820800781e-06, + "ewc_loss_parallel": 6.645917892456055e-06, + "grad_norm": 7.105025291442871, + "learning_rate": 8.969902501059771e-07, + "loss": 0.3671, + "mean_token_accuracy": 0.8784976601600647, + "num_tokens": 82374178.0, + "step": 2117 + }, + { + "epoch": 0.2694313700547004, + "ewc_loss": 0.007476806640625, + "ewc_loss_diag": 7.748603820800781e-06, + "ewc_loss_parallel": 6.705522537231445e-06, + "grad_norm": 6.9591498374938965, + "learning_rate": 8.97414158541755e-07, + "loss": 0.3271, + "mean_token_accuracy": 0.8861902952194214, + "num_tokens": 82412582.0, + "step": 2118 + }, + { + "epoch": 0.26955858033329094, + "ewc_loss": 0.007476806640625, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.67572021484375e-06, + "grad_norm": 7.169095516204834, + "learning_rate": 8.978380669775328e-07, + "loss": 0.4078, + "mean_token_accuracy": 0.8623920679092407, + "num_tokens": 82451649.0, + "step": 2119 + }, + { + "epoch": 0.2696857906118814, + "ewc_loss": 0.007537841796875, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.735324859619141e-06, + "grad_norm": 7.100421905517578, + "learning_rate": 8.982619754133107e-07, + "loss": 0.4113, + "mean_token_accuracy": 0.8628168106079102, + "num_tokens": 82489280.0, + "step": 2120 + }, + { + "epoch": 0.26981300089047194, + "ewc_loss": 0.007476806640625, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.67572021484375e-06, + "grad_norm": 7.019321441650391, + "learning_rate": 8.986858838490886e-07, + "loss": 0.4054, + "mean_token_accuracy": 0.8647245764732361, + "num_tokens": 82534057.0, + "step": 2121 + }, + { + "epoch": 0.26994021116906247, + "ewc_loss": 0.00738525390625, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.586313247680664e-06, + "grad_norm": 6.9338860511779785, + "learning_rate": 8.991097922848663e-07, + "loss": 0.3657, + "mean_token_accuracy": 0.8774978518486023, + "num_tokens": 82576441.0, + "step": 2122 + }, + { + "epoch": 0.27006742144765294, + "ewc_loss": 0.007354736328125, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.556510925292969e-06, + "grad_norm": 7.010534763336182, + "learning_rate": 8.995337007206443e-07, + "loss": 0.3879, + "mean_token_accuracy": 0.870905876159668, + "num_tokens": 82618009.0, + "step": 2123 + }, + { + "epoch": 0.2701946317262435, + "ewc_loss": 0.007476806640625, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.67572021484375e-06, + "grad_norm": 7.002044200897217, + "learning_rate": 8.999576091564221e-07, + "loss": 0.3722, + "mean_token_accuracy": 0.8780365586280823, + "num_tokens": 82654426.0, + "step": 2124 + }, + { + "epoch": 0.270321842004834, + "ewc_loss": 0.007476806640625, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.67572021484375e-06, + "grad_norm": 7.124518871307373, + "learning_rate": 9.003815175922001e-07, + "loss": 0.3857, + "mean_token_accuracy": 0.868434488773346, + "num_tokens": 82687673.0, + "step": 2125 + }, + { + "epoch": 0.2704490522834245, + "ewc_loss": 0.00750732421875, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.705522537231445e-06, + "grad_norm": 7.048102378845215, + "learning_rate": 9.008054260279779e-07, + "loss": 0.3823, + "mean_token_accuracy": 0.8713070154190063, + "num_tokens": 82730543.0, + "step": 2126 + }, + { + "epoch": 0.270576262562015, + "ewc_loss": 0.007415771484375, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.616115570068359e-06, + "grad_norm": 7.116008281707764, + "learning_rate": 9.012293344637558e-07, + "loss": 0.4019, + "mean_token_accuracy": 0.8679141998291016, + "num_tokens": 82768119.0, + "step": 2127 + }, + { + "epoch": 0.27070347284060553, + "ewc_loss": 0.007476806640625, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.67572021484375e-06, + "grad_norm": 7.067548751831055, + "learning_rate": 9.016532428995337e-07, + "loss": 0.3618, + "mean_token_accuracy": 0.8779735565185547, + "num_tokens": 82806559.0, + "step": 2128 + }, + { + "epoch": 0.27083068311919606, + "ewc_loss": 0.00750732421875, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.705522537231445e-06, + "grad_norm": 6.970828056335449, + "learning_rate": 9.020771513353115e-07, + "loss": 0.3968, + "mean_token_accuracy": 0.8662363290786743, + "num_tokens": 82849408.0, + "step": 2129 + }, + { + "epoch": 0.27095789339778653, + "ewc_loss": 0.007415771484375, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.616115570068359e-06, + "grad_norm": 7.113757133483887, + "learning_rate": 9.025010597710894e-07, + "loss": 0.4047, + "mean_token_accuracy": 0.8651547431945801, + "num_tokens": 82887795.0, + "step": 2130 + }, + { + "epoch": 0.27108510367637706, + "ewc_loss": 0.007537841796875, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.735324859619141e-06, + "grad_norm": 7.073990821838379, + "learning_rate": 9.029249682068673e-07, + "loss": 0.3778, + "mean_token_accuracy": 0.8745488524436951, + "num_tokens": 82925903.0, + "step": 2131 + }, + { + "epoch": 0.2712123139549676, + "ewc_loss": 0.007476806640625, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.67572021484375e-06, + "grad_norm": 7.001782417297363, + "learning_rate": 9.033488766426451e-07, + "loss": 0.3776, + "mean_token_accuracy": 0.8714873790740967, + "num_tokens": 82963940.0, + "step": 2132 + }, + { + "epoch": 0.27133952423355806, + "ewc_loss": 0.007415771484375, + "ewc_loss_diag": 7.808208465576172e-06, + "ewc_loss_parallel": 6.616115570068359e-06, + "grad_norm": 7.093019008636475, + "learning_rate": 9.037727850784231e-07, + "loss": 0.355, + "mean_token_accuracy": 0.879099428653717, + "num_tokens": 83000150.0, + "step": 2133 + }, + { + "epoch": 0.2714667345121486, + "ewc_loss": 0.0074462890625, + "ewc_loss_diag": 7.867813110351562e-06, + "ewc_loss_parallel": 6.645917892456055e-06, + "grad_norm": 7.192010402679443, + "learning_rate": 9.041966935142009e-07, + "loss": 0.383, + "mean_token_accuracy": 0.8709235191345215, + "num_tokens": 83040239.0, + "step": 2134 + }, + { + "epoch": 0.2715939447907391, + "ewc_loss": 0.007537841796875, + "ewc_loss_diag": 7.867813110351562e-06, + "ewc_loss_parallel": 6.735324859619141e-06, + "grad_norm": 7.0176801681518555, + "learning_rate": 9.046206019499788e-07, + "loss": 0.3576, + "mean_token_accuracy": 0.8805094361305237, + "num_tokens": 83081629.0, + "step": 2135 + }, + { + "epoch": 0.2717211550693296, + "ewc_loss": 0.007354736328125, + "ewc_loss_diag": 7.867813110351562e-06, + "ewc_loss_parallel": 6.556510925292969e-06, + "grad_norm": 7.091657638549805, + "learning_rate": 9.050445103857567e-07, + "loss": 0.4306, + "mean_token_accuracy": 0.8603963851928711, + "num_tokens": 83121865.0, + "step": 2136 + }, + { + "epoch": 0.2718483653479201, + "ewc_loss": 0.007476806640625, + "ewc_loss_diag": 7.867813110351562e-06, + "ewc_loss_parallel": 6.67572021484375e-06, + "grad_norm": 7.081111907958984, + "learning_rate": 9.054684188215344e-07, + "loss": 0.3336, + "mean_token_accuracy": 0.8853213787078857, + "num_tokens": 83156588.0, + "step": 2137 + }, + { + "epoch": 0.27197557562651065, + "ewc_loss": 0.007476806640625, + "ewc_loss_diag": 7.867813110351562e-06, + "ewc_loss_parallel": 6.67572021484375e-06, + "grad_norm": 7.194619655609131, + "learning_rate": 9.058923272573124e-07, + "loss": 0.4004, + "mean_token_accuracy": 0.8650175333023071, + "num_tokens": 83192670.0, + "step": 2138 + }, + { + "epoch": 0.2721027859051011, + "ewc_loss": 0.00750732421875, + "ewc_loss_diag": 7.867813110351562e-06, + "ewc_loss_parallel": 6.705522537231445e-06, + "grad_norm": 7.0174336433410645, + "learning_rate": 9.063162356930902e-07, + "loss": 0.3693, + "mean_token_accuracy": 0.8771669864654541, + "num_tokens": 83233266.0, + "step": 2139 + }, + { + "epoch": 0.27222999618369165, + "ewc_loss": 0.007415771484375, + "ewc_loss_diag": 7.867813110351562e-06, + "ewc_loss_parallel": 6.616115570068359e-06, + "grad_norm": 7.13344669342041, + "learning_rate": 9.067401441288681e-07, + "loss": 0.401, + "mean_token_accuracy": 0.867654025554657, + "num_tokens": 83270720.0, + "step": 2140 + }, + { + "epoch": 0.2723572064622822, + "ewc_loss": 0.007659912109375, + "ewc_loss_diag": 7.867813110351562e-06, + "ewc_loss_parallel": 6.854534149169922e-06, + "grad_norm": 7.123126029968262, + "learning_rate": 9.07164052564646e-07, + "loss": 0.3754, + "mean_token_accuracy": 0.8708333373069763, + "num_tokens": 83310303.0, + "step": 2141 + }, + { + "epoch": 0.27248441674087265, + "ewc_loss": 0.00750732421875, + "ewc_loss_diag": 7.867813110351562e-06, + "ewc_loss_parallel": 6.705522537231445e-06, + "grad_norm": 7.155616760253906, + "learning_rate": 9.075879610004239e-07, + "loss": 0.4085, + "mean_token_accuracy": 0.8612874746322632, + "num_tokens": 83347222.0, + "step": 2142 + }, + { + "epoch": 0.2726116270194632, + "ewc_loss": 0.007537841796875, + "ewc_loss_diag": 7.867813110351562e-06, + "ewc_loss_parallel": 6.735324859619141e-06, + "grad_norm": 7.0869574546813965, + "learning_rate": 9.080118694362017e-07, + "loss": 0.3992, + "mean_token_accuracy": 0.8634506464004517, + "num_tokens": 83390610.0, + "step": 2143 + }, + { + "epoch": 0.2727388372980537, + "ewc_loss": 0.007476806640625, + "ewc_loss_diag": 7.987022399902344e-06, + "ewc_loss_parallel": 6.67572021484375e-06, + "grad_norm": 7.199553489685059, + "learning_rate": 9.084357778719796e-07, + "loss": 0.4131, + "mean_token_accuracy": 0.8598467111587524, + "num_tokens": 83427175.0, + "step": 2144 + }, + { + "epoch": 0.2728660475766442, + "ewc_loss": 0.007598876953125, + "ewc_loss_diag": 7.987022399902344e-06, + "ewc_loss_parallel": 6.794929504394531e-06, + "grad_norm": 7.086575984954834, + "learning_rate": 9.088596863077574e-07, + "loss": 0.3856, + "mean_token_accuracy": 0.8737586736679077, + "num_tokens": 83464021.0, + "step": 2145 + }, + { + "epoch": 0.2729932578552347, + "ewc_loss": 0.007598876953125, + "ewc_loss_diag": 7.987022399902344e-06, + "ewc_loss_parallel": 6.794929504394531e-06, + "grad_norm": 7.114500045776367, + "learning_rate": 9.092835947435354e-07, + "loss": 0.3846, + "mean_token_accuracy": 0.8710569143295288, + "num_tokens": 83502386.0, + "step": 2146 + }, + { + "epoch": 0.27312046813382523, + "ewc_loss": 0.007659912109375, + "ewc_loss_diag": 7.987022399902344e-06, + "ewc_loss_parallel": 6.854534149169922e-06, + "grad_norm": 7.199038505554199, + "learning_rate": 9.097075031793132e-07, + "loss": 0.4262, + "mean_token_accuracy": 0.856854259967804, + "num_tokens": 83539893.0, + "step": 2147 + }, + { + "epoch": 0.2732476784124157, + "ewc_loss": 0.007659912109375, + "ewc_loss_diag": 7.987022399902344e-06, + "ewc_loss_parallel": 6.854534149169922e-06, + "grad_norm": 7.114885330200195, + "learning_rate": 9.101314116150911e-07, + "loss": 0.3532, + "mean_token_accuracy": 0.8837012052536011, + "num_tokens": 83577026.0, + "step": 2148 + }, + { + "epoch": 0.27337488869100623, + "ewc_loss": 0.00762939453125, + "ewc_loss_diag": 7.987022399902344e-06, + "ewc_loss_parallel": 6.8247318267822266e-06, + "grad_norm": 7.110232830047607, + "learning_rate": 9.10555320050869e-07, + "loss": 0.3657, + "mean_token_accuracy": 0.8781549334526062, + "num_tokens": 83614410.0, + "step": 2149 + }, + { + "epoch": 0.27350209896959676, + "ewc_loss": 0.007659912109375, + "ewc_loss_diag": 7.987022399902344e-06, + "ewc_loss_parallel": 6.854534149169922e-06, + "grad_norm": 7.033297538757324, + "learning_rate": 9.109792284866469e-07, + "loss": 0.3875, + "mean_token_accuracy": 0.8718622326850891, + "num_tokens": 83657209.0, + "step": 2150 + }, + { + "epoch": 0.27362930924818724, + "ewc_loss": 0.0076904296875, + "ewc_loss_diag": 7.987022399902344e-06, + "ewc_loss_parallel": 6.884336471557617e-06, + "grad_norm": 7.133808612823486, + "learning_rate": 9.114031369224247e-07, + "loss": 0.3907, + "mean_token_accuracy": 0.8692183494567871, + "num_tokens": 83696719.0, + "step": 2151 + }, + { + "epoch": 0.27375651952677776, + "ewc_loss": 0.007598876953125, + "ewc_loss_diag": 7.987022399902344e-06, + "ewc_loss_parallel": 6.794929504394531e-06, + "grad_norm": 7.160823345184326, + "learning_rate": 9.118270453582026e-07, + "loss": 0.443, + "mean_token_accuracy": 0.8563807606697083, + "num_tokens": 83736514.0, + "step": 2152 + }, + { + "epoch": 0.2738837298053683, + "ewc_loss": 0.007598876953125, + "ewc_loss_diag": 7.987022399902344e-06, + "ewc_loss_parallel": 6.794929504394531e-06, + "grad_norm": 7.05399751663208, + "learning_rate": 9.122509537939804e-07, + "loss": 0.3764, + "mean_token_accuracy": 0.8713032007217407, + "num_tokens": 83780010.0, + "step": 2153 + }, + { + "epoch": 0.27401094008395877, + "ewc_loss": 0.00762939453125, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.8247318267822266e-06, + "grad_norm": 7.18313455581665, + "learning_rate": 9.126748622297584e-07, + "loss": 0.4325, + "mean_token_accuracy": 0.8533236980438232, + "num_tokens": 83817152.0, + "step": 2154 + }, + { + "epoch": 0.2741381503625493, + "ewc_loss": 0.007598876953125, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.794929504394531e-06, + "grad_norm": 7.174239158630371, + "learning_rate": 9.130987706655362e-07, + "loss": 0.4128, + "mean_token_accuracy": 0.8636316061019897, + "num_tokens": 83855567.0, + "step": 2155 + }, + { + "epoch": 0.2742653606411398, + "ewc_loss": 0.007659912109375, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.854534149169922e-06, + "grad_norm": 7.135774612426758, + "learning_rate": 9.135226791013141e-07, + "loss": 0.433, + "mean_token_accuracy": 0.8614330291748047, + "num_tokens": 83895454.0, + "step": 2156 + }, + { + "epoch": 0.2743925709197303, + "ewc_loss": 0.007659912109375, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.854534149169922e-06, + "grad_norm": 7.298967361450195, + "learning_rate": 9.13946587537092e-07, + "loss": 0.3915, + "mean_token_accuracy": 0.8640426397323608, + "num_tokens": 83923803.0, + "step": 2157 + }, + { + "epoch": 0.2745197811983208, + "ewc_loss": 0.007781982421875, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.973743438720703e-06, + "grad_norm": 7.10221004486084, + "learning_rate": 9.143704959728699e-07, + "loss": 0.3985, + "mean_token_accuracy": 0.8655958771705627, + "num_tokens": 83960872.0, + "step": 2158 + }, + { + "epoch": 0.27464699147691135, + "ewc_loss": 0.007598876953125, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.794929504394531e-06, + "grad_norm": 7.2766499519348145, + "learning_rate": 9.147944044086476e-07, + "loss": 0.4094, + "mean_token_accuracy": 0.8615827560424805, + "num_tokens": 83993986.0, + "step": 2159 + }, + { + "epoch": 0.2747742017555018, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 7.033348083496094e-06, + "grad_norm": 7.096640586853027, + "learning_rate": 9.152183128444255e-07, + "loss": 0.421, + "mean_token_accuracy": 0.8647584915161133, + "num_tokens": 84035494.0, + "step": 2160 + }, + { + "epoch": 0.27490141203409235, + "ewc_loss": 0.0076904296875, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.884336471557617e-06, + "grad_norm": 7.100311756134033, + "learning_rate": 9.156422212802034e-07, + "loss": 0.4216, + "mean_token_accuracy": 0.8556040525436401, + "num_tokens": 84075481.0, + "step": 2161 + }, + { + "epoch": 0.2750286223126829, + "ewc_loss": 0.007720947265625, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.9141387939453125e-06, + "grad_norm": 7.182697296142578, + "learning_rate": 9.160661297159813e-07, + "loss": 0.3586, + "mean_token_accuracy": 0.8791064620018005, + "num_tokens": 84111106.0, + "step": 2162 + }, + { + "epoch": 0.27515583259127335, + "ewc_loss": 0.007781982421875, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 7.0035457611083984e-06, + "grad_norm": 7.087684631347656, + "learning_rate": 9.164900381517592e-07, + "loss": 0.3839, + "mean_token_accuracy": 0.8721009492874146, + "num_tokens": 84149990.0, + "step": 2163 + }, + { + "epoch": 0.2752830428698639, + "ewc_loss": 0.007659912109375, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.854534149169922e-06, + "grad_norm": 7.074515342712402, + "learning_rate": 9.16913946587537e-07, + "loss": 0.3769, + "mean_token_accuracy": 0.8716834783554077, + "num_tokens": 84193182.0, + "step": 2164 + }, + { + "epoch": 0.2754102531484544, + "ewc_loss": 0.007781982421875, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 7.0035457611083984e-06, + "grad_norm": 7.277979373931885, + "learning_rate": 9.17337855023315e-07, + "loss": 0.3993, + "mean_token_accuracy": 0.8655611276626587, + "num_tokens": 84229545.0, + "step": 2165 + }, + { + "epoch": 0.2755374634270449, + "ewc_loss": 0.00775146484375, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.943941116333008e-06, + "grad_norm": 7.189537048339844, + "learning_rate": 9.177617634590928e-07, + "loss": 0.3679, + "mean_token_accuracy": 0.8758453726768494, + "num_tokens": 84267638.0, + "step": 2166 + }, + { + "epoch": 0.2756646737056354, + "ewc_loss": 0.007720947265625, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.9141387939453125e-06, + "grad_norm": 7.072775363922119, + "learning_rate": 9.181856718948706e-07, + "loss": 0.3381, + "mean_token_accuracy": 0.8834407329559326, + "num_tokens": 84305004.0, + "step": 2167 + }, + { + "epoch": 0.27579188398422594, + "ewc_loss": 0.007598876953125, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.794929504394531e-06, + "grad_norm": 7.222319602966309, + "learning_rate": 9.186095803306485e-07, + "loss": 0.374, + "mean_token_accuracy": 0.8733553886413574, + "num_tokens": 84341469.0, + "step": 2168 + }, + { + "epoch": 0.2759190942628164, + "ewc_loss": 0.00775146484375, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.943941116333008e-06, + "grad_norm": 7.1809000968933105, + "learning_rate": 9.190334887664264e-07, + "loss": 0.4229, + "mean_token_accuracy": 0.860668420791626, + "num_tokens": 84384654.0, + "step": 2169 + }, + { + "epoch": 0.27604630454140694, + "ewc_loss": 0.007720947265625, + "ewc_loss_diag": 8.046627044677734e-06, + "ewc_loss_parallel": 6.9141387939453125e-06, + "grad_norm": 7.337833404541016, + "learning_rate": 9.194573972022043e-07, + "loss": 0.4713, + "mean_token_accuracy": 0.8501824140548706, + "num_tokens": 84420885.0, + "step": 2170 + }, + { + "epoch": 0.27617351481999747, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 7.033348083496094e-06, + "grad_norm": 7.1133317947387695, + "learning_rate": 9.198813056379822e-07, + "loss": 0.4415, + "mean_token_accuracy": 0.8525910377502441, + "num_tokens": 84465398.0, + "step": 2171 + }, + { + "epoch": 0.27630072509858794, + "ewc_loss": 0.00775146484375, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 6.9141387939453125e-06, + "grad_norm": 7.306756496429443, + "learning_rate": 9.2030521407376e-07, + "loss": 0.4161, + "mean_token_accuracy": 0.8620688915252686, + "num_tokens": 84500751.0, + "step": 2172 + }, + { + "epoch": 0.27642793537717847, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 7.033348083496094e-06, + "grad_norm": 7.0961151123046875, + "learning_rate": 9.20729122509538e-07, + "loss": 0.341, + "mean_token_accuracy": 0.8867446184158325, + "num_tokens": 84540688.0, + "step": 2173 + }, + { + "epoch": 0.276555145655769, + "ewc_loss": 0.00775146484375, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 6.943941116333008e-06, + "grad_norm": 7.162700653076172, + "learning_rate": 9.211530309453158e-07, + "loss": 0.3999, + "mean_token_accuracy": 0.8655584454536438, + "num_tokens": 84580002.0, + "step": 2174 + }, + { + "epoch": 0.27668235593435947, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 6.973743438720703e-06, + "grad_norm": 7.268883228302002, + "learning_rate": 9.215769393810936e-07, + "loss": 0.3948, + "mean_token_accuracy": 0.8674705028533936, + "num_tokens": 84617009.0, + "step": 2175 + }, + { + "epoch": 0.27680956621295, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 6.973743438720703e-06, + "grad_norm": 7.275196552276611, + "learning_rate": 9.220008478168715e-07, + "loss": 0.4853, + "mean_token_accuracy": 0.8401732444763184, + "num_tokens": 84657941.0, + "step": 2176 + }, + { + "epoch": 0.2769367764915405, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 7.0035457611083984e-06, + "grad_norm": 7.108083724975586, + "learning_rate": 9.224247562526494e-07, + "loss": 0.3964, + "mean_token_accuracy": 0.8655259609222412, + "num_tokens": 84699363.0, + "step": 2177 + }, + { + "epoch": 0.277063986770131, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 6.973743438720703e-06, + "grad_norm": 7.13894510269165, + "learning_rate": 9.228486646884273e-07, + "loss": 0.3833, + "mean_token_accuracy": 0.8699549436569214, + "num_tokens": 84742203.0, + "step": 2178 + }, + { + "epoch": 0.2771911970487215, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 6.973743438720703e-06, + "grad_norm": 7.168878078460693, + "learning_rate": 9.232725731242052e-07, + "loss": 0.4364, + "mean_token_accuracy": 0.8521541357040405, + "num_tokens": 84783021.0, + "step": 2179 + }, + { + "epoch": 0.27731840732731206, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 7.033348083496094e-06, + "grad_norm": 7.195292949676514, + "learning_rate": 9.23696481559983e-07, + "loss": 0.42, + "mean_token_accuracy": 0.8585022687911987, + "num_tokens": 84821297.0, + "step": 2180 + }, + { + "epoch": 0.2774456176059026, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 7.0035457611083984e-06, + "grad_norm": 7.098839282989502, + "learning_rate": 9.24120389995761e-07, + "loss": 0.3696, + "mean_token_accuracy": 0.8772865533828735, + "num_tokens": 84863429.0, + "step": 2181 + }, + { + "epoch": 0.27757282788449306, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 6.973743438720703e-06, + "grad_norm": 7.217189788818359, + "learning_rate": 9.245442984315387e-07, + "loss": 0.3735, + "mean_token_accuracy": 0.8718879222869873, + "num_tokens": 84900107.0, + "step": 2182 + }, + { + "epoch": 0.2777000381630836, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 7.033348083496094e-06, + "grad_norm": 7.200204849243164, + "learning_rate": 9.249682068673165e-07, + "loss": 0.4165, + "mean_token_accuracy": 0.8604515790939331, + "num_tokens": 84939017.0, + "step": 2183 + }, + { + "epoch": 0.2778272484416741, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 7.033348083496094e-06, + "grad_norm": 7.1896867752075195, + "learning_rate": 9.253921153030945e-07, + "loss": 0.426, + "mean_token_accuracy": 0.8576807379722595, + "num_tokens": 84981245.0, + "step": 2184 + }, + { + "epoch": 0.2779544587202646, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 7.0035457611083984e-06, + "grad_norm": 7.149019718170166, + "learning_rate": 9.258160237388723e-07, + "loss": 0.3707, + "mean_token_accuracy": 0.877909243106842, + "num_tokens": 85017640.0, + "step": 2185 + }, + { + "epoch": 0.2780816689988551, + "ewc_loss": 0.00775146484375, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 6.9141387939453125e-06, + "grad_norm": 7.200776100158691, + "learning_rate": 9.262399321746503e-07, + "loss": 0.3771, + "mean_token_accuracy": 0.8727139830589294, + "num_tokens": 85052786.0, + "step": 2186 + }, + { + "epoch": 0.27820887927744564, + "ewc_loss": 0.00775146484375, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 6.943941116333008e-06, + "grad_norm": 7.266270637512207, + "learning_rate": 9.266638406104281e-07, + "loss": 0.4427, + "mean_token_accuracy": 0.8518072366714478, + "num_tokens": 85092067.0, + "step": 2187 + }, + { + "epoch": 0.2783360895560361, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 7.033348083496094e-06, + "grad_norm": 7.336269378662109, + "learning_rate": 9.27087749046206e-07, + "loss": 0.4137, + "mean_token_accuracy": 0.8628650903701782, + "num_tokens": 85123420.0, + "step": 2188 + }, + { + "epoch": 0.27846329983462664, + "ewc_loss": 0.00787353515625, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.206249237060547, + "learning_rate": 9.275116574819839e-07, + "loss": 0.3937, + "mean_token_accuracy": 0.8667415976524353, + "num_tokens": 85160190.0, + "step": 2189 + }, + { + "epoch": 0.27859051011321717, + "ewc_loss": 0.00787353515625, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 7.063150405883789e-06, + "grad_norm": 7.233577728271484, + "learning_rate": 9.279355659177617e-07, + "loss": 0.3966, + "mean_token_accuracy": 0.8696531653404236, + "num_tokens": 85195615.0, + "step": 2190 + }, + { + "epoch": 0.27871772039180764, + "ewc_loss": 0.00787353515625, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 7.063150405883789e-06, + "grad_norm": 7.278502941131592, + "learning_rate": 9.283594743535395e-07, + "loss": 0.4336, + "mean_token_accuracy": 0.8580841422080994, + "num_tokens": 85229870.0, + "step": 2191 + }, + { + "epoch": 0.2788449306703982, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.106231689453125e-06, + "ewc_loss_parallel": 7.033348083496094e-06, + "grad_norm": 7.185497283935547, + "learning_rate": 9.287833827893175e-07, + "loss": 0.3997, + "mean_token_accuracy": 0.865141749382019, + "num_tokens": 85269832.0, + "step": 2192 + }, + { + "epoch": 0.2789721409489887, + "ewc_loss": 0.00787353515625, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.063150405883789e-06, + "grad_norm": 7.2934184074401855, + "learning_rate": 9.292072912250953e-07, + "loss": 0.4083, + "mean_token_accuracy": 0.8618708848953247, + "num_tokens": 85303025.0, + "step": 2193 + }, + { + "epoch": 0.2790993512275792, + "ewc_loss": 0.00787353515625, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.0341644287109375, + "learning_rate": 9.296311996608733e-07, + "loss": 0.3425, + "mean_token_accuracy": 0.8838375806808472, + "num_tokens": 85343056.0, + "step": 2194 + }, + { + "epoch": 0.2792265615061697, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.033348083496094e-06, + "grad_norm": 7.19369649887085, + "learning_rate": 9.300551080966511e-07, + "loss": 0.446, + "mean_token_accuracy": 0.8484153747558594, + "num_tokens": 85384629.0, + "step": 2195 + }, + { + "epoch": 0.27935377178476023, + "ewc_loss": 0.00787353515625, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.28695821762085, + "learning_rate": 9.30479016532429e-07, + "loss": 0.3846, + "mean_token_accuracy": 0.8665584325790405, + "num_tokens": 85418755.0, + "step": 2196 + }, + { + "epoch": 0.2794809820633507, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.12275505065918e-06, + "grad_norm": 7.156136512756348, + "learning_rate": 9.309029249682068e-07, + "loss": 0.3855, + "mean_token_accuracy": 0.8707978129386902, + "num_tokens": 85458113.0, + "step": 2197 + }, + { + "epoch": 0.27960819234194123, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.0035457611083984e-06, + "grad_norm": 7.171924591064453, + "learning_rate": 9.313268334039847e-07, + "loss": 0.3964, + "mean_token_accuracy": 0.8633602261543274, + "num_tokens": 85496848.0, + "step": 2198 + }, + { + "epoch": 0.27973540262053176, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.033348083496094e-06, + "grad_norm": 7.259672164916992, + "learning_rate": 9.317507418397625e-07, + "loss": 0.3349, + "mean_token_accuracy": 0.8880071640014648, + "num_tokens": 85535917.0, + "step": 2199 + }, + { + "epoch": 0.27986261289912223, + "ewc_loss": 0.00787353515625, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.154356479644775, + "learning_rate": 9.321746502755404e-07, + "loss": 0.4363, + "mean_token_accuracy": 0.8570516705513, + "num_tokens": 85576900.0, + "step": 2200 + }, + { + "epoch": 0.27998982317771276, + "ewc_loss": 0.0078125, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.033348083496094e-06, + "grad_norm": 7.262918949127197, + "learning_rate": 9.325985587113183e-07, + "loss": 0.3956, + "mean_token_accuracy": 0.8678098320960999, + "num_tokens": 85611544.0, + "step": 2201 + }, + { + "epoch": 0.2801170334563033, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.12275505065918e-06, + "grad_norm": 7.302351474761963, + "learning_rate": 9.330224671470962e-07, + "loss": 0.4139, + "mean_token_accuracy": 0.8662232160568237, + "num_tokens": 85645351.0, + "step": 2202 + }, + { + "epoch": 0.28024424373489376, + "ewc_loss": 0.00787353515625, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.31718111038208, + "learning_rate": 9.334463755828741e-07, + "loss": 0.4259, + "mean_token_accuracy": 0.8543539047241211, + "num_tokens": 85676832.0, + "step": 2203 + }, + { + "epoch": 0.2803714540134843, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.12275505065918e-06, + "grad_norm": 7.197866439819336, + "learning_rate": 9.338702840186519e-07, + "loss": 0.4319, + "mean_token_accuracy": 0.8540631532669067, + "num_tokens": 85716168.0, + "step": 2204 + }, + { + "epoch": 0.2804986642920748, + "ewc_loss": 0.00787353515625, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.063150405883789e-06, + "grad_norm": 7.189391613006592, + "learning_rate": 9.342941924544298e-07, + "loss": 0.4108, + "mean_token_accuracy": 0.8616660833358765, + "num_tokens": 85756370.0, + "step": 2205 + }, + { + "epoch": 0.2806258745706653, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.12275505065918e-06, + "grad_norm": 7.207852840423584, + "learning_rate": 9.347181008902076e-07, + "loss": 0.3565, + "mean_token_accuracy": 0.8832454681396484, + "num_tokens": 85793086.0, + "step": 2206 + }, + { + "epoch": 0.2807530848492558, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.3206892013549805, + "learning_rate": 9.351420093259855e-07, + "loss": 0.3656, + "mean_token_accuracy": 0.8773383498191833, + "num_tokens": 85826545.0, + "step": 2207 + }, + { + "epoch": 0.28088029512784635, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.2601847648620605, + "learning_rate": 9.355659177617634e-07, + "loss": 0.3517, + "mean_token_accuracy": 0.8775844573974609, + "num_tokens": 85858091.0, + "step": 2208 + }, + { + "epoch": 0.2810075054064368, + "ewc_loss": 0.00787353515625, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.112934112548828, + "learning_rate": 9.359898261975413e-07, + "loss": 0.341, + "mean_token_accuracy": 0.8832402229309082, + "num_tokens": 85900383.0, + "step": 2209 + }, + { + "epoch": 0.28113471568502735, + "ewc_loss": 0.00787353515625, + "ewc_loss_diag": 8.165836334228516e-06, + "ewc_loss_parallel": 7.063150405883789e-06, + "grad_norm": 7.30222225189209, + "learning_rate": 9.364137346333192e-07, + "loss": 0.456, + "mean_token_accuracy": 0.850481390953064, + "num_tokens": 85940800.0, + "step": 2210 + }, + { + "epoch": 0.2812619259636179, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.225440979003906e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.284467697143555, + "learning_rate": 9.368376430690971e-07, + "loss": 0.3695, + "mean_token_accuracy": 0.8745250105857849, + "num_tokens": 85974210.0, + "step": 2211 + }, + { + "epoch": 0.28138913624220835, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.225440979003906e-06, + "ewc_loss_parallel": 7.12275505065918e-06, + "grad_norm": 7.302996635437012, + "learning_rate": 9.372615515048749e-07, + "loss": 0.4122, + "mean_token_accuracy": 0.8627787232398987, + "num_tokens": 86009646.0, + "step": 2212 + }, + { + "epoch": 0.2815163465207989, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.225440979003906e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.184664249420166, + "learning_rate": 9.376854599406528e-07, + "loss": 0.4211, + "mean_token_accuracy": 0.8610948324203491, + "num_tokens": 86054828.0, + "step": 2213 + }, + { + "epoch": 0.2816435567993894, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.225440979003906e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.361013889312744, + "learning_rate": 9.381093683764306e-07, + "loss": 0.4047, + "mean_token_accuracy": 0.8663246035575867, + "num_tokens": 86091115.0, + "step": 2214 + }, + { + "epoch": 0.2817707670779799, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.225440979003906e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.123264312744141, + "learning_rate": 9.385332768122085e-07, + "loss": 0.3839, + "mean_token_accuracy": 0.869566798210144, + "num_tokens": 86132417.0, + "step": 2215 + }, + { + "epoch": 0.2818979773565704, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.225440979003906e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.193667888641357, + "learning_rate": 9.389571852479864e-07, + "loss": 0.3842, + "mean_token_accuracy": 0.872151255607605, + "num_tokens": 86174260.0, + "step": 2216 + }, + { + "epoch": 0.28202518763516093, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.225440979003906e-06, + "ewc_loss_parallel": 7.12275505065918e-06, + "grad_norm": 7.2522454261779785, + "learning_rate": 9.393810936837643e-07, + "loss": 0.3821, + "mean_token_accuracy": 0.8713043928146362, + "num_tokens": 86214673.0, + "step": 2217 + }, + { + "epoch": 0.2821523979137514, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.225440979003906e-06, + "ewc_loss_parallel": 7.12275505065918e-06, + "grad_norm": 7.156287670135498, + "learning_rate": 9.398050021195422e-07, + "loss": 0.4183, + "mean_token_accuracy": 0.8628495931625366, + "num_tokens": 86256550.0, + "step": 2218 + }, + { + "epoch": 0.28227960819234194, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.225440979003906e-06, + "ewc_loss_parallel": 7.12275505065918e-06, + "grad_norm": 7.1658034324646, + "learning_rate": 9.402289105553201e-07, + "loss": 0.3722, + "mean_token_accuracy": 0.8741382956504822, + "num_tokens": 86300101.0, + "step": 2219 + }, + { + "epoch": 0.28240681847093246, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.238125324249268, + "learning_rate": 9.406528189910978e-07, + "loss": 0.3767, + "mean_token_accuracy": 0.874977707862854, + "num_tokens": 86337626.0, + "step": 2220 + }, + { + "epoch": 0.28253402874952294, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.212854385375977, + "learning_rate": 9.410767274268757e-07, + "loss": 0.362, + "mean_token_accuracy": 0.8757625222206116, + "num_tokens": 86377728.0, + "step": 2221 + }, + { + "epoch": 0.28266123902811346, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.293026447296143, + "learning_rate": 9.415006358626536e-07, + "loss": 0.4502, + "mean_token_accuracy": 0.8492158651351929, + "num_tokens": 86418372.0, + "step": 2222 + }, + { + "epoch": 0.282788449306704, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.387664794921875, + "learning_rate": 9.419245442984314e-07, + "loss": 0.4386, + "mean_token_accuracy": 0.8529976010322571, + "num_tokens": 86454228.0, + "step": 2223 + }, + { + "epoch": 0.28291565958529447, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.327908515930176, + "learning_rate": 9.423484527342094e-07, + "loss": 0.3956, + "mean_token_accuracy": 0.8656541705131531, + "num_tokens": 86489264.0, + "step": 2224 + }, + { + "epoch": 0.283042869863885, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.301905155181885, + "learning_rate": 9.427723611699872e-07, + "loss": 0.4031, + "mean_token_accuracy": 0.8657931089401245, + "num_tokens": 86524130.0, + "step": 2225 + }, + { + "epoch": 0.2831700801424755, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.248824119567871, + "learning_rate": 9.431962696057652e-07, + "loss": 0.4025, + "mean_token_accuracy": 0.8653252124786377, + "num_tokens": 86568254.0, + "step": 2226 + }, + { + "epoch": 0.283297290421066, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.2674560546875, + "learning_rate": 9.43620178041543e-07, + "loss": 0.3697, + "mean_token_accuracy": 0.8771240711212158, + "num_tokens": 86605459.0, + "step": 2227 + }, + { + "epoch": 0.2834245006996565, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.2090606689453125, + "learning_rate": 9.440440864773208e-07, + "loss": 0.3987, + "mean_token_accuracy": 0.8651308417320251, + "num_tokens": 86642948.0, + "step": 2228 + }, + { + "epoch": 0.28355171097824705, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.248514175415039, + "learning_rate": 9.444679949130987e-07, + "loss": 0.4015, + "mean_token_accuracy": 0.8625521659851074, + "num_tokens": 86685290.0, + "step": 2229 + }, + { + "epoch": 0.2836789212568376, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.379108428955078, + "learning_rate": 9.448919033488766e-07, + "loss": 0.4298, + "mean_token_accuracy": 0.857889711856842, + "num_tokens": 86722276.0, + "step": 2230 + }, + { + "epoch": 0.28380613153542805, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.231592655181885, + "learning_rate": 9.453158117846544e-07, + "loss": 0.4156, + "mean_token_accuracy": 0.8633605241775513, + "num_tokens": 86761038.0, + "step": 2231 + }, + { + "epoch": 0.2839333418140186, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.255916595458984, + "learning_rate": 9.457397202204324e-07, + "loss": 0.4419, + "mean_token_accuracy": 0.8538174629211426, + "num_tokens": 86803785.0, + "step": 2232 + }, + { + "epoch": 0.2840605520926091, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.304425239562988, + "learning_rate": 9.461636286562102e-07, + "loss": 0.3563, + "mean_token_accuracy": 0.8812113404273987, + "num_tokens": 86840435.0, + "step": 2233 + }, + { + "epoch": 0.2841877623711996, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.12275505065918e-06, + "grad_norm": 7.3400444984436035, + "learning_rate": 9.465875370919882e-07, + "loss": 0.4289, + "mean_token_accuracy": 0.8597321510314941, + "num_tokens": 86879027.0, + "step": 2234 + }, + { + "epoch": 0.2843149726497901, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.212162017822266e-06, + "grad_norm": 7.334221363067627, + "learning_rate": 9.470114455277659e-07, + "loss": 0.4212, + "mean_token_accuracy": 0.8582116961479187, + "num_tokens": 86915114.0, + "step": 2235 + }, + { + "epoch": 0.28444218292838064, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.12275505065918e-06, + "grad_norm": 7.1961565017700195, + "learning_rate": 9.474353539635438e-07, + "loss": 0.3789, + "mean_token_accuracy": 0.8713911771774292, + "num_tokens": 86958246.0, + "step": 2236 + }, + { + "epoch": 0.2845693932069711, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.333535194396973, + "learning_rate": 9.478592623993217e-07, + "loss": 0.3908, + "mean_token_accuracy": 0.8674294948577881, + "num_tokens": 86991426.0, + "step": 2237 + }, + { + "epoch": 0.28469660348556164, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.254756450653076, + "learning_rate": 9.482831708350996e-07, + "loss": 0.3789, + "mean_token_accuracy": 0.8740077614784241, + "num_tokens": 87030989.0, + "step": 2238 + }, + { + "epoch": 0.28482381376415217, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.253384113311768, + "learning_rate": 9.487070792708775e-07, + "loss": 0.4308, + "mean_token_accuracy": 0.8591904640197754, + "num_tokens": 87070846.0, + "step": 2239 + }, + { + "epoch": 0.28495102404274264, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.227680206298828, + "learning_rate": 9.491309877066554e-07, + "loss": 0.4039, + "mean_token_accuracy": 0.8649631142616272, + "num_tokens": 87115199.0, + "step": 2240 + }, + { + "epoch": 0.28507823432133317, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.30406379699707, + "learning_rate": 9.495548961424332e-07, + "loss": 0.4237, + "mean_token_accuracy": 0.8585636615753174, + "num_tokens": 87158823.0, + "step": 2241 + }, + { + "epoch": 0.2852054445999237, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.285045623779297e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.331402778625488, + "learning_rate": 9.499788045782111e-07, + "loss": 0.391, + "mean_token_accuracy": 0.868955135345459, + "num_tokens": 87196389.0, + "step": 2242 + }, + { + "epoch": 0.28533265487851417, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.344650268554688e-06, + "ewc_loss_parallel": 7.212162017822266e-06, + "grad_norm": 7.312140941619873, + "learning_rate": 9.504027130139889e-07, + "loss": 0.4076, + "mean_token_accuracy": 0.8634090423583984, + "num_tokens": 87239335.0, + "step": 2243 + }, + { + "epoch": 0.2854598651571047, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.344650268554688e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.325005531311035, + "learning_rate": 9.508266214497667e-07, + "loss": 0.3679, + "mean_token_accuracy": 0.8779312372207642, + "num_tokens": 87276401.0, + "step": 2244 + }, + { + "epoch": 0.2855870754356952, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.344650268554688e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.21989631652832, + "learning_rate": 9.512505298855447e-07, + "loss": 0.3855, + "mean_token_accuracy": 0.872244656085968, + "num_tokens": 87320945.0, + "step": 2245 + }, + { + "epoch": 0.2857142857142857, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.344650268554688e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.308084964752197, + "learning_rate": 9.516744383213225e-07, + "loss": 0.3505, + "mean_token_accuracy": 0.8794517517089844, + "num_tokens": 87357834.0, + "step": 2246 + }, + { + "epoch": 0.2858414959928762, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.344650268554688e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.269008636474609, + "learning_rate": 9.520983467571005e-07, + "loss": 0.4102, + "mean_token_accuracy": 0.8627452850341797, + "num_tokens": 87398587.0, + "step": 2247 + }, + { + "epoch": 0.28596870627146675, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.344650268554688e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.420757293701172, + "learning_rate": 9.525222551928783e-07, + "loss": 0.4461, + "mean_token_accuracy": 0.8589967489242554, + "num_tokens": 87438517.0, + "step": 2248 + }, + { + "epoch": 0.2860959165500572, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.344650268554688e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.374533653259277, + "learning_rate": 9.529461636286562e-07, + "loss": 0.3624, + "mean_token_accuracy": 0.8779188394546509, + "num_tokens": 87470296.0, + "step": 2249 + }, + { + "epoch": 0.28622312682864776, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.344650268554688e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.312405586242676, + "learning_rate": 9.533700720644341e-07, + "loss": 0.3528, + "mean_token_accuracy": 0.8818312287330627, + "num_tokens": 87507046.0, + "step": 2250 + }, + { + "epoch": 0.2863503371072383, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.344650268554688e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.336441993713379, + "learning_rate": 9.537939805002118e-07, + "loss": 0.3968, + "mean_token_accuracy": 0.8676362037658691, + "num_tokens": 87550505.0, + "step": 2251 + }, + { + "epoch": 0.28647754738582876, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.335220813751221, + "learning_rate": 9.542178889359898e-07, + "loss": 0.3791, + "mean_token_accuracy": 0.873744010925293, + "num_tokens": 87588560.0, + "step": 2252 + }, + { + "epoch": 0.2866047576644193, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.12275505065918e-06, + "grad_norm": 7.299487113952637, + "learning_rate": 9.546417973717677e-07, + "loss": 0.4032, + "mean_token_accuracy": 0.8664001226425171, + "num_tokens": 87628862.0, + "step": 2253 + }, + { + "epoch": 0.2867319679430098, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.245275020599365, + "learning_rate": 9.550657058075455e-07, + "loss": 0.4125, + "mean_token_accuracy": 0.8635746240615845, + "num_tokens": 87673852.0, + "step": 2254 + }, + { + "epoch": 0.2868591782216003, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.092952728271484e-06, + "grad_norm": 7.314792156219482, + "learning_rate": 9.554896142433234e-07, + "loss": 0.4249, + "mean_token_accuracy": 0.8564401865005493, + "num_tokens": 87712206.0, + "step": 2255 + }, + { + "epoch": 0.2869863885001908, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.356495380401611, + "learning_rate": 9.559135226791012e-07, + "loss": 0.4355, + "mean_token_accuracy": 0.8511729836463928, + "num_tokens": 87751245.0, + "step": 2256 + }, + { + "epoch": 0.28711359877878134, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.191709041595459, + "learning_rate": 9.563374311148793e-07, + "loss": 0.4247, + "mean_token_accuracy": 0.8575533628463745, + "num_tokens": 87800624.0, + "step": 2257 + }, + { + "epoch": 0.2872408090573718, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.212162017822266e-06, + "grad_norm": 7.423187255859375, + "learning_rate": 9.56761339550657e-07, + "loss": 0.3767, + "mean_token_accuracy": 0.8739029765129089, + "num_tokens": 87832271.0, + "step": 2258 + }, + { + "epoch": 0.28736801933596234, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.212162017822266e-06, + "grad_norm": 7.24691915512085, + "learning_rate": 9.57185247986435e-07, + "loss": 0.3627, + "mean_token_accuracy": 0.8775566220283508, + "num_tokens": 87871031.0, + "step": 2259 + }, + { + "epoch": 0.28749522961455287, + "ewc_loss": 0.0079345703125, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.12275505065918e-06, + "grad_norm": 7.316500663757324, + "learning_rate": 9.576091564222128e-07, + "loss": 0.3414, + "mean_token_accuracy": 0.8827610611915588, + "num_tokens": 87908814.0, + "step": 2260 + }, + { + "epoch": 0.28762243989314334, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.334413051605225, + "learning_rate": 9.580330648579906e-07, + "loss": 0.4064, + "mean_token_accuracy": 0.8616983890533447, + "num_tokens": 87947076.0, + "step": 2261 + }, + { + "epoch": 0.2877496501717339, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.314282417297363, + "learning_rate": 9.584569732937685e-07, + "loss": 0.4083, + "mean_token_accuracy": 0.8635187149047852, + "num_tokens": 87987447.0, + "step": 2262 + }, + { + "epoch": 0.2878768604503244, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.212162017822266e-06, + "grad_norm": 7.377484321594238, + "learning_rate": 9.588808817295463e-07, + "loss": 0.3958, + "mean_token_accuracy": 0.8635646104812622, + "num_tokens": 88021793.0, + "step": 2263 + }, + { + "epoch": 0.2880040707289149, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.346151828765869, + "learning_rate": 9.593047901653242e-07, + "loss": 0.3913, + "mean_token_accuracy": 0.8664106726646423, + "num_tokens": 88056873.0, + "step": 2264 + }, + { + "epoch": 0.2881312810075054, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.267880439758301, + "learning_rate": 9.597286986011022e-07, + "loss": 0.3559, + "mean_token_accuracy": 0.8796122670173645, + "num_tokens": 88093868.0, + "step": 2265 + }, + { + "epoch": 0.28825849128609593, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.390333652496338, + "learning_rate": 9.601526070368799e-07, + "loss": 0.4311, + "mean_token_accuracy": 0.858174741268158, + "num_tokens": 88131358.0, + "step": 2266 + }, + { + "epoch": 0.2883857015646864, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.212162017822266e-06, + "grad_norm": 7.355099678039551, + "learning_rate": 9.60576515472658e-07, + "loss": 0.3917, + "mean_token_accuracy": 0.8681386113166809, + "num_tokens": 88168137.0, + "step": 2267 + }, + { + "epoch": 0.28851291184327693, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.306422710418701, + "learning_rate": 9.610004239084358e-07, + "loss": 0.3961, + "mean_token_accuracy": 0.8667972683906555, + "num_tokens": 88209897.0, + "step": 2268 + }, + { + "epoch": 0.28864012212186746, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.423868179321289, + "learning_rate": 9.614243323442136e-07, + "loss": 0.4526, + "mean_token_accuracy": 0.8509752154350281, + "num_tokens": 88248034.0, + "step": 2269 + }, + { + "epoch": 0.28876733240045793, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.350575923919678, + "learning_rate": 9.618482407799915e-07, + "loss": 0.377, + "mean_token_accuracy": 0.8708261847496033, + "num_tokens": 88284542.0, + "step": 2270 + }, + { + "epoch": 0.28889454267904846, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.212162017822266e-06, + "grad_norm": 7.280081272125244, + "learning_rate": 9.622721492157693e-07, + "loss": 0.3464, + "mean_token_accuracy": 0.8824067711830139, + "num_tokens": 88323664.0, + "step": 2271 + }, + { + "epoch": 0.289021752957639, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.4354143142700195, + "learning_rate": 9.626960576515472e-07, + "loss": 0.3876, + "mean_token_accuracy": 0.869256317615509, + "num_tokens": 88355467.0, + "step": 2272 + }, + { + "epoch": 0.28914896323622946, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.212162017822266e-06, + "grad_norm": 7.291471004486084, + "learning_rate": 9.63119966087325e-07, + "loss": 0.3716, + "mean_token_accuracy": 0.874352216720581, + "num_tokens": 88394375.0, + "step": 2273 + }, + { + "epoch": 0.28927617351482, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.152557373046875e-06, + "grad_norm": 7.247459888458252, + "learning_rate": 9.635438745231029e-07, + "loss": 0.3768, + "mean_token_accuracy": 0.8729293346405029, + "num_tokens": 88435377.0, + "step": 2274 + }, + { + "epoch": 0.2894033837934105, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.271766662597656e-06, + "grad_norm": 7.403828144073486, + "learning_rate": 9.63967782958881e-07, + "loss": 0.4057, + "mean_token_accuracy": 0.8619694709777832, + "num_tokens": 88474205.0, + "step": 2275 + }, + { + "epoch": 0.289530594072001, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.353227615356445, + "learning_rate": 9.643916913946588e-07, + "loss": 0.3762, + "mean_token_accuracy": 0.874026894569397, + "num_tokens": 88509955.0, + "step": 2276 + }, + { + "epoch": 0.2896578043505915, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.212162017822266e-06, + "grad_norm": 7.309587001800537, + "learning_rate": 9.648155998304366e-07, + "loss": 0.3455, + "mean_token_accuracy": 0.8827158212661743, + "num_tokens": 88549575.0, + "step": 2277 + }, + { + "epoch": 0.28978501462918205, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.404254913330078e-06, + "ewc_loss_parallel": 7.271766662597656e-06, + "grad_norm": 7.290575981140137, + "learning_rate": 9.652395082662145e-07, + "loss": 0.4378, + "mean_token_accuracy": 0.8529007434844971, + "num_tokens": 88593883.0, + "step": 2278 + }, + { + "epoch": 0.2899122249077726, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.463859558105469e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.461091995239258, + "learning_rate": 9.656634167019923e-07, + "loss": 0.4046, + "mean_token_accuracy": 0.8636393547058105, + "num_tokens": 88631702.0, + "step": 2279 + }, + { + "epoch": 0.29003943518636305, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.463859558105469e-06, + "ewc_loss_parallel": 7.271766662597656e-06, + "grad_norm": 7.302731990814209, + "learning_rate": 9.660873251377701e-07, + "loss": 0.4103, + "mean_token_accuracy": 0.8609197735786438, + "num_tokens": 88671583.0, + "step": 2280 + }, + { + "epoch": 0.2901666454649536, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.463859558105469e-06, + "ewc_loss_parallel": 7.271766662597656e-06, + "grad_norm": 7.444035053253174, + "learning_rate": 9.66511233573548e-07, + "loss": 0.4349, + "mean_token_accuracy": 0.8598601818084717, + "num_tokens": 88708241.0, + "step": 2281 + }, + { + "epoch": 0.2902938557435441, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.463859558105469e-06, + "ewc_loss_parallel": 7.271766662597656e-06, + "grad_norm": 7.37495231628418, + "learning_rate": 9.669351420093258e-07, + "loss": 0.3676, + "mean_token_accuracy": 0.8783340454101562, + "num_tokens": 88742947.0, + "step": 2282 + }, + { + "epoch": 0.2904210660221346, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.463859558105469e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.436962604522705, + "learning_rate": 9.67359050445104e-07, + "loss": 0.3905, + "mean_token_accuracy": 0.8677840828895569, + "num_tokens": 88775899.0, + "step": 2283 + }, + { + "epoch": 0.2905482763007251, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.463859558105469e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.392199993133545, + "learning_rate": 9.677829588808817e-07, + "loss": 0.3882, + "mean_token_accuracy": 0.8661458492279053, + "num_tokens": 88808360.0, + "step": 2284 + }, + { + "epoch": 0.29067548657931563, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.463859558105469e-06, + "ewc_loss_parallel": 7.212162017822266e-06, + "grad_norm": 7.4305291175842285, + "learning_rate": 9.682068673166596e-07, + "loss": 0.4413, + "mean_token_accuracy": 0.8564242720603943, + "num_tokens": 88842589.0, + "step": 2285 + }, + { + "epoch": 0.2908026968579061, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.463859558105469e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.32792854309082, + "learning_rate": 9.686307757524374e-07, + "loss": 0.3776, + "mean_token_accuracy": 0.8716217279434204, + "num_tokens": 88881722.0, + "step": 2286 + }, + { + "epoch": 0.29092990713649663, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.463859558105469e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.281277179718018, + "learning_rate": 9.690546841882153e-07, + "loss": 0.3311, + "mean_token_accuracy": 0.8877597451210022, + "num_tokens": 88919355.0, + "step": 2287 + }, + { + "epoch": 0.29105711741508716, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.463859558105469e-06, + "ewc_loss_parallel": 7.212162017822266e-06, + "grad_norm": 7.3026628494262695, + "learning_rate": 9.694785926239931e-07, + "loss": 0.392, + "mean_token_accuracy": 0.8679623603820801, + "num_tokens": 88960699.0, + "step": 2288 + }, + { + "epoch": 0.29118432769367764, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.271766662597656e-06, + "grad_norm": 7.47360897064209, + "learning_rate": 9.69902501059771e-07, + "loss": 0.4321, + "mean_token_accuracy": 0.8549717664718628, + "num_tokens": 88996020.0, + "step": 2289 + }, + { + "epoch": 0.29131153797226816, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.3361310958862305, + "learning_rate": 9.703264094955488e-07, + "loss": 0.4089, + "mean_token_accuracy": 0.8625437021255493, + "num_tokens": 89035458.0, + "step": 2290 + }, + { + "epoch": 0.2914387482508587, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.336990833282471, + "learning_rate": 9.707503179313269e-07, + "loss": 0.3769, + "mean_token_accuracy": 0.8769417405128479, + "num_tokens": 89076779.0, + "step": 2291 + }, + { + "epoch": 0.29156595852944917, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.346866130828857, + "learning_rate": 9.711742263671047e-07, + "loss": 0.3711, + "mean_token_accuracy": 0.8724681735038757, + "num_tokens": 89115312.0, + "step": 2292 + }, + { + "epoch": 0.2916931688080397, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.317037582397461, + "learning_rate": 9.715981348028826e-07, + "loss": 0.3804, + "mean_token_accuracy": 0.8708534836769104, + "num_tokens": 89156703.0, + "step": 2293 + }, + { + "epoch": 0.2918203790866302, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.481770038604736, + "learning_rate": 9.720220432386604e-07, + "loss": 0.3899, + "mean_token_accuracy": 0.8692764043807983, + "num_tokens": 89190346.0, + "step": 2294 + }, + { + "epoch": 0.2919475893652207, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.361173629760742e-06, + "grad_norm": 7.292775630950928, + "learning_rate": 9.724459516744383e-07, + "loss": 0.3628, + "mean_token_accuracy": 0.8764686584472656, + "num_tokens": 89231095.0, + "step": 2295 + }, + { + "epoch": 0.2920747996438112, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.373158931732178, + "learning_rate": 9.728698601102161e-07, + "loss": 0.3548, + "mean_token_accuracy": 0.8811297416687012, + "num_tokens": 89271094.0, + "step": 2296 + }, + { + "epoch": 0.29220200992240175, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.455630302429199, + "learning_rate": 9.73293768545994e-07, + "loss": 0.3924, + "mean_token_accuracy": 0.864696741104126, + "num_tokens": 89305570.0, + "step": 2297 + }, + { + "epoch": 0.2923292202009922, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.271766662597656e-06, + "grad_norm": 7.287979602813721, + "learning_rate": 9.737176769817718e-07, + "loss": 0.4149, + "mean_token_accuracy": 0.8620771169662476, + "num_tokens": 89351838.0, + "step": 2298 + }, + { + "epoch": 0.29245643047958275, + "ewc_loss": 0.00799560546875, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.18235969543457e-06, + "grad_norm": 7.474092483520508, + "learning_rate": 9.741415854175499e-07, + "loss": 0.4614, + "mean_token_accuracy": 0.8504995107650757, + "num_tokens": 89389895.0, + "step": 2299 + }, + { + "epoch": 0.2925836407581733, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.415060520172119, + "learning_rate": 9.745654938533277e-07, + "loss": 0.4072, + "mean_token_accuracy": 0.8615137338638306, + "num_tokens": 89427683.0, + "step": 2300 + }, + { + "epoch": 0.29271085103676375, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.412556171417236, + "learning_rate": 9.749894022891056e-07, + "loss": 0.4329, + "mean_token_accuracy": 0.8578433990478516, + "num_tokens": 89468082.0, + "step": 2301 + }, + { + "epoch": 0.2928380613153543, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.52346420288086e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.350097179412842, + "learning_rate": 9.754133107248834e-07, + "loss": 0.3582, + "mean_token_accuracy": 0.8783398866653442, + "num_tokens": 89507564.0, + "step": 2302 + }, + { + "epoch": 0.2929652715939448, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.212162017822266e-06, + "grad_norm": 7.428549766540527, + "learning_rate": 9.758372191606612e-07, + "loss": 0.4362, + "mean_token_accuracy": 0.852567195892334, + "num_tokens": 89544551.0, + "step": 2303 + }, + { + "epoch": 0.2930924818725353, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.323663234710693, + "learning_rate": 9.76261127596439e-07, + "loss": 0.3446, + "mean_token_accuracy": 0.8819814920425415, + "num_tokens": 89580665.0, + "step": 2304 + }, + { + "epoch": 0.2932196921511258, + "ewc_loss": 0.008056640625, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.212162017822266e-06, + "grad_norm": 7.416016101837158, + "learning_rate": 9.76685036032217e-07, + "loss": 0.4175, + "mean_token_accuracy": 0.8600893616676331, + "num_tokens": 89617842.0, + "step": 2305 + }, + { + "epoch": 0.29334690242971634, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.268280506134033, + "learning_rate": 9.771089444679948e-07, + "loss": 0.378, + "mean_token_accuracy": 0.8737208843231201, + "num_tokens": 89659908.0, + "step": 2306 + }, + { + "epoch": 0.2934741127083068, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.4479660987854, + "learning_rate": 9.775328529037728e-07, + "loss": 0.3962, + "mean_token_accuracy": 0.8654872179031372, + "num_tokens": 89697659.0, + "step": 2307 + }, + { + "epoch": 0.29360132298689734, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.569732666015625, + "learning_rate": 9.779567613395507e-07, + "loss": 0.4214, + "mean_token_accuracy": 0.8608795404434204, + "num_tokens": 89730143.0, + "step": 2308 + }, + { + "epoch": 0.29372853326548787, + "ewc_loss": 0.00823974609375, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.361173629760742e-06, + "grad_norm": 7.389254570007324, + "learning_rate": 9.783806697753285e-07, + "loss": 0.3903, + "mean_token_accuracy": 0.8690339922904968, + "num_tokens": 89767674.0, + "step": 2309 + }, + { + "epoch": 0.29385574354407834, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.326120376586914, + "learning_rate": 9.788045782111064e-07, + "loss": 0.3659, + "mean_token_accuracy": 0.8759152293205261, + "num_tokens": 89811830.0, + "step": 2310 + }, + { + "epoch": 0.29398295382266887, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.271766662597656e-06, + "grad_norm": 7.325500011444092, + "learning_rate": 9.792284866468842e-07, + "loss": 0.35, + "mean_token_accuracy": 0.8786364793777466, + "num_tokens": 89849939.0, + "step": 2311 + }, + { + "epoch": 0.2941101641012594, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.442232608795166, + "learning_rate": 9.79652395082662e-07, + "loss": 0.3594, + "mean_token_accuracy": 0.8750684261322021, + "num_tokens": 89882571.0, + "step": 2312 + }, + { + "epoch": 0.29423737437984987, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.5324788093566895, + "learning_rate": 9.8007630351844e-07, + "loss": 0.4308, + "mean_token_accuracy": 0.8567706346511841, + "num_tokens": 89916242.0, + "step": 2313 + }, + { + "epoch": 0.2943645846584404, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.493861198425293, + "learning_rate": 9.805002119542178e-07, + "loss": 0.4341, + "mean_token_accuracy": 0.8560778498649597, + "num_tokens": 89952262.0, + "step": 2314 + }, + { + "epoch": 0.2944917949370309, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.271766662597656e-06, + "grad_norm": 7.312155246734619, + "learning_rate": 9.809241203899958e-07, + "loss": 0.3874, + "mean_token_accuracy": 0.8710397481918335, + "num_tokens": 89994817.0, + "step": 2315 + }, + { + "epoch": 0.2946190052156214, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.271766662597656e-06, + "grad_norm": 7.489168167114258, + "learning_rate": 9.813480288257737e-07, + "loss": 0.4163, + "mean_token_accuracy": 0.857021689414978, + "num_tokens": 90029953.0, + "step": 2316 + }, + { + "epoch": 0.2947462154942119, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.321094036102295, + "learning_rate": 9.817719372615515e-07, + "loss": 0.3652, + "mean_token_accuracy": 0.8767204284667969, + "num_tokens": 90069333.0, + "step": 2317 + }, + { + "epoch": 0.29487342577280246, + "ewc_loss": 0.00823974609375, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.361173629760742e-06, + "grad_norm": 7.389962196350098, + "learning_rate": 9.821958456973294e-07, + "loss": 0.338, + "mean_token_accuracy": 0.8855011463165283, + "num_tokens": 90104220.0, + "step": 2318 + }, + { + "epoch": 0.29500063605139293, + "ewc_loss": 0.00823974609375, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.3909759521484375e-06, + "grad_norm": 7.401897430419922, + "learning_rate": 9.826197541331072e-07, + "loss": 0.3566, + "mean_token_accuracy": 0.8804423213005066, + "num_tokens": 90140071.0, + "step": 2319 + }, + { + "epoch": 0.29512784632998346, + "ewc_loss": 0.00823974609375, + "ewc_loss_diag": 8.58306884765625e-06, + "ewc_loss_parallel": 7.361173629760742e-06, + "grad_norm": 7.402684211730957, + "learning_rate": 9.83043662568885e-07, + "loss": 0.4017, + "mean_token_accuracy": 0.8656095266342163, + "num_tokens": 90179835.0, + "step": 2320 + }, + { + "epoch": 0.295255056608574, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.378591537475586, + "learning_rate": 9.83467571004663e-07, + "loss": 0.3679, + "mean_token_accuracy": 0.8760769367218018, + "num_tokens": 90221177.0, + "step": 2321 + }, + { + "epoch": 0.29538226688716446, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.345334529876709, + "learning_rate": 9.838914794404407e-07, + "loss": 0.3475, + "mean_token_accuracy": 0.8825345039367676, + "num_tokens": 90263373.0, + "step": 2322 + }, + { + "epoch": 0.295509477165755, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.506104469299316, + "learning_rate": 9.843153878762188e-07, + "loss": 0.415, + "mean_token_accuracy": 0.8637019991874695, + "num_tokens": 90300951.0, + "step": 2323 + }, + { + "epoch": 0.2956366874443455, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.405440330505371, + "learning_rate": 9.847392963119966e-07, + "loss": 0.3753, + "mean_token_accuracy": 0.8722925782203674, + "num_tokens": 90340076.0, + "step": 2324 + }, + { + "epoch": 0.295763897722936, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.271766662597656e-06, + "grad_norm": 7.302032470703125, + "learning_rate": 9.851632047477745e-07, + "loss": 0.3836, + "mean_token_accuracy": 0.8687724471092224, + "num_tokens": 90383662.0, + "step": 2325 + }, + { + "epoch": 0.2958911080015265, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.444728374481201, + "learning_rate": 9.855871131835523e-07, + "loss": 0.3397, + "mean_token_accuracy": 0.8829261064529419, + "num_tokens": 90419715.0, + "step": 2326 + }, + { + "epoch": 0.29601831828011704, + "ewc_loss": 0.00823974609375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.361173629760742e-06, + "grad_norm": 7.381535053253174, + "learning_rate": 9.860110216193302e-07, + "loss": 0.3448, + "mean_token_accuracy": 0.8828949332237244, + "num_tokens": 90461151.0, + "step": 2327 + }, + { + "epoch": 0.2961455285587075, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.450798034667969, + "learning_rate": 9.86434930055108e-07, + "loss": 0.3544, + "mean_token_accuracy": 0.8816796541213989, + "num_tokens": 90494807.0, + "step": 2328 + }, + { + "epoch": 0.29627273883729804, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.271766662597656e-06, + "grad_norm": 7.457634449005127, + "learning_rate": 9.868588384908859e-07, + "loss": 0.3496, + "mean_token_accuracy": 0.8805110454559326, + "num_tokens": 90530374.0, + "step": 2329 + }, + { + "epoch": 0.2963999491158886, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.371271133422852, + "learning_rate": 9.872827469266637e-07, + "loss": 0.356, + "mean_token_accuracy": 0.8767834901809692, + "num_tokens": 90572040.0, + "step": 2330 + }, + { + "epoch": 0.2965271593944791, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.241964340209961e-06, + "grad_norm": 7.395590782165527, + "learning_rate": 9.877066553624418e-07, + "loss": 0.3696, + "mean_token_accuracy": 0.8751010894775391, + "num_tokens": 90608918.0, + "step": 2331 + }, + { + "epoch": 0.2966543696730696, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.4622392654418945, + "learning_rate": 9.881305637982196e-07, + "loss": 0.3799, + "mean_token_accuracy": 0.8724024891853333, + "num_tokens": 90650206.0, + "step": 2332 + }, + { + "epoch": 0.2967815799516601, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.481339931488037, + "learning_rate": 9.885544722339975e-07, + "loss": 0.3719, + "mean_token_accuracy": 0.8723055124282837, + "num_tokens": 90686240.0, + "step": 2333 + }, + { + "epoch": 0.29690879023025063, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.404257297515869, + "learning_rate": 9.889783806697753e-07, + "loss": 0.352, + "mean_token_accuracy": 0.8801442384719849, + "num_tokens": 90723220.0, + "step": 2334 + }, + { + "epoch": 0.2970360005088411, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.4568915367126465, + "learning_rate": 9.894022891055532e-07, + "loss": 0.3403, + "mean_token_accuracy": 0.8853002190589905, + "num_tokens": 90760617.0, + "step": 2335 + }, + { + "epoch": 0.29716321078743163, + "ewc_loss": 0.00811767578125, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.271766662597656e-06, + "grad_norm": 7.438316345214844, + "learning_rate": 9.89826197541331e-07, + "loss": 0.3317, + "mean_token_accuracy": 0.8878291249275208, + "num_tokens": 90794641.0, + "step": 2336 + }, + { + "epoch": 0.29729042106602216, + "ewc_loss": 0.00823974609375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.361173629760742e-06, + "grad_norm": 7.484931468963623, + "learning_rate": 9.902501059771089e-07, + "loss": 0.3819, + "mean_token_accuracy": 0.8717444539070129, + "num_tokens": 90830067.0, + "step": 2337 + }, + { + "epoch": 0.29741763134461263, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.4872565269470215, + "learning_rate": 9.906740144128867e-07, + "loss": 0.4002, + "mean_token_accuracy": 0.8622658848762512, + "num_tokens": 90867570.0, + "step": 2338 + }, + { + "epoch": 0.29754484162320316, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.477321624755859, + "learning_rate": 9.910979228486648e-07, + "loss": 0.3393, + "mean_token_accuracy": 0.8852907419204712, + "num_tokens": 90904348.0, + "step": 2339 + }, + { + "epoch": 0.2976720519017937, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.481971263885498, + "learning_rate": 9.915218312844426e-07, + "loss": 0.4097, + "mean_token_accuracy": 0.862967848777771, + "num_tokens": 90943089.0, + "step": 2340 + }, + { + "epoch": 0.29779926218038416, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.4256391525268555, + "learning_rate": 9.919457397202205e-07, + "loss": 0.4179, + "mean_token_accuracy": 0.8598314523696899, + "num_tokens": 90981914.0, + "step": 2341 + }, + { + "epoch": 0.2979264724589747, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.64267349243164e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.581008434295654, + "learning_rate": 9.923696481559983e-07, + "loss": 0.3557, + "mean_token_accuracy": 0.8788869976997375, + "num_tokens": 91013159.0, + "step": 2342 + }, + { + "epoch": 0.2980536827375652, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.702278137207031e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.499526023864746, + "learning_rate": 9.927935565917761e-07, + "loss": 0.4047, + "mean_token_accuracy": 0.8629431128501892, + "num_tokens": 91050743.0, + "step": 2343 + }, + { + "epoch": 0.2981808930161557, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.702278137207031e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.407937049865723, + "learning_rate": 9.93217465027554e-07, + "loss": 0.3973, + "mean_token_accuracy": 0.8689194321632385, + "num_tokens": 91094278.0, + "step": 2344 + }, + { + "epoch": 0.2983081032947462, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.761882781982422e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.628378391265869, + "learning_rate": 9.936413734633318e-07, + "loss": 0.4502, + "mean_token_accuracy": 0.854846179485321, + "num_tokens": 91126779.0, + "step": 2345 + }, + { + "epoch": 0.29843531357333675, + "ewc_loss": 0.00823974609375, + "ewc_loss_diag": 8.761882781982422e-06, + "ewc_loss_parallel": 7.3909759521484375e-06, + "grad_norm": 7.319370746612549, + "learning_rate": 9.940652818991097e-07, + "loss": 0.355, + "mean_token_accuracy": 0.880833625793457, + "num_tokens": 91167573.0, + "step": 2346 + }, + { + "epoch": 0.2985625238519272, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.761882781982422e-06, + "ewc_loss_parallel": 7.331371307373047e-06, + "grad_norm": 7.499589443206787, + "learning_rate": 9.944891903348877e-07, + "loss": 0.4091, + "mean_token_accuracy": 0.8637481331825256, + "num_tokens": 91210262.0, + "step": 2347 + }, + { + "epoch": 0.29868973413051775, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 8.761882781982422e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.406425476074219, + "learning_rate": 9.949130987706656e-07, + "loss": 0.3639, + "mean_token_accuracy": 0.8775191903114319, + "num_tokens": 91252016.0, + "step": 2348 + }, + { + "epoch": 0.2988169444091083, + "ewc_loss": 0.00823974609375, + "ewc_loss_diag": 8.761882781982422e-06, + "ewc_loss_parallel": 7.361173629760742e-06, + "grad_norm": 7.43483304977417, + "learning_rate": 9.953370072064432e-07, + "loss": 0.3557, + "mean_token_accuracy": 0.8786563277244568, + "num_tokens": 91291720.0, + "step": 2349 + }, + { + "epoch": 0.29894415468769875, + "ewc_loss": 0.00823974609375, + "ewc_loss_diag": 8.761882781982422e-06, + "ewc_loss_parallel": 7.3909759521484375e-06, + "grad_norm": 7.446240425109863, + "learning_rate": 9.957609156422213e-07, + "loss": 0.4301, + "mean_token_accuracy": 0.8549740314483643, + "num_tokens": 91334741.0, + "step": 2350 + }, + { + "epoch": 0.2990713649662893, + "ewc_loss": 0.0081787109375, + "ewc_loss_diag": 8.821487426757812e-06, + "ewc_loss_parallel": 7.3015689849853516e-06, + "grad_norm": 7.663629055023193, + "learning_rate": 9.961848240779991e-07, + "loss": 0.3732, + "mean_token_accuracy": 0.8755067586898804, + "num_tokens": 91365360.0, + "step": 2351 + }, + { + "epoch": 0.2991985752448798, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 8.821487426757812e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.415438175201416, + "learning_rate": 9.96608732513777e-07, + "loss": 0.367, + "mean_token_accuracy": 0.8778489828109741, + "num_tokens": 91406239.0, + "step": 2352 + }, + { + "epoch": 0.2993257855234703, + "ewc_loss": 0.00823974609375, + "ewc_loss_diag": 8.821487426757812e-06, + "ewc_loss_parallel": 7.361173629760742e-06, + "grad_norm": 7.435232162475586, + "learning_rate": 9.970326409495548e-07, + "loss": 0.4232, + "mean_token_accuracy": 0.8548664450645447, + "num_tokens": 91446525.0, + "step": 2353 + }, + { + "epoch": 0.2994529958020608, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 8.761882781982422e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.501745700836182, + "learning_rate": 9.974565493853327e-07, + "loss": 0.4373, + "mean_token_accuracy": 0.8520427942276001, + "num_tokens": 91486324.0, + "step": 2354 + }, + { + "epoch": 0.29958020608065133, + "ewc_loss": 0.00823974609375, + "ewc_loss_diag": 8.821487426757812e-06, + "ewc_loss_parallel": 7.361173629760742e-06, + "grad_norm": 7.635654449462891, + "learning_rate": 9.978804578211107e-07, + "loss": 0.4097, + "mean_token_accuracy": 0.8625116944313049, + "num_tokens": 91522994.0, + "step": 2355 + }, + { + "epoch": 0.2997074163592418, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 8.821487426757812e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.576040744781494, + "learning_rate": 9.983043662568886e-07, + "loss": 0.4048, + "mean_token_accuracy": 0.8649944067001343, + "num_tokens": 91556579.0, + "step": 2356 + }, + { + "epoch": 0.29983462663783234, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 8.821487426757812e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.498610496520996, + "learning_rate": 9.987282746926662e-07, + "loss": 0.3749, + "mean_token_accuracy": 0.8746602535247803, + "num_tokens": 91593015.0, + "step": 2357 + }, + { + "epoch": 0.29996183691642286, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 8.821487426757812e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.449615478515625, + "learning_rate": 9.991521831284443e-07, + "loss": 0.386, + "mean_token_accuracy": 0.8716095089912415, + "num_tokens": 91633391.0, + "step": 2358 + }, + { + "epoch": 0.30008904719501334, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 8.821487426757812e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.528264999389648, + "learning_rate": 9.995760915642221e-07, + "loss": 0.3379, + "mean_token_accuracy": 0.8849537372589111, + "num_tokens": 91664017.0, + "step": 2359 + }, + { + "epoch": 0.30021625747360386, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 8.821487426757812e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.486080169677734, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8811303973197937, + "num_tokens": 91701327.0, + "step": 2360 + }, + { + "epoch": 0.3003434677521944, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.487917423248291, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8687684535980225, + "num_tokens": 91738255.0, + "step": 2361 + }, + { + "epoch": 0.30047067803078487, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.518035411834717, + "learning_rate": 1e-06, + "loss": 0.445, + "mean_token_accuracy": 0.8509821891784668, + "num_tokens": 91778318.0, + "step": 2362 + }, + { + "epoch": 0.3005978883093754, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.437720775604248, + "learning_rate": 1e-06, + "loss": 0.4073, + "mean_token_accuracy": 0.8664047718048096, + "num_tokens": 91827088.0, + "step": 2363 + }, + { + "epoch": 0.3007250985879659, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 8.821487426757812e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.698140621185303, + "learning_rate": 1e-06, + "loss": 0.4876, + "mean_token_accuracy": 0.844501256942749, + "num_tokens": 91860367.0, + "step": 2364 + }, + { + "epoch": 0.3008523088665564, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 8.821487426757812e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.467785835266113, + "learning_rate": 1e-06, + "loss": 0.3945, + "mean_token_accuracy": 0.8710734844207764, + "num_tokens": 91896605.0, + "step": 2365 + }, + { + "epoch": 0.3009795191451469, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 8.821487426757812e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.442690849304199, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8772076964378357, + "num_tokens": 91938376.0, + "step": 2366 + }, + { + "epoch": 0.30110672942373745, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.493662357330322, + "learning_rate": 1e-06, + "loss": 0.4097, + "mean_token_accuracy": 0.8606441617012024, + "num_tokens": 91982394.0, + "step": 2367 + }, + { + "epoch": 0.3012339397023279, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.608535289764404, + "learning_rate": 1e-06, + "loss": 0.4023, + "mean_token_accuracy": 0.862450361251831, + "num_tokens": 92014454.0, + "step": 2368 + }, + { + "epoch": 0.30136114998091845, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.580708980560303, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8704277276992798, + "num_tokens": 92049638.0, + "step": 2369 + }, + { + "epoch": 0.301488360259509, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.499816417694092, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8802058696746826, + "num_tokens": 92086351.0, + "step": 2370 + }, + { + "epoch": 0.30161557053809945, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.508693695068359, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8778162002563477, + "num_tokens": 92125193.0, + "step": 2371 + }, + { + "epoch": 0.30174278081669, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.618085861206055, + "learning_rate": 1e-06, + "loss": 0.4283, + "mean_token_accuracy": 0.8573007583618164, + "num_tokens": 92169625.0, + "step": 2372 + }, + { + "epoch": 0.3018699910952805, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.4311747550964355, + "learning_rate": 1e-06, + "loss": 0.4019, + "mean_token_accuracy": 0.8665385246276855, + "num_tokens": 92212151.0, + "step": 2373 + }, + { + "epoch": 0.301997201373871, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.489933967590332, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8837987184524536, + "num_tokens": 92251870.0, + "step": 2374 + }, + { + "epoch": 0.3021244116524615, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.447360992431641, + "learning_rate": 1e-06, + "loss": 0.3977, + "mean_token_accuracy": 0.8656007647514343, + "num_tokens": 92293322.0, + "step": 2375 + }, + { + "epoch": 0.30225162193105204, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.580033302307129, + "learning_rate": 1e-06, + "loss": 0.4075, + "mean_token_accuracy": 0.8646604418754578, + "num_tokens": 92330301.0, + "step": 2376 + }, + { + "epoch": 0.3023788322096425, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.523468971252441, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.8740296363830566, + "num_tokens": 92370979.0, + "step": 2377 + }, + { + "epoch": 0.30250604248823304, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.579728126525879, + "learning_rate": 1e-06, + "loss": 0.4171, + "mean_token_accuracy": 0.862889289855957, + "num_tokens": 92411348.0, + "step": 2378 + }, + { + "epoch": 0.30263325276682357, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.569806098937988, + "learning_rate": 1e-06, + "loss": 0.426, + "mean_token_accuracy": 0.8559798002243042, + "num_tokens": 92447127.0, + "step": 2379 + }, + { + "epoch": 0.3027604630454141, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.524415969848633, + "learning_rate": 1e-06, + "loss": 0.398, + "mean_token_accuracy": 0.8698348999023438, + "num_tokens": 92485812.0, + "step": 2380 + }, + { + "epoch": 0.30288767332400457, + "ewc_loss": 0.00823974609375, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.361173629760742e-06, + "grad_norm": 7.423532009124756, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8739687204360962, + "num_tokens": 92525123.0, + "step": 2381 + }, + { + "epoch": 0.3030148836025951, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.548840045928955, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8726800084114075, + "num_tokens": 92559035.0, + "step": 2382 + }, + { + "epoch": 0.3031420938811856, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.4379682540893555, + "learning_rate": 1e-06, + "loss": 0.361, + "mean_token_accuracy": 0.8788601756095886, + "num_tokens": 92597543.0, + "step": 2383 + }, + { + "epoch": 0.3032693041597761, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.3909759521484375e-06, + "grad_norm": 7.671151161193848, + "learning_rate": 1e-06, + "loss": 0.4508, + "mean_token_accuracy": 0.8508923053741455, + "num_tokens": 92629940.0, + "step": 2384 + }, + { + "epoch": 0.3033965144383666, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 8.940696716308594e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.611550331115723, + "learning_rate": 1e-06, + "loss": 0.4227, + "mean_token_accuracy": 0.8598064184188843, + "num_tokens": 92665943.0, + "step": 2385 + }, + { + "epoch": 0.30352372471695716, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.534645080566406, + "learning_rate": 1e-06, + "loss": 0.3752, + "mean_token_accuracy": 0.8715506792068481, + "num_tokens": 92700229.0, + "step": 2386 + }, + { + "epoch": 0.30365093499554763, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.3909759521484375e-06, + "grad_norm": 7.4550700187683105, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8723326325416565, + "num_tokens": 92737674.0, + "step": 2387 + }, + { + "epoch": 0.30377814527413816, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.3909759521484375e-06, + "grad_norm": 7.449907302856445, + "learning_rate": 1e-06, + "loss": 0.413, + "mean_token_accuracy": 0.8626583814620972, + "num_tokens": 92780087.0, + "step": 2388 + }, + { + "epoch": 0.3039053555527287, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.483075141906738, + "learning_rate": 1e-06, + "loss": 0.3945, + "mean_token_accuracy": 0.8655621409416199, + "num_tokens": 92822656.0, + "step": 2389 + }, + { + "epoch": 0.30403256583131916, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.498039722442627, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8879784345626831, + "num_tokens": 92859648.0, + "step": 2390 + }, + { + "epoch": 0.3041597761099097, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.574606895446777, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8672532439231873, + "num_tokens": 92893924.0, + "step": 2391 + }, + { + "epoch": 0.3042869863885002, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.483346462249756, + "learning_rate": 1e-06, + "loss": 0.4207, + "mean_token_accuracy": 0.8581749796867371, + "num_tokens": 92934260.0, + "step": 2392 + }, + { + "epoch": 0.3044141966670907, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.485086441040039, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8739913105964661, + "num_tokens": 92974179.0, + "step": 2393 + }, + { + "epoch": 0.3045414069456812, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.707939147949219, + "learning_rate": 1e-06, + "loss": 0.4128, + "mean_token_accuracy": 0.8603091239929199, + "num_tokens": 93006409.0, + "step": 2394 + }, + { + "epoch": 0.30466861722427174, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.41212797164917, + "learning_rate": 1e-06, + "loss": 0.3823, + "mean_token_accuracy": 0.8699539303779602, + "num_tokens": 93047492.0, + "step": 2395 + }, + { + "epoch": 0.3047958275028622, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.5027241706848145, + "learning_rate": 1e-06, + "loss": 0.4034, + "mean_token_accuracy": 0.8650416731834412, + "num_tokens": 93087472.0, + "step": 2396 + }, + { + "epoch": 0.30492303778145274, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.5080437660217285, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.8747020959854126, + "num_tokens": 93124395.0, + "step": 2397 + }, + { + "epoch": 0.30505024806004327, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.4650068283081055, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.8714202642440796, + "num_tokens": 93167997.0, + "step": 2398 + }, + { + "epoch": 0.30517745833863374, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.5507426261901855, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8731139898300171, + "num_tokens": 93204877.0, + "step": 2399 + }, + { + "epoch": 0.3053046686172243, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.586262226104736, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.878512442111969, + "num_tokens": 93238074.0, + "step": 2400 + }, + { + "epoch": 0.3054318788958148, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.41689395904541, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.8728439807891846, + "num_tokens": 93282913.0, + "step": 2401 + }, + { + "epoch": 0.3055590891744053, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.485907554626465, + "learning_rate": 1e-06, + "loss": 0.3971, + "mean_token_accuracy": 0.8650209903717041, + "num_tokens": 93327168.0, + "step": 2402 + }, + { + "epoch": 0.3056862994529958, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.555721282958984, + "learning_rate": 1e-06, + "loss": 0.3926, + "mean_token_accuracy": 0.8673493266105652, + "num_tokens": 93365838.0, + "step": 2403 + }, + { + "epoch": 0.30581350973158633, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.537426471710205, + "learning_rate": 1e-06, + "loss": 0.4128, + "mean_token_accuracy": 0.8630878925323486, + "num_tokens": 93411416.0, + "step": 2404 + }, + { + "epoch": 0.3059407200101768, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.568943500518799, + "learning_rate": 1e-06, + "loss": 0.4384, + "mean_token_accuracy": 0.8601731061935425, + "num_tokens": 93449360.0, + "step": 2405 + }, + { + "epoch": 0.30606793028876733, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.7621941566467285, + "learning_rate": 1e-06, + "loss": 0.4227, + "mean_token_accuracy": 0.8591558337211609, + "num_tokens": 93489861.0, + "step": 2406 + }, + { + "epoch": 0.30619514056735786, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.059906005859375e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.48041296005249, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.869113564491272, + "num_tokens": 93529406.0, + "step": 2407 + }, + { + "epoch": 0.30632235084594833, + "ewc_loss": 0.00823974609375, + "ewc_loss_diag": 9.119510650634766e-06, + "ewc_loss_parallel": 7.361173629760742e-06, + "grad_norm": 7.511773109436035, + "learning_rate": 1e-06, + "loss": 0.3948, + "mean_token_accuracy": 0.8678555488586426, + "num_tokens": 93569798.0, + "step": 2408 + }, + { + "epoch": 0.30644956112453886, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.119510650634766e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.525622367858887, + "learning_rate": 1e-06, + "loss": 0.3858, + "mean_token_accuracy": 0.8694885969161987, + "num_tokens": 93607188.0, + "step": 2409 + }, + { + "epoch": 0.3065767714031294, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 9.119510650634766e-06, + "ewc_loss_parallel": 7.3909759521484375e-06, + "grad_norm": 7.518735885620117, + "learning_rate": 1e-06, + "loss": 0.3848, + "mean_token_accuracy": 0.8685702085494995, + "num_tokens": 93642023.0, + "step": 2410 + }, + { + "epoch": 0.30670398168171986, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 9.119510650634766e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.444423675537109, + "learning_rate": 1e-06, + "loss": 0.3936, + "mean_token_accuracy": 0.8665072917938232, + "num_tokens": 93687482.0, + "step": 2411 + }, + { + "epoch": 0.3068311919603104, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.179115295410156e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.513831615447998, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8710516095161438, + "num_tokens": 93722846.0, + "step": 2412 + }, + { + "epoch": 0.3069584022389009, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.179115295410156e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.729153633117676, + "learning_rate": 1e-06, + "loss": 0.3888, + "mean_token_accuracy": 0.8707176446914673, + "num_tokens": 93757850.0, + "step": 2413 + }, + { + "epoch": 0.3070856125174914, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.179115295410156e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.395181655883789, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8759065866470337, + "num_tokens": 93801515.0, + "step": 2414 + }, + { + "epoch": 0.3072128227960819, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.179115295410156e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.505819797515869, + "learning_rate": 1e-06, + "loss": 0.4036, + "mean_token_accuracy": 0.8649196028709412, + "num_tokens": 93840866.0, + "step": 2415 + }, + { + "epoch": 0.30734003307467245, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.179115295410156e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.529305934906006, + "learning_rate": 1e-06, + "loss": 0.4307, + "mean_token_accuracy": 0.8576827049255371, + "num_tokens": 93880561.0, + "step": 2416 + }, + { + "epoch": 0.3074672433532629, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.179115295410156e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.4885406494140625, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8755981922149658, + "num_tokens": 93918766.0, + "step": 2417 + }, + { + "epoch": 0.30759445363185345, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.238719940185547e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.48654317855835, + "learning_rate": 1e-06, + "loss": 0.3928, + "mean_token_accuracy": 0.8693071007728577, + "num_tokens": 93961056.0, + "step": 2418 + }, + { + "epoch": 0.307721663910444, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.238719940185547e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.598165035247803, + "learning_rate": 1e-06, + "loss": 0.3734, + "mean_token_accuracy": 0.8706556558609009, + "num_tokens": 93992146.0, + "step": 2419 + }, + { + "epoch": 0.30784887418903445, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.238719940185547e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.488929748535156, + "learning_rate": 1e-06, + "loss": 0.4055, + "mean_token_accuracy": 0.8698865175247192, + "num_tokens": 94031814.0, + "step": 2420 + }, + { + "epoch": 0.307976084467625, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.238719940185547e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.577199459075928, + "learning_rate": 1e-06, + "loss": 0.4123, + "mean_token_accuracy": 0.8628017902374268, + "num_tokens": 94067135.0, + "step": 2421 + }, + { + "epoch": 0.3081032947462155, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.238719940185547e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.588486671447754, + "learning_rate": 1e-06, + "loss": 0.3993, + "mean_token_accuracy": 0.8643782138824463, + "num_tokens": 94104491.0, + "step": 2422 + }, + { + "epoch": 0.308230505024806, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.238719940185547e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.741032123565674, + "learning_rate": 1e-06, + "loss": 0.4216, + "mean_token_accuracy": 0.8609689474105835, + "num_tokens": 94142115.0, + "step": 2423 + }, + { + "epoch": 0.3083577153033965, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.238719940185547e-06, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.449822902679443, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8757709264755249, + "num_tokens": 94180084.0, + "step": 2424 + }, + { + "epoch": 0.30848492558198704, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.238719940185547e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.647130966186523, + "learning_rate": 1e-06, + "loss": 0.443, + "mean_token_accuracy": 0.849969744682312, + "num_tokens": 94215936.0, + "step": 2425 + }, + { + "epoch": 0.3086121358605775, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.238719940185547e-06, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.463711261749268, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8872987627983093, + "num_tokens": 94255632.0, + "step": 2426 + }, + { + "epoch": 0.30873934613916804, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.238719940185547e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.448123455047607, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8762103319168091, + "num_tokens": 94294933.0, + "step": 2427 + }, + { + "epoch": 0.30886655641775856, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.238719940185547e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.502749443054199, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8798749446868896, + "num_tokens": 94332781.0, + "step": 2428 + }, + { + "epoch": 0.3089937666963491, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.298324584960938e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.60953950881958, + "learning_rate": 1e-06, + "loss": 0.3951, + "mean_token_accuracy": 0.8653694987297058, + "num_tokens": 94368234.0, + "step": 2429 + }, + { + "epoch": 0.30912097697493957, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.494380474090576, + "learning_rate": 1e-06, + "loss": 0.3823, + "mean_token_accuracy": 0.8729141354560852, + "num_tokens": 94410410.0, + "step": 2430 + }, + { + "epoch": 0.3092481872535301, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.645374298095703, + "learning_rate": 1e-06, + "loss": 0.4172, + "mean_token_accuracy": 0.8606089949607849, + "num_tokens": 94445246.0, + "step": 2431 + }, + { + "epoch": 0.3093753975321206, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.398632049560547, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8720411658287048, + "num_tokens": 94493511.0, + "step": 2432 + }, + { + "epoch": 0.3095026078107111, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.557537078857422, + "learning_rate": 1e-06, + "loss": 0.4009, + "mean_token_accuracy": 0.8655914664268494, + "num_tokens": 94528618.0, + "step": 2433 + }, + { + "epoch": 0.3096298180893016, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.5508503913879395, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8774699568748474, + "num_tokens": 94565541.0, + "step": 2434 + }, + { + "epoch": 0.30975702836789215, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.5593485832214355, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.8700553178787231, + "num_tokens": 94604855.0, + "step": 2435 + }, + { + "epoch": 0.3098842386464826, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.743494033813477, + "learning_rate": 1e-06, + "loss": 0.428, + "mean_token_accuracy": 0.8588621616363525, + "num_tokens": 94639460.0, + "step": 2436 + }, + { + "epoch": 0.31001144892507315, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.413235664367676, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8779095411300659, + "num_tokens": 94679165.0, + "step": 2437 + }, + { + "epoch": 0.3101386592036637, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.501370906829834, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8767446279525757, + "num_tokens": 94715972.0, + "step": 2438 + }, + { + "epoch": 0.31026586948225415, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.583670139312744, + "learning_rate": 1e-06, + "loss": 0.403, + "mean_token_accuracy": 0.8632200956344604, + "num_tokens": 94756046.0, + "step": 2439 + }, + { + "epoch": 0.3103930797608447, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.477607250213623, + "learning_rate": 1e-06, + "loss": 0.41, + "mean_token_accuracy": 0.8637591600418091, + "num_tokens": 94797799.0, + "step": 2440 + }, + { + "epoch": 0.3105202900394352, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.711360454559326, + "learning_rate": 1e-06, + "loss": 0.4005, + "mean_token_accuracy": 0.8600545525550842, + "num_tokens": 94832982.0, + "step": 2441 + }, + { + "epoch": 0.3106475003180257, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.581712245941162, + "learning_rate": 1e-06, + "loss": 0.4179, + "mean_token_accuracy": 0.8611665964126587, + "num_tokens": 94868505.0, + "step": 2442 + }, + { + "epoch": 0.3107747105966162, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.357929229736328e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.559977054595947, + "learning_rate": 1e-06, + "loss": 0.4231, + "mean_token_accuracy": 0.8614797592163086, + "num_tokens": 94906917.0, + "step": 2443 + }, + { + "epoch": 0.31090192087520674, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.447793483734131, + "learning_rate": 1e-06, + "loss": 0.4086, + "mean_token_accuracy": 0.8638389110565186, + "num_tokens": 94953878.0, + "step": 2444 + }, + { + "epoch": 0.3110291311537972, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.499410629272461, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8773418664932251, + "num_tokens": 94990643.0, + "step": 2445 + }, + { + "epoch": 0.31115634143238774, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.528757572174072, + "learning_rate": 1e-06, + "loss": 0.3925, + "mean_token_accuracy": 0.8671489953994751, + "num_tokens": 95032027.0, + "step": 2446 + }, + { + "epoch": 0.31128355171097827, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.516521453857422, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8833976984024048, + "num_tokens": 95065628.0, + "step": 2447 + }, + { + "epoch": 0.31141076198956874, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.54987907409668, + "learning_rate": 1e-06, + "loss": 0.4026, + "mean_token_accuracy": 0.8659734725952148, + "num_tokens": 95105422.0, + "step": 2448 + }, + { + "epoch": 0.31153797226815927, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.5494890213012695, + "learning_rate": 1e-06, + "loss": 0.3949, + "mean_token_accuracy": 0.8677153587341309, + "num_tokens": 95141735.0, + "step": 2449 + }, + { + "epoch": 0.3116651825467498, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.460811614990234, + "learning_rate": 1e-06, + "loss": 0.3979, + "mean_token_accuracy": 0.8610491752624512, + "num_tokens": 95184335.0, + "step": 2450 + }, + { + "epoch": 0.31179239282534027, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.611898899078369, + "learning_rate": 1e-06, + "loss": 0.4142, + "mean_token_accuracy": 0.8638794422149658, + "num_tokens": 95220055.0, + "step": 2451 + }, + { + "epoch": 0.3119196031039308, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.548903465270996, + "learning_rate": 1e-06, + "loss": 0.4306, + "mean_token_accuracy": 0.8581551313400269, + "num_tokens": 95261027.0, + "step": 2452 + }, + { + "epoch": 0.3120468133825213, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.5103583335876465, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.8717371821403503, + "num_tokens": 95301739.0, + "step": 2453 + }, + { + "epoch": 0.3121740236611118, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.521583080291748, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.873916745185852, + "num_tokens": 95338913.0, + "step": 2454 + }, + { + "epoch": 0.3123012339397023, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.523457050323486, + "learning_rate": 1e-06, + "loss": 0.3811, + "mean_token_accuracy": 0.874649167060852, + "num_tokens": 95376056.0, + "step": 2455 + }, + { + "epoch": 0.31242844421829286, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.41703462600708, + "learning_rate": 1e-06, + "loss": 0.3858, + "mean_token_accuracy": 0.8715753555297852, + "num_tokens": 95418942.0, + "step": 2456 + }, + { + "epoch": 0.31255565449688333, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.522547721862793, + "learning_rate": 1e-06, + "loss": 0.4269, + "mean_token_accuracy": 0.854695200920105, + "num_tokens": 95461304.0, + "step": 2457 + }, + { + "epoch": 0.31268286477547386, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.666431903839111, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8795475363731384, + "num_tokens": 95496141.0, + "step": 2458 + }, + { + "epoch": 0.3128100750540644, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.645341396331787, + "learning_rate": 1e-06, + "loss": 0.3969, + "mean_token_accuracy": 0.8681216835975647, + "num_tokens": 95533568.0, + "step": 2459 + }, + { + "epoch": 0.31293728533265486, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.417533874511719e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.434242248535156, + "learning_rate": 1e-06, + "loss": 0.4178, + "mean_token_accuracy": 0.8595179319381714, + "num_tokens": 95581458.0, + "step": 2460 + }, + { + "epoch": 0.3130644956112454, + "ewc_loss": 0.00830078125, + "ewc_loss_diag": 9.47713851928711e-06, + "ewc_loss_parallel": 7.361173629760742e-06, + "grad_norm": 7.622156143188477, + "learning_rate": 1e-06, + "loss": 0.4116, + "mean_token_accuracy": 0.8629108667373657, + "num_tokens": 95616232.0, + "step": 2461 + }, + { + "epoch": 0.3131917058898359, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.47713851928711e-06, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.594399452209473, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8694548606872559, + "num_tokens": 95651456.0, + "step": 2462 + }, + { + "epoch": 0.3133189161684264, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.47713851928711e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.485760688781738, + "learning_rate": 1e-06, + "loss": 0.3943, + "mean_token_accuracy": 0.8698185682296753, + "num_tokens": 95691982.0, + "step": 2463 + }, + { + "epoch": 0.3134461264470169, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.47713851928711e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.55197286605835, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8736786842346191, + "num_tokens": 95729420.0, + "step": 2464 + }, + { + "epoch": 0.31357333672560744, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.47713851928711e-06, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.547673225402832, + "learning_rate": 1e-06, + "loss": 0.4479, + "mean_token_accuracy": 0.8487233519554138, + "num_tokens": 95771382.0, + "step": 2465 + }, + { + "epoch": 0.3137005470041979, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.47713851928711e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.4692487716674805, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8780521154403687, + "num_tokens": 95815213.0, + "step": 2466 + }, + { + "epoch": 0.31382775728278844, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.47713851928711e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.5018310546875, + "learning_rate": 1e-06, + "loss": 0.3811, + "mean_token_accuracy": 0.8728528022766113, + "num_tokens": 95856470.0, + "step": 2467 + }, + { + "epoch": 0.313954967561379, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.47713851928711e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.518369197845459, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8851626515388489, + "num_tokens": 95891724.0, + "step": 2468 + }, + { + "epoch": 0.31408217783996945, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.47713851928711e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.536131858825684, + "learning_rate": 1e-06, + "loss": 0.3921, + "mean_token_accuracy": 0.8678386807441711, + "num_tokens": 95929793.0, + "step": 2469 + }, + { + "epoch": 0.31420938811856, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.59634780883789e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.465862274169922, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8838942050933838, + "num_tokens": 95970349.0, + "step": 2470 + }, + { + "epoch": 0.3143365983971505, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.59634780883789e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.473900318145752, + "learning_rate": 1e-06, + "loss": 0.401, + "mean_token_accuracy": 0.8648214340209961, + "num_tokens": 96013220.0, + "step": 2471 + }, + { + "epoch": 0.314463808675741, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.59634780883789e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.507443428039551, + "learning_rate": 1e-06, + "loss": 0.384, + "mean_token_accuracy": 0.8726764917373657, + "num_tokens": 96053533.0, + "step": 2472 + }, + { + "epoch": 0.3145910189543315, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.59634780883789e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.497747898101807, + "learning_rate": 1e-06, + "loss": 0.4258, + "mean_token_accuracy": 0.8580586910247803, + "num_tokens": 96094430.0, + "step": 2473 + }, + { + "epoch": 0.31471822923292203, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.47713851928711e-06, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.5071611404418945, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8720821142196655, + "num_tokens": 96135328.0, + "step": 2474 + }, + { + "epoch": 0.3148454395115125, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.47713851928711e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.414997100830078, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8816152811050415, + "num_tokens": 96178057.0, + "step": 2475 + }, + { + "epoch": 0.31497264979010303, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.59634780883789e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.609273910522461, + "learning_rate": 1e-06, + "loss": 0.414, + "mean_token_accuracy": 0.8651179075241089, + "num_tokens": 96218366.0, + "step": 2476 + }, + { + "epoch": 0.31509986006869356, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.59634780883789e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.486147880554199, + "learning_rate": 1e-06, + "loss": 0.4433, + "mean_token_accuracy": 0.8521022200584412, + "num_tokens": 96260819.0, + "step": 2477 + }, + { + "epoch": 0.31522707034728403, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.59634780883789e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.565946102142334, + "learning_rate": 1e-06, + "loss": 0.4145, + "mean_token_accuracy": 0.8635809421539307, + "num_tokens": 96303139.0, + "step": 2478 + }, + { + "epoch": 0.31535428062587456, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.556459426879883, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8744314312934875, + "num_tokens": 96341181.0, + "step": 2479 + }, + { + "epoch": 0.3154814909044651, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.5250749588012695, + "learning_rate": 1e-06, + "loss": 0.402, + "mean_token_accuracy": 0.8701781630516052, + "num_tokens": 96384874.0, + "step": 2480 + }, + { + "epoch": 0.3156087011830556, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.528633117675781, + "learning_rate": 1e-06, + "loss": 0.4018, + "mean_token_accuracy": 0.8677489757537842, + "num_tokens": 96426317.0, + "step": 2481 + }, + { + "epoch": 0.3157359114616461, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.606277942657471, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8758945465087891, + "num_tokens": 96464509.0, + "step": 2482 + }, + { + "epoch": 0.3158631217402366, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.752448558807373, + "learning_rate": 1e-06, + "loss": 0.3948, + "mean_token_accuracy": 0.8624531626701355, + "num_tokens": 96497073.0, + "step": 2483 + }, + { + "epoch": 0.31599033201882715, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.584987640380859, + "learning_rate": 1e-06, + "loss": 0.3902, + "mean_token_accuracy": 0.8687594532966614, + "num_tokens": 96531763.0, + "step": 2484 + }, + { + "epoch": 0.3161175422974176, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.598886489868164, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.8705766201019287, + "num_tokens": 96568141.0, + "step": 2485 + }, + { + "epoch": 0.31624475257600815, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.432494640350342, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.880581796169281, + "num_tokens": 96615181.0, + "step": 2486 + }, + { + "epoch": 0.3163719628545987, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.565806865692139, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8787471055984497, + "num_tokens": 96652141.0, + "step": 2487 + }, + { + "epoch": 0.31649917313318915, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.711948394775391, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8744775056838989, + "num_tokens": 96682817.0, + "step": 2488 + }, + { + "epoch": 0.3166263834117797, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.523670196533203, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8828182816505432, + "num_tokens": 96716351.0, + "step": 2489 + }, + { + "epoch": 0.3167535936903702, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.649722576141357, + "learning_rate": 1e-06, + "loss": 0.4396, + "mean_token_accuracy": 0.8536866307258606, + "num_tokens": 96750796.0, + "step": 2490 + }, + { + "epoch": 0.3168808039689607, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.454076766967773, + "learning_rate": 1e-06, + "loss": 0.3732, + "mean_token_accuracy": 0.8777316808700562, + "num_tokens": 96789862.0, + "step": 2491 + }, + { + "epoch": 0.3170080142475512, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.609253406524658, + "learning_rate": 1e-06, + "loss": 0.4294, + "mean_token_accuracy": 0.8568145036697388, + "num_tokens": 96829452.0, + "step": 2492 + }, + { + "epoch": 0.31713522452614173, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.609766960144043, + "learning_rate": 1e-06, + "loss": 0.4549, + "mean_token_accuracy": 0.8506406545639038, + "num_tokens": 96869788.0, + "step": 2493 + }, + { + "epoch": 0.3172624348047322, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.655952453613281e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.548423767089844, + "learning_rate": 1e-06, + "loss": 0.4251, + "mean_token_accuracy": 0.858869194984436, + "num_tokens": 96908999.0, + "step": 2494 + }, + { + "epoch": 0.31738964508332274, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.464921951293945, + "learning_rate": 1e-06, + "loss": 0.4042, + "mean_token_accuracy": 0.863034725189209, + "num_tokens": 96951520.0, + "step": 2495 + }, + { + "epoch": 0.31751685536191326, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.468168258666992, + "learning_rate": 1e-06, + "loss": 0.3768, + "mean_token_accuracy": 0.8724980354309082, + "num_tokens": 96993906.0, + "step": 2496 + }, + { + "epoch": 0.31764406564050374, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.491030693054199, + "learning_rate": 1e-06, + "loss": 0.4377, + "mean_token_accuracy": 0.8516460657119751, + "num_tokens": 97035949.0, + "step": 2497 + }, + { + "epoch": 0.31777127591909426, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.437542915344238, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8879547715187073, + "num_tokens": 97077882.0, + "step": 2498 + }, + { + "epoch": 0.3178984861976848, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.55288553237915, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8852750658988953, + "num_tokens": 97112567.0, + "step": 2499 + }, + { + "epoch": 0.31802569647627527, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.541783809661865, + "learning_rate": 1e-06, + "loss": 0.4139, + "mean_token_accuracy": 0.8623970150947571, + "num_tokens": 97152652.0, + "step": 2500 + }, + { + "epoch": 0.3181529067548658, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.638354301452637, + "learning_rate": 1e-06, + "loss": 0.4512, + "mean_token_accuracy": 0.8550732135772705, + "num_tokens": 97193840.0, + "step": 2501 + }, + { + "epoch": 0.3182801170334563, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.606803894042969, + "learning_rate": 1e-06, + "loss": 0.4768, + "mean_token_accuracy": 0.8417602777481079, + "num_tokens": 97235623.0, + "step": 2502 + }, + { + "epoch": 0.3184073273120468, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.4850592613220215, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8761946558952332, + "num_tokens": 97275797.0, + "step": 2503 + }, + { + "epoch": 0.3185345375906373, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.42116117477417, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8732513189315796, + "num_tokens": 97323595.0, + "step": 2504 + }, + { + "epoch": 0.31866174786922785, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.569940567016602, + "learning_rate": 1e-06, + "loss": 0.4054, + "mean_token_accuracy": 0.8614550828933716, + "num_tokens": 97364264.0, + "step": 2505 + }, + { + "epoch": 0.3187889581478183, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.452927112579346, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.876934289932251, + "num_tokens": 97406120.0, + "step": 2506 + }, + { + "epoch": 0.31891616842640885, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.684818744659424, + "learning_rate": 1e-06, + "loss": 0.4016, + "mean_token_accuracy": 0.8639846444129944, + "num_tokens": 97441177.0, + "step": 2507 + }, + { + "epoch": 0.3190433787049994, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.621974945068359, + "learning_rate": 1e-06, + "loss": 0.408, + "mean_token_accuracy": 0.8652708530426025, + "num_tokens": 97479175.0, + "step": 2508 + }, + { + "epoch": 0.31917058898358985, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.443897724151611, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8715355396270752, + "num_tokens": 97520029.0, + "step": 2509 + }, + { + "epoch": 0.3192977992621804, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.716929912567139, + "learning_rate": 1e-06, + "loss": 0.4338, + "mean_token_accuracy": 0.8615498542785645, + "num_tokens": 97563140.0, + "step": 2510 + }, + { + "epoch": 0.3194250095407709, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.439749240875244, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.889445960521698, + "num_tokens": 97601435.0, + "step": 2511 + }, + { + "epoch": 0.3195522198193614, + "ewc_loss": 0.00836181640625, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.3909759521484375e-06, + "grad_norm": 7.449441432952881, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8759980201721191, + "num_tokens": 97647244.0, + "step": 2512 + }, + { + "epoch": 0.3196794300979519, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.56927490234375, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.870639979839325, + "num_tokens": 97683555.0, + "step": 2513 + }, + { + "epoch": 0.31980664037654244, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.425847053527832, + "learning_rate": 1e-06, + "loss": 0.3689, + "mean_token_accuracy": 0.875848650932312, + "num_tokens": 97725911.0, + "step": 2514 + }, + { + "epoch": 0.3199338506551329, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.596730709075928, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8876931667327881, + "num_tokens": 97757263.0, + "step": 2515 + }, + { + "epoch": 0.32006106093372344, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.512461185455322, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.8717689514160156, + "num_tokens": 97798185.0, + "step": 2516 + }, + { + "epoch": 0.32018827121231397, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.450580596923828e-06, + "grad_norm": 7.516190528869629, + "learning_rate": 1e-06, + "loss": 0.3825, + "mean_token_accuracy": 0.8718395233154297, + "num_tokens": 97834786.0, + "step": 2517 + }, + { + "epoch": 0.32031548149090444, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.486724853515625, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8709874153137207, + "num_tokens": 97875854.0, + "step": 2518 + }, + { + "epoch": 0.32044269176949497, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.473814010620117, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8852027654647827, + "num_tokens": 97912710.0, + "step": 2519 + }, + { + "epoch": 0.3205699020480855, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.4803829193115234e-06, + "grad_norm": 7.542573928833008, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8794789910316467, + "num_tokens": 97950414.0, + "step": 2520 + }, + { + "epoch": 0.32069711232667597, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.775161743164062e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.409811496734619, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8790501952171326, + "num_tokens": 97994794.0, + "step": 2521 + }, + { + "epoch": 0.3208243226052665, + "ewc_loss": 0.0084228515625, + "ewc_loss_diag": 9.834766387939453e-06, + "ewc_loss_parallel": 7.420778274536133e-06, + "grad_norm": 7.655200004577637, + "learning_rate": 1e-06, + "loss": 0.4038, + "mean_token_accuracy": 0.8626139760017395, + "num_tokens": 98029312.0, + "step": 2522 + }, + { + "epoch": 0.320951532883857, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 9.834766387939453e-06, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.629438877105713, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8803645968437195, + "num_tokens": 98063318.0, + "step": 2523 + }, + { + "epoch": 0.3210787431624475, + "ewc_loss": 0.00848388671875, + "ewc_loss_diag": 9.834766387939453e-06, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.510642051696777, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.871606707572937, + "num_tokens": 98099728.0, + "step": 2524 + }, + { + "epoch": 0.32120595344103803, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.834766387939453e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.560025691986084, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.8684700727462769, + "num_tokens": 98142034.0, + "step": 2525 + }, + { + "epoch": 0.32133316371962856, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.834766387939453e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.649590015411377, + "learning_rate": 1e-06, + "loss": 0.4156, + "mean_token_accuracy": 0.8578243255615234, + "num_tokens": 98178045.0, + "step": 2526 + }, + { + "epoch": 0.32146037399821903, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 9.834766387939453e-06, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.60377311706543, + "learning_rate": 1e-06, + "loss": 0.4341, + "mean_token_accuracy": 0.8626608848571777, + "num_tokens": 98214486.0, + "step": 2527 + }, + { + "epoch": 0.32158758427680956, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 9.834766387939453e-06, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.449207782745361, + "learning_rate": 1e-06, + "loss": 0.4139, + "mean_token_accuracy": 0.8630481362342834, + "num_tokens": 98263686.0, + "step": 2528 + }, + { + "epoch": 0.3217147945554001, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.834766387939453e-06, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.6259894371032715, + "learning_rate": 1e-06, + "loss": 0.409, + "mean_token_accuracy": 0.8634606599807739, + "num_tokens": 98295583.0, + "step": 2529 + }, + { + "epoch": 0.3218420048339906, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.590800762176514, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8750566840171814, + "num_tokens": 98330038.0, + "step": 2530 + }, + { + "epoch": 0.3219692151125811, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.487833499908447, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8784090280532837, + "num_tokens": 98371253.0, + "step": 2531 + }, + { + "epoch": 0.3220964253911716, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.548442363739014, + "learning_rate": 1e-06, + "loss": 0.3724, + "mean_token_accuracy": 0.878692626953125, + "num_tokens": 98412779.0, + "step": 2532 + }, + { + "epoch": 0.32222363566976214, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.585991382598877, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.870633602142334, + "num_tokens": 98452940.0, + "step": 2533 + }, + { + "epoch": 0.3223508459483526, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.703675746917725, + "learning_rate": 1e-06, + "loss": 0.4095, + "mean_token_accuracy": 0.8685876131057739, + "num_tokens": 98486933.0, + "step": 2534 + }, + { + "epoch": 0.32247805622694314, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.588365077972412, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.8728238344192505, + "num_tokens": 98522058.0, + "step": 2535 + }, + { + "epoch": 0.32260526650553367, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.674785614013672, + "learning_rate": 1e-06, + "loss": 0.4108, + "mean_token_accuracy": 0.8678942322731018, + "num_tokens": 98556498.0, + "step": 2536 + }, + { + "epoch": 0.32273247678412414, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.517261981964111, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8703749179840088, + "num_tokens": 98594553.0, + "step": 2537 + }, + { + "epoch": 0.3228596870627147, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.496396541595459, + "learning_rate": 1e-06, + "loss": 0.3969, + "mean_token_accuracy": 0.8653783202171326, + "num_tokens": 98635234.0, + "step": 2538 + }, + { + "epoch": 0.3229868973413052, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.587076663970947, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8738930225372314, + "num_tokens": 98674359.0, + "step": 2539 + }, + { + "epoch": 0.3231141076198957, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.6237592697143555, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.8664442300796509, + "num_tokens": 98709914.0, + "step": 2540 + }, + { + "epoch": 0.3232413178984862, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.486352920532227, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8915165662765503, + "num_tokens": 98746656.0, + "step": 2541 + }, + { + "epoch": 0.32336852817707673, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.531822681427002, + "learning_rate": 1e-06, + "loss": 0.3923, + "mean_token_accuracy": 0.8682360649108887, + "num_tokens": 98785244.0, + "step": 2542 + }, + { + "epoch": 0.3234957384556672, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 9.894371032714844e-06, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.5626220703125, + "learning_rate": 1e-06, + "loss": 0.4044, + "mean_token_accuracy": 0.866996705532074, + "num_tokens": 98827498.0, + "step": 2543 + }, + { + "epoch": 0.32362294873425773, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.0013580322265625e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.561588764190674, + "learning_rate": 1e-06, + "loss": 0.3937, + "mean_token_accuracy": 0.8647831082344055, + "num_tokens": 98865251.0, + "step": 2544 + }, + { + "epoch": 0.32375015901284826, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0013580322265625e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.624792575836182, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8738698363304138, + "num_tokens": 98900682.0, + "step": 2545 + }, + { + "epoch": 0.32387736929143873, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0013580322265625e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.7858405113220215, + "learning_rate": 1e-06, + "loss": 0.4226, + "mean_token_accuracy": 0.85840904712677, + "num_tokens": 98930500.0, + "step": 2546 + }, + { + "epoch": 0.32400457957002926, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0013580322265625e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.513503074645996, + "learning_rate": 1e-06, + "loss": 0.3908, + "mean_token_accuracy": 0.8704754710197449, + "num_tokens": 98975114.0, + "step": 2547 + }, + { + "epoch": 0.3241317898486198, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 1.0073184967041016e-05, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.58216667175293, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8720370531082153, + "num_tokens": 99014544.0, + "step": 2548 + }, + { + "epoch": 0.32425900012721026, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0073184967041016e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.7464776039123535, + "learning_rate": 1e-06, + "loss": 0.4273, + "mean_token_accuracy": 0.8548153042793274, + "num_tokens": 99046737.0, + "step": 2549 + }, + { + "epoch": 0.3243862104058008, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0132789611816406e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.493777275085449, + "learning_rate": 1e-06, + "loss": 0.3729, + "mean_token_accuracy": 0.8763061761856079, + "num_tokens": 99086107.0, + "step": 2550 + }, + { + "epoch": 0.3245134206843913, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0132789611816406e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.88053035736084, + "learning_rate": 1e-06, + "loss": 0.3869, + "mean_token_accuracy": 0.8680364489555359, + "num_tokens": 99126961.0, + "step": 2551 + }, + { + "epoch": 0.3246406309629818, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0132789611816406e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.6043853759765625, + "learning_rate": 1e-06, + "loss": 0.3753, + "mean_token_accuracy": 0.8721059560775757, + "num_tokens": 99160157.0, + "step": 2552 + }, + { + "epoch": 0.3247678412415723, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.0132789611816406e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.5212602615356445, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8779325485229492, + "num_tokens": 99199206.0, + "step": 2553 + }, + { + "epoch": 0.32489505152016285, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0132789611816406e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.514306545257568, + "learning_rate": 1e-06, + "loss": 0.3959, + "mean_token_accuracy": 0.8663221597671509, + "num_tokens": 99240391.0, + "step": 2554 + }, + { + "epoch": 0.3250222617987533, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0132789611816406e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.533328056335449, + "learning_rate": 1e-06, + "loss": 0.4609, + "mean_token_accuracy": 0.8459266424179077, + "num_tokens": 99280975.0, + "step": 2555 + }, + { + "epoch": 0.32514947207734385, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0192394256591797e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.536372661590576, + "learning_rate": 1e-06, + "loss": 0.3716, + "mean_token_accuracy": 0.8739777207374573, + "num_tokens": 99320253.0, + "step": 2556 + }, + { + "epoch": 0.3252766823559344, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0192394256591797e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.427249431610107, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8879112005233765, + "num_tokens": 99361541.0, + "step": 2557 + }, + { + "epoch": 0.32540389263452485, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0192394256591797e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.540071964263916, + "learning_rate": 1e-06, + "loss": 0.4143, + "mean_token_accuracy": 0.8633575439453125, + "num_tokens": 99405249.0, + "step": 2558 + }, + { + "epoch": 0.3255311029131154, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0192394256591797e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.653625965118408, + "learning_rate": 1e-06, + "loss": 0.4263, + "mean_token_accuracy": 0.8589763641357422, + "num_tokens": 99439797.0, + "step": 2559 + }, + { + "epoch": 0.3256583131917059, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.0192394256591797e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.585277080535889, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8650252223014832, + "num_tokens": 99481169.0, + "step": 2560 + }, + { + "epoch": 0.3257855234702964, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0192394256591797e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.513261795043945, + "learning_rate": 1e-06, + "loss": 0.3972, + "mean_token_accuracy": 0.8648650050163269, + "num_tokens": 99521356.0, + "step": 2561 + }, + { + "epoch": 0.3259127337488869, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0311603546142578e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.550704002380371, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8700301051139832, + "num_tokens": 99559166.0, + "step": 2562 + }, + { + "epoch": 0.32603994402747744, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0311603546142578e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.507277965545654, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.883060872554779, + "num_tokens": 99600340.0, + "step": 2563 + }, + { + "epoch": 0.3261671543060679, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0311603546142578e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.4798736572265625, + "learning_rate": 1e-06, + "loss": 0.3905, + "mean_token_accuracy": 0.8683741092681885, + "num_tokens": 99645019.0, + "step": 2564 + }, + { + "epoch": 0.32629436458465844, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0311603546142578e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.570505619049072, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8729000687599182, + "num_tokens": 99683665.0, + "step": 2565 + }, + { + "epoch": 0.32642157486324896, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0311603546142578e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.712043762207031, + "learning_rate": 1e-06, + "loss": 0.395, + "mean_token_accuracy": 0.8668538331985474, + "num_tokens": 99723170.0, + "step": 2566 + }, + { + "epoch": 0.32654878514183944, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.56054162979126, + "learning_rate": 1e-06, + "loss": 0.3717, + "mean_token_accuracy": 0.8727425932884216, + "num_tokens": 99763064.0, + "step": 2567 + }, + { + "epoch": 0.32667599542042997, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.510185241699219e-06, + "grad_norm": 7.542374134063721, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8790431618690491, + "num_tokens": 99798808.0, + "step": 2568 + }, + { + "epoch": 0.3268032056990205, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.581881999969482, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8763419985771179, + "num_tokens": 99838160.0, + "step": 2569 + }, + { + "epoch": 0.32693041597761097, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.605208396911621, + "learning_rate": 1e-06, + "loss": 0.3913, + "mean_token_accuracy": 0.8672505021095276, + "num_tokens": 99873630.0, + "step": 2570 + }, + { + "epoch": 0.3270576262562015, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.630815029144287, + "learning_rate": 1e-06, + "loss": 0.4387, + "mean_token_accuracy": 0.8546981811523438, + "num_tokens": 99914570.0, + "step": 2571 + }, + { + "epoch": 0.327184836534792, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.521724224090576, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.8781139850616455, + "num_tokens": 99954941.0, + "step": 2572 + }, + { + "epoch": 0.3273120468133825, + "ewc_loss": 0.008544921875, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.539987564086914e-06, + "grad_norm": 7.502418041229248, + "learning_rate": 1e-06, + "loss": 0.4114, + "mean_token_accuracy": 0.8615571856498718, + "num_tokens": 100000594.0, + "step": 2573 + }, + { + "epoch": 0.327439257091973, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.570131301879883, + "learning_rate": 1e-06, + "loss": 0.4317, + "mean_token_accuracy": 0.8596502542495728, + "num_tokens": 100043999.0, + "step": 2574 + }, + { + "epoch": 0.32756646737056355, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.51523494720459, + "learning_rate": 1e-06, + "loss": 0.4045, + "mean_token_accuracy": 0.8702216148376465, + "num_tokens": 100086411.0, + "step": 2575 + }, + { + "epoch": 0.327693677649154, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.643556594848633, + "learning_rate": 1e-06, + "loss": 0.415, + "mean_token_accuracy": 0.8619162440299988, + "num_tokens": 100124889.0, + "step": 2576 + }, + { + "epoch": 0.32782088792774455, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.546497344970703, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8763239979743958, + "num_tokens": 100161089.0, + "step": 2577 + }, + { + "epoch": 0.3279480982063351, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.528654098510742, + "learning_rate": 1e-06, + "loss": 0.4236, + "mean_token_accuracy": 0.857375979423523, + "num_tokens": 100203509.0, + "step": 2578 + }, + { + "epoch": 0.3280753084849256, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.585977077484131, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8770535588264465, + "num_tokens": 100241472.0, + "step": 2579 + }, + { + "epoch": 0.3282025187635161, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.6602983474731445, + "learning_rate": 1e-06, + "loss": 0.415, + "mean_token_accuracy": 0.8635198473930359, + "num_tokens": 100276596.0, + "step": 2580 + }, + { + "epoch": 0.3283297290421066, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.558976173400879, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8827617764472961, + "num_tokens": 100313227.0, + "step": 2581 + }, + { + "epoch": 0.32845693932069714, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0371208190917969e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.595657825469971, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.877097487449646, + "num_tokens": 100345757.0, + "step": 2582 + }, + { + "epoch": 0.3285841495992876, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.58703088760376, + "learning_rate": 1e-06, + "loss": 0.4245, + "mean_token_accuracy": 0.8581898808479309, + "num_tokens": 100387641.0, + "step": 2583 + }, + { + "epoch": 0.32871135987787814, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.409019470214844, + "learning_rate": 1e-06, + "loss": 0.3732, + "mean_token_accuracy": 0.8747061491012573, + "num_tokens": 100432491.0, + "step": 2584 + }, + { + "epoch": 0.32883857015646867, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.552314758300781, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.874894380569458, + "num_tokens": 100472106.0, + "step": 2585 + }, + { + "epoch": 0.32896578043505914, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.508569717407227, + "learning_rate": 1e-06, + "loss": 0.4024, + "mean_token_accuracy": 0.8637256622314453, + "num_tokens": 100514143.0, + "step": 2586 + }, + { + "epoch": 0.32909299071364967, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.617504596710205, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8731189966201782, + "num_tokens": 100549687.0, + "step": 2587 + }, + { + "epoch": 0.3292202009922402, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.586958408355713, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8793472051620483, + "num_tokens": 100586818.0, + "step": 2588 + }, + { + "epoch": 0.32934741127083067, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.557088375091553, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8731950521469116, + "num_tokens": 100625823.0, + "step": 2589 + }, + { + "epoch": 0.3294746215494212, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.568978786468506, + "learning_rate": 1e-06, + "loss": 0.4162, + "mean_token_accuracy": 0.8619479537010193, + "num_tokens": 100664674.0, + "step": 2590 + }, + { + "epoch": 0.3296018318280117, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.583780765533447, + "learning_rate": 1e-06, + "loss": 0.4602, + "mean_token_accuracy": 0.8491166830062866, + "num_tokens": 100704958.0, + "step": 2591 + }, + { + "epoch": 0.3297290421066022, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.5419721603393555, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8804278373718262, + "num_tokens": 100741430.0, + "step": 2592 + }, + { + "epoch": 0.3298562523851927, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.470602035522461, + "learning_rate": 1e-06, + "loss": 0.4024, + "mean_token_accuracy": 0.8643832206726074, + "num_tokens": 100784400.0, + "step": 2593 + }, + { + "epoch": 0.32998346266378326, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.599592208862305e-06, + "grad_norm": 7.5131964683532715, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8689631223678589, + "num_tokens": 100825623.0, + "step": 2594 + }, + { + "epoch": 0.33011067294237373, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.535388946533203, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8869863748550415, + "num_tokens": 100865865.0, + "step": 2595 + }, + { + "epoch": 0.33023788322096426, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.504837989807129, + "learning_rate": 1e-06, + "loss": 0.3993, + "mean_token_accuracy": 0.8661288022994995, + "num_tokens": 100909884.0, + "step": 2596 + }, + { + "epoch": 0.3303650934995548, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.594959735870361, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8674635291099548, + "num_tokens": 100946391.0, + "step": 2597 + }, + { + "epoch": 0.33049230377814526, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.4739532470703125, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.8757014870643616, + "num_tokens": 100989173.0, + "step": 2598 + }, + { + "epoch": 0.3306195140567358, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.5170464515686035, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8782679438591003, + "num_tokens": 101032271.0, + "step": 2599 + }, + { + "epoch": 0.3307467243353263, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.6402587890625, + "learning_rate": 1e-06, + "loss": 0.4073, + "mean_token_accuracy": 0.8641906976699829, + "num_tokens": 101068626.0, + "step": 2600 + }, + { + "epoch": 0.3308739346139168, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.585248947143555, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8762795925140381, + "num_tokens": 101105456.0, + "step": 2601 + }, + { + "epoch": 0.3310011448925073, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.662184238433838, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.876117467880249, + "num_tokens": 101141057.0, + "step": 2602 + }, + { + "epoch": 0.33112835517109784, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.530699729919434, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.8737056255340576, + "num_tokens": 101182296.0, + "step": 2603 + }, + { + "epoch": 0.3312555654496883, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.459941864013672, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.871234655380249, + "num_tokens": 101229841.0, + "step": 2604 + }, + { + "epoch": 0.33138277572827884, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.043081283569336e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.683436393737793, + "learning_rate": 1e-06, + "loss": 0.4252, + "mean_token_accuracy": 0.8602399826049805, + "num_tokens": 101267268.0, + "step": 2605 + }, + { + "epoch": 0.3315099860068694, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.049041748046875e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.577578544616699, + "learning_rate": 1e-06, + "loss": 0.3757, + "mean_token_accuracy": 0.8738489151000977, + "num_tokens": 101303961.0, + "step": 2606 + }, + { + "epoch": 0.33163719628545985, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.049041748046875e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.680794715881348, + "learning_rate": 1e-06, + "loss": 0.3964, + "mean_token_accuracy": 0.8687471747398376, + "num_tokens": 101339300.0, + "step": 2607 + }, + { + "epoch": 0.3317644065640504, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.049041748046875e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.532513618469238, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8738548159599304, + "num_tokens": 101377719.0, + "step": 2608 + }, + { + "epoch": 0.3318916168426409, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.049041748046875e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.565467357635498, + "learning_rate": 1e-06, + "loss": 0.45, + "mean_token_accuracy": 0.8498971462249756, + "num_tokens": 101420083.0, + "step": 2609 + }, + { + "epoch": 0.3320188271212314, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.049041748046875e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.6525349617004395, + "learning_rate": 1e-06, + "loss": 0.4109, + "mean_token_accuracy": 0.8641234636306763, + "num_tokens": 101459247.0, + "step": 2610 + }, + { + "epoch": 0.3321460373998219, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.055002212524414e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.576632499694824, + "learning_rate": 1e-06, + "loss": 0.3858, + "mean_token_accuracy": 0.8690061569213867, + "num_tokens": 101496125.0, + "step": 2611 + }, + { + "epoch": 0.33227324767841243, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.055002212524414e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.7236857414245605, + "learning_rate": 1e-06, + "loss": 0.3834, + "mean_token_accuracy": 0.872252881526947, + "num_tokens": 101527143.0, + "step": 2612 + }, + { + "epoch": 0.3324004579570029, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.055002212524414e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.540499210357666, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8749127388000488, + "num_tokens": 101569263.0, + "step": 2613 + }, + { + "epoch": 0.33252766823559343, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.055002212524414e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.474708080291748, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8848528265953064, + "num_tokens": 101608880.0, + "step": 2614 + }, + { + "epoch": 0.33265487851418396, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.055002212524414e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.557743072509766, + "learning_rate": 1e-06, + "loss": 0.403, + "mean_token_accuracy": 0.8665643930435181, + "num_tokens": 101651771.0, + "step": 2615 + }, + { + "epoch": 0.33278208879277443, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.055002212524414e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.523478031158447, + "learning_rate": 1e-06, + "loss": 0.3943, + "mean_token_accuracy": 0.8676626086235046, + "num_tokens": 101695051.0, + "step": 2616 + }, + { + "epoch": 0.33290929907136496, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.055002212524414e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.601721286773682, + "learning_rate": 1e-06, + "loss": 0.3853, + "mean_token_accuracy": 0.871306300163269, + "num_tokens": 101734377.0, + "step": 2617 + }, + { + "epoch": 0.3330365093499555, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.055002212524414e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.554170608520508, + "learning_rate": 1e-06, + "loss": 0.401, + "mean_token_accuracy": 0.8633938431739807, + "num_tokens": 101779352.0, + "step": 2618 + }, + { + "epoch": 0.33316371962854596, + "ewc_loss": 0.00860595703125, + "ewc_loss_diag": 1.055002212524414e-05, + "ewc_loss_parallel": 7.569789886474609e-06, + "grad_norm": 7.6190619468688965, + "learning_rate": 1e-06, + "loss": 0.3995, + "mean_token_accuracy": 0.8664408326148987, + "num_tokens": 101815850.0, + "step": 2619 + }, + { + "epoch": 0.3332909299071365, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.055002212524414e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.55398416519165, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8739746809005737, + "num_tokens": 101855435.0, + "step": 2620 + }, + { + "epoch": 0.333418140185727, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.055002212524414e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.751657009124756, + "learning_rate": 1e-06, + "loss": 0.4535, + "mean_token_accuracy": 0.8457505702972412, + "num_tokens": 101887581.0, + "step": 2621 + }, + { + "epoch": 0.3335453504643175, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.055002212524414e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.511294364929199, + "learning_rate": 1e-06, + "loss": 0.3933, + "mean_token_accuracy": 0.8646544814109802, + "num_tokens": 101925777.0, + "step": 2622 + }, + { + "epoch": 0.333672560742908, + "ewc_loss": 0.0086669921875, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.62939453125e-06, + "grad_norm": 7.567924499511719, + "learning_rate": 1e-06, + "loss": 0.4104, + "mean_token_accuracy": 0.8642420768737793, + "num_tokens": 101963346.0, + "step": 2623 + }, + { + "epoch": 0.33379977102149855, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.5427632331848145, + "learning_rate": 1e-06, + "loss": 0.4168, + "mean_token_accuracy": 0.8626515865325928, + "num_tokens": 102003368.0, + "step": 2624 + }, + { + "epoch": 0.333926981300089, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.6844868659973145, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8779536485671997, + "num_tokens": 102040048.0, + "step": 2625 + }, + { + "epoch": 0.33405419157867955, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.730281829833984, + "learning_rate": 1e-06, + "loss": 0.4074, + "mean_token_accuracy": 0.8596675395965576, + "num_tokens": 102072045.0, + "step": 2626 + }, + { + "epoch": 0.3341814018572701, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.675539970397949, + "learning_rate": 1e-06, + "loss": 0.4066, + "mean_token_accuracy": 0.8636436462402344, + "num_tokens": 102105874.0, + "step": 2627 + }, + { + "epoch": 0.33430861213586055, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.731113433837891, + "learning_rate": 1e-06, + "loss": 0.4401, + "mean_token_accuracy": 0.8543220162391663, + "num_tokens": 102140550.0, + "step": 2628 + }, + { + "epoch": 0.3344358224144511, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.6144843101501465, + "learning_rate": 1e-06, + "loss": 0.4565, + "mean_token_accuracy": 0.8493676781654358, + "num_tokens": 102180960.0, + "step": 2629 + }, + { + "epoch": 0.3345630326930416, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.581299781799316, + "learning_rate": 1e-06, + "loss": 0.4004, + "mean_token_accuracy": 0.8614577054977417, + "num_tokens": 102215862.0, + "step": 2630 + }, + { + "epoch": 0.33469024297163213, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.481729984283447, + "learning_rate": 1e-06, + "loss": 0.4194, + "mean_token_accuracy": 0.8610286116600037, + "num_tokens": 102262797.0, + "step": 2631 + }, + { + "epoch": 0.3348174532502226, + "ewc_loss": 0.00872802734375, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.48825740814209, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.8795326948165894, + "num_tokens": 102306364.0, + "step": 2632 + }, + { + "epoch": 0.33494466352881314, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.599786281585693, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8727888464927673, + "num_tokens": 102348690.0, + "step": 2633 + }, + { + "epoch": 0.33507187380740366, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.659467697143555, + "learning_rate": 1e-06, + "loss": 0.4028, + "mean_token_accuracy": 0.8648726940155029, + "num_tokens": 102383997.0, + "step": 2634 + }, + { + "epoch": 0.33519908408599414, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.648813724517822, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8751769065856934, + "num_tokens": 102416547.0, + "step": 2635 + }, + { + "epoch": 0.33532629436458466, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.62302303314209, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8765127658843994, + "num_tokens": 102453462.0, + "step": 2636 + }, + { + "epoch": 0.3354535046431752, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.573343753814697, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8727591037750244, + "num_tokens": 102494103.0, + "step": 2637 + }, + { + "epoch": 0.33558071492176567, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.482504367828369, + "learning_rate": 1e-06, + "loss": 0.3773, + "mean_token_accuracy": 0.8749690651893616, + "num_tokens": 102536897.0, + "step": 2638 + }, + { + "epoch": 0.3357079252003562, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.607448577880859, + "learning_rate": 1e-06, + "loss": 0.432, + "mean_token_accuracy": 0.8541440367698669, + "num_tokens": 102578310.0, + "step": 2639 + }, + { + "epoch": 0.3358351354789467, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.658758640289307, + "learning_rate": 1e-06, + "loss": 0.4311, + "mean_token_accuracy": 0.8581652641296387, + "num_tokens": 102616846.0, + "step": 2640 + }, + { + "epoch": 0.3359623457575372, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.734246730804443, + "learning_rate": 1e-06, + "loss": 0.4157, + "mean_token_accuracy": 0.8630906343460083, + "num_tokens": 102648006.0, + "step": 2641 + }, + { + "epoch": 0.3360895560361277, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.620901107788086, + "learning_rate": 1e-06, + "loss": 0.4187, + "mean_token_accuracy": 0.8581463694572449, + "num_tokens": 102688933.0, + "step": 2642 + }, + { + "epoch": 0.33621676631471825, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0609626770019531e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.6570658683776855, + "learning_rate": 1e-06, + "loss": 0.454, + "mean_token_accuracy": 0.8469760417938232, + "num_tokens": 102724163.0, + "step": 2643 + }, + { + "epoch": 0.3363439765933087, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.555212497711182, + "learning_rate": 1e-06, + "loss": 0.3885, + "mean_token_accuracy": 0.8722062706947327, + "num_tokens": 102766271.0, + "step": 2644 + }, + { + "epoch": 0.33647118687189925, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.61129093170166, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8825510740280151, + "num_tokens": 102808265.0, + "step": 2645 + }, + { + "epoch": 0.3365983971504898, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.624608039855957, + "learning_rate": 1e-06, + "loss": 0.396, + "mean_token_accuracy": 0.8648508787155151, + "num_tokens": 102850040.0, + "step": 2646 + }, + { + "epoch": 0.33672560742908025, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.637430667877197, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.873191237449646, + "num_tokens": 102884445.0, + "step": 2647 + }, + { + "epoch": 0.3368528177076708, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.499728202819824, + "learning_rate": 1e-06, + "loss": 0.3732, + "mean_token_accuracy": 0.8765681385993958, + "num_tokens": 102929555.0, + "step": 2648 + }, + { + "epoch": 0.3369800279862613, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.525660514831543, + "learning_rate": 1e-06, + "loss": 0.3823, + "mean_token_accuracy": 0.8724027276039124, + "num_tokens": 102972091.0, + "step": 2649 + }, + { + "epoch": 0.3371072382648518, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.6960649490356445, + "learning_rate": 1e-06, + "loss": 0.3959, + "mean_token_accuracy": 0.8669700026512146, + "num_tokens": 103009185.0, + "step": 2650 + }, + { + "epoch": 0.3372344485434423, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.533579349517822, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8838945627212524, + "num_tokens": 103047501.0, + "step": 2651 + }, + { + "epoch": 0.33736165882203284, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.539438247680664, + "learning_rate": 1e-06, + "loss": 0.4025, + "mean_token_accuracy": 0.8643026947975159, + "num_tokens": 103092254.0, + "step": 2652 + }, + { + "epoch": 0.3374888691006233, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.595415115356445, + "learning_rate": 1e-06, + "loss": 0.4275, + "mean_token_accuracy": 0.8591341972351074, + "num_tokens": 103134129.0, + "step": 2653 + }, + { + "epoch": 0.33761607937921384, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.7708563804626465, + "learning_rate": 1e-06, + "loss": 0.4515, + "mean_token_accuracy": 0.8464423418045044, + "num_tokens": 103169299.0, + "step": 2654 + }, + { + "epoch": 0.33774328965780437, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.504724502563477, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8780738115310669, + "num_tokens": 103209109.0, + "step": 2655 + }, + { + "epoch": 0.33787049993639484, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.66405725479126, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.874073326587677, + "num_tokens": 103244433.0, + "step": 2656 + }, + { + "epoch": 0.33799771021498537, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.746828556060791, + "learning_rate": 1e-06, + "loss": 0.4349, + "mean_token_accuracy": 0.8573893308639526, + "num_tokens": 103278733.0, + "step": 2657 + }, + { + "epoch": 0.3381249204935759, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.669604778289795, + "learning_rate": 1e-06, + "loss": 0.4068, + "mean_token_accuracy": 0.8638922572135925, + "num_tokens": 103316377.0, + "step": 2658 + }, + { + "epoch": 0.33825213077216637, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.695592880249023, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.8678833246231079, + "num_tokens": 103348221.0, + "step": 2659 + }, + { + "epoch": 0.3383793410507569, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.512617588043213, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8857741355895996, + "num_tokens": 103385909.0, + "step": 2660 + }, + { + "epoch": 0.3385065513293474, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0669231414794922e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.44778299331665, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8869094848632812, + "num_tokens": 103426257.0, + "step": 2661 + }, + { + "epoch": 0.3386337616079379, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0728836059570312e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.728811740875244, + "learning_rate": 1e-06, + "loss": 0.4295, + "mean_token_accuracy": 0.8573477864265442, + "num_tokens": 103461713.0, + "step": 2662 + }, + { + "epoch": 0.33876097188652843, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0728836059570312e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.5597028732299805, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8838735818862915, + "num_tokens": 103500483.0, + "step": 2663 + }, + { + "epoch": 0.33888818216511896, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0728836059570312e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.524056911468506, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8817573189735413, + "num_tokens": 103540469.0, + "step": 2664 + }, + { + "epoch": 0.33901539244370943, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0728836059570312e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.523318767547607, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.8810907602310181, + "num_tokens": 103582123.0, + "step": 2665 + }, + { + "epoch": 0.33914260272229996, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0728836059570312e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.561904430389404, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8752709627151489, + "num_tokens": 103620813.0, + "step": 2666 + }, + { + "epoch": 0.3392698130008905, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0728836059570312e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.543495178222656, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8767654299736023, + "num_tokens": 103661714.0, + "step": 2667 + }, + { + "epoch": 0.33939702327948096, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0728836059570312e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.6753034591674805, + "learning_rate": 1e-06, + "loss": 0.3814, + "mean_token_accuracy": 0.8691225051879883, + "num_tokens": 103695426.0, + "step": 2668 + }, + { + "epoch": 0.3395242335580715, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0728836059570312e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.58280611038208, + "learning_rate": 1e-06, + "loss": 0.4259, + "mean_token_accuracy": 0.8615080118179321, + "num_tokens": 103737028.0, + "step": 2669 + }, + { + "epoch": 0.339651443836662, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0728836059570312e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.703315258026123, + "learning_rate": 1e-06, + "loss": 0.4439, + "mean_token_accuracy": 0.851719856262207, + "num_tokens": 103773485.0, + "step": 2670 + }, + { + "epoch": 0.3397786541152525, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0728836059570312e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.816741466522217, + "learning_rate": 1e-06, + "loss": 0.4209, + "mean_token_accuracy": 0.8578455448150635, + "num_tokens": 103804343.0, + "step": 2671 + }, + { + "epoch": 0.339905864393843, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0728836059570312e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.553274154663086, + "learning_rate": 1e-06, + "loss": 0.3954, + "mean_token_accuracy": 0.87312912940979, + "num_tokens": 103843093.0, + "step": 2672 + }, + { + "epoch": 0.34003307467243354, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0788440704345703e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.764373779296875, + "learning_rate": 1e-06, + "loss": 0.4242, + "mean_token_accuracy": 0.8583660125732422, + "num_tokens": 103874558.0, + "step": 2673 + }, + { + "epoch": 0.340160284951024, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0788440704345703e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.581416130065918, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8751459717750549, + "num_tokens": 103909084.0, + "step": 2674 + }, + { + "epoch": 0.34028749522961454, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0788440704345703e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.517902851104736, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8759835362434387, + "num_tokens": 103954971.0, + "step": 2675 + }, + { + "epoch": 0.3404147055082051, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0788440704345703e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.6817402839660645, + "learning_rate": 1e-06, + "loss": 0.4444, + "mean_token_accuracy": 0.8570122718811035, + "num_tokens": 103994974.0, + "step": 2676 + }, + { + "epoch": 0.34054191578679555, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.605978012084961, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.877485990524292, + "num_tokens": 104031644.0, + "step": 2677 + }, + { + "epoch": 0.3406691260653861, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.627417087554932, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.8730522394180298, + "num_tokens": 104071744.0, + "step": 2678 + }, + { + "epoch": 0.3407963363439766, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.697822570800781, + "learning_rate": 1e-06, + "loss": 0.4541, + "mean_token_accuracy": 0.8514183163642883, + "num_tokens": 104109606.0, + "step": 2679 + }, + { + "epoch": 0.34092354662256713, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.642206192016602, + "learning_rate": 1e-06, + "loss": 0.3788, + "mean_token_accuracy": 0.8728987574577332, + "num_tokens": 104147481.0, + "step": 2680 + }, + { + "epoch": 0.3410507569011576, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.624407768249512, + "learning_rate": 1e-06, + "loss": 0.4251, + "mean_token_accuracy": 0.8562802076339722, + "num_tokens": 104184605.0, + "step": 2681 + }, + { + "epoch": 0.34117796717974813, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.502374172210693, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8867009878158569, + "num_tokens": 104225340.0, + "step": 2682 + }, + { + "epoch": 0.34130517745833866, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0788440704345703e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.561455726623535, + "learning_rate": 1e-06, + "loss": 0.3761, + "mean_token_accuracy": 0.8723883628845215, + "num_tokens": 104267427.0, + "step": 2683 + }, + { + "epoch": 0.34143238773692913, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.582569599151611, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.875115692615509, + "num_tokens": 104307202.0, + "step": 2684 + }, + { + "epoch": 0.34155959801551966, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.647529125213623, + "learning_rate": 1e-06, + "loss": 0.3981, + "mean_token_accuracy": 0.8693771958351135, + "num_tokens": 104341930.0, + "step": 2685 + }, + { + "epoch": 0.3416868082941102, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.609391212463379, + "learning_rate": 1e-06, + "loss": 0.3925, + "mean_token_accuracy": 0.8652248382568359, + "num_tokens": 104382340.0, + "step": 2686 + }, + { + "epoch": 0.34181401857270066, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.605303764343262, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8736619353294373, + "num_tokens": 104418939.0, + "step": 2687 + }, + { + "epoch": 0.3419412288512912, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.532474517822266, + "learning_rate": 1e-06, + "loss": 0.3907, + "mean_token_accuracy": 0.8684887886047363, + "num_tokens": 104466154.0, + "step": 2688 + }, + { + "epoch": 0.3420684391298817, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.527118682861328, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8759481906890869, + "num_tokens": 104507087.0, + "step": 2689 + }, + { + "epoch": 0.3421956494084722, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.60673189163208, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.875276505947113, + "num_tokens": 104540767.0, + "step": 2690 + }, + { + "epoch": 0.3423228596870627, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.700962543487549, + "learning_rate": 1e-06, + "loss": 0.4069, + "mean_token_accuracy": 0.8640363812446594, + "num_tokens": 104578778.0, + "step": 2691 + }, + { + "epoch": 0.34245006996565325, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.518406391143799, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8685904741287231, + "num_tokens": 104619753.0, + "step": 2692 + }, + { + "epoch": 0.3425772802442437, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.592108249664307, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8728994727134705, + "num_tokens": 104659852.0, + "step": 2693 + }, + { + "epoch": 0.34270449052283425, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.503115177154541, + "learning_rate": 1e-06, + "loss": 0.3848, + "mean_token_accuracy": 0.8708961606025696, + "num_tokens": 104704760.0, + "step": 2694 + }, + { + "epoch": 0.3428317008014248, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.619411945343018, + "learning_rate": 1e-06, + "loss": 0.3961, + "mean_token_accuracy": 0.8666502237319946, + "num_tokens": 104743137.0, + "step": 2695 + }, + { + "epoch": 0.34295891108001525, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.620064735412598, + "learning_rate": 1e-06, + "loss": 0.3796, + "mean_token_accuracy": 0.877292275428772, + "num_tokens": 104780838.0, + "step": 2696 + }, + { + "epoch": 0.3430861213586058, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0848045349121094e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.584505081176758, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8774021863937378, + "num_tokens": 104817233.0, + "step": 2697 + }, + { + "epoch": 0.3432133316371963, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0907649993896484e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.594371795654297, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8791037797927856, + "num_tokens": 104853374.0, + "step": 2698 + }, + { + "epoch": 0.3433405419157868, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0907649993896484e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.585534572601318, + "learning_rate": 1e-06, + "loss": 0.3689, + "mean_token_accuracy": 0.8760519027709961, + "num_tokens": 104894733.0, + "step": 2699 + }, + { + "epoch": 0.3434677521943773, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0907649993896484e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.6534318923950195, + "learning_rate": 1e-06, + "loss": 0.4113, + "mean_token_accuracy": 0.865561306476593, + "num_tokens": 104933621.0, + "step": 2700 + }, + { + "epoch": 0.34359496247296784, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0907649993896484e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.552499771118164, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8749390840530396, + "num_tokens": 104974310.0, + "step": 2701 + }, + { + "epoch": 0.3437221727515583, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.0907649993896484e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.591403961181641, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.8696312308311462, + "num_tokens": 105014121.0, + "step": 2702 + }, + { + "epoch": 0.34384938303014884, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0907649993896484e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.742321968078613, + "learning_rate": 1e-06, + "loss": 0.3968, + "mean_token_accuracy": 0.8664556741714478, + "num_tokens": 105046878.0, + "step": 2703 + }, + { + "epoch": 0.34397659330873936, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0907649993896484e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.652307510375977, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8705271482467651, + "num_tokens": 105082008.0, + "step": 2704 + }, + { + "epoch": 0.34410380358732984, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0907649993896484e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.591411113739014, + "learning_rate": 1e-06, + "loss": 0.4026, + "mean_token_accuracy": 0.8633947968482971, + "num_tokens": 105123393.0, + "step": 2705 + }, + { + "epoch": 0.34423101386592037, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0907649993896484e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.5215744972229, + "learning_rate": 1e-06, + "loss": 0.4191, + "mean_token_accuracy": 0.8604246377944946, + "num_tokens": 105169268.0, + "step": 2706 + }, + { + "epoch": 0.3443582241445109, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0907649993896484e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.581902027130127, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8724560141563416, + "num_tokens": 105210054.0, + "step": 2707 + }, + { + "epoch": 0.34448543442310137, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0907649993896484e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.624118328094482, + "learning_rate": 1e-06, + "loss": 0.4323, + "mean_token_accuracy": 0.8542169332504272, + "num_tokens": 105251174.0, + "step": 2708 + }, + { + "epoch": 0.3446126447016919, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.0907649993896484e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.599056720733643, + "learning_rate": 1e-06, + "loss": 0.3959, + "mean_token_accuracy": 0.8668100833892822, + "num_tokens": 105292668.0, + "step": 2709 + }, + { + "epoch": 0.3447398549802824, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.0907649993896484e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.60264778137207, + "learning_rate": 1e-06, + "loss": 0.4187, + "mean_token_accuracy": 0.8606610298156738, + "num_tokens": 105336745.0, + "step": 2710 + }, + { + "epoch": 0.3448670652588729, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.1026859283447266e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.79030704498291, + "learning_rate": 1e-06, + "loss": 0.4039, + "mean_token_accuracy": 0.8672845363616943, + "num_tokens": 105372737.0, + "step": 2711 + }, + { + "epoch": 0.3449942755374634, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.1026859283447266e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.584246635437012, + "learning_rate": 1e-06, + "loss": 0.3667, + "mean_token_accuracy": 0.8765166997909546, + "num_tokens": 105413155.0, + "step": 2712 + }, + { + "epoch": 0.34512148581605395, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.1026859283447266e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.48798131942749, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8792983293533325, + "num_tokens": 105456038.0, + "step": 2713 + }, + { + "epoch": 0.3452486960946444, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.1026859283447266e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.705107688903809, + "learning_rate": 1e-06, + "loss": 0.3738, + "mean_token_accuracy": 0.8720338940620422, + "num_tokens": 105489254.0, + "step": 2714 + }, + { + "epoch": 0.34537590637323495, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.1026859283447266e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.546050548553467, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8790151476860046, + "num_tokens": 105530747.0, + "step": 2715 + }, + { + "epoch": 0.3455031166518255, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.1026859283447266e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.6592864990234375, + "learning_rate": 1e-06, + "loss": 0.4026, + "mean_token_accuracy": 0.8623263835906982, + "num_tokens": 105566192.0, + "step": 2716 + }, + { + "epoch": 0.34563032693041595, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.1026859283447266e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.632691383361816, + "learning_rate": 1e-06, + "loss": 0.4379, + "mean_token_accuracy": 0.8546448945999146, + "num_tokens": 105602088.0, + "step": 2717 + }, + { + "epoch": 0.3457575372090065, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.1026859283447266e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.5759663581848145, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8758077025413513, + "num_tokens": 105642112.0, + "step": 2718 + }, + { + "epoch": 0.345884747487597, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.1026859283447266e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.539854049682617, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8863275051116943, + "num_tokens": 105679937.0, + "step": 2719 + }, + { + "epoch": 0.3460119577661875, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.1026859283447266e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.556033611297607, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8832646608352661, + "num_tokens": 105717533.0, + "step": 2720 + }, + { + "epoch": 0.346139168044778, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.1026859283447266e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.586929798126221, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8883399367332458, + "num_tokens": 105750659.0, + "step": 2721 + }, + { + "epoch": 0.34626637832336854, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.1146068572998047e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.601329803466797, + "learning_rate": 1e-06, + "loss": 0.412, + "mean_token_accuracy": 0.863176167011261, + "num_tokens": 105789098.0, + "step": 2722 + }, + { + "epoch": 0.346393588601959, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1146068572998047e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.6351165771484375, + "learning_rate": 1e-06, + "loss": 0.4096, + "mean_token_accuracy": 0.8613196015357971, + "num_tokens": 105827102.0, + "step": 2723 + }, + { + "epoch": 0.34652079888054954, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.1146068572998047e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.51561164855957, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8799445033073425, + "num_tokens": 105863712.0, + "step": 2724 + }, + { + "epoch": 0.34664800915914007, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.1146068572998047e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.611454486846924, + "learning_rate": 1e-06, + "loss": 0.412, + "mean_token_accuracy": 0.8617314100265503, + "num_tokens": 105903979.0, + "step": 2725 + }, + { + "epoch": 0.34677521943773054, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.1146068572998047e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.680671215057373, + "learning_rate": 1e-06, + "loss": 0.3887, + "mean_token_accuracy": 0.8724627494812012, + "num_tokens": 105940803.0, + "step": 2726 + }, + { + "epoch": 0.34690242971632107, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.1146068572998047e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.561944007873535, + "learning_rate": 1e-06, + "loss": 0.4018, + "mean_token_accuracy": 0.8661465644836426, + "num_tokens": 105979669.0, + "step": 2727 + }, + { + "epoch": 0.3470296399949116, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.1146068572998047e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.552434921264648, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8743308782577515, + "num_tokens": 106022428.0, + "step": 2728 + }, + { + "epoch": 0.3471568502735021, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1146068572998047e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.642486095428467, + "learning_rate": 1e-06, + "loss": 0.385, + "mean_token_accuracy": 0.8688479065895081, + "num_tokens": 106059749.0, + "step": 2729 + }, + { + "epoch": 0.3472840605520926, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1146068572998047e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.556640148162842, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8774818778038025, + "num_tokens": 106098546.0, + "step": 2730 + }, + { + "epoch": 0.3474112708306831, + "ewc_loss": 0.0087890625, + "ewc_loss_diag": 1.1146068572998047e-05, + "ewc_loss_parallel": 7.68899917602539e-06, + "grad_norm": 7.674807071685791, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8827501535415649, + "num_tokens": 106133161.0, + "step": 2731 + }, + { + "epoch": 0.34753848110927366, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.1146068572998047e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.589765548706055, + "learning_rate": 1e-06, + "loss": 0.4202, + "mean_token_accuracy": 0.8590543866157532, + "num_tokens": 106169948.0, + "step": 2732 + }, + { + "epoch": 0.34766569138786413, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.716592311859131, + "learning_rate": 1e-06, + "loss": 0.3909, + "mean_token_accuracy": 0.87281733751297, + "num_tokens": 106207838.0, + "step": 2733 + }, + { + "epoch": 0.34779290166645466, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.5684099197387695, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8764468431472778, + "num_tokens": 106243297.0, + "step": 2734 + }, + { + "epoch": 0.3479201119450452, + "ewc_loss": 0.00885009765625, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.578939437866211, + "learning_rate": 1e-06, + "loss": 0.4214, + "mean_token_accuracy": 0.8607608079910278, + "num_tokens": 106282150.0, + "step": 2735 + }, + { + "epoch": 0.34804732222363566, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.637901306152344, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8742189407348633, + "num_tokens": 106319194.0, + "step": 2736 + }, + { + "epoch": 0.3481745325022262, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.591878890991211, + "learning_rate": 1e-06, + "loss": 0.4125, + "mean_token_accuracy": 0.8617151975631714, + "num_tokens": 106356775.0, + "step": 2737 + }, + { + "epoch": 0.3483017427808167, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.693885803222656, + "learning_rate": 1e-06, + "loss": 0.4048, + "mean_token_accuracy": 0.8656759262084961, + "num_tokens": 106398616.0, + "step": 2738 + }, + { + "epoch": 0.3484289530594072, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.650910377502441, + "learning_rate": 1e-06, + "loss": 0.4276, + "mean_token_accuracy": 0.8554842472076416, + "num_tokens": 106438804.0, + "step": 2739 + }, + { + "epoch": 0.3485561633379977, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.70546293258667, + "learning_rate": 1e-06, + "loss": 0.4037, + "mean_token_accuracy": 0.8679620027542114, + "num_tokens": 106483627.0, + "step": 2740 + }, + { + "epoch": 0.34868337361658824, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.495845794677734, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8721256256103516, + "num_tokens": 106521634.0, + "step": 2741 + }, + { + "epoch": 0.3488105838951787, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.607360363006592, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.879959225654602, + "num_tokens": 106557540.0, + "step": 2742 + }, + { + "epoch": 0.34893779417376924, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.606328010559082, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8778268098831177, + "num_tokens": 106596009.0, + "step": 2743 + }, + { + "epoch": 0.3490650044523598, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.607058525085449, + "learning_rate": 1e-06, + "loss": 0.3893, + "mean_token_accuracy": 0.8704316020011902, + "num_tokens": 106637043.0, + "step": 2744 + }, + { + "epoch": 0.34919221473095025, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.519133567810059, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.8735594749450684, + "num_tokens": 106679269.0, + "step": 2745 + }, + { + "epoch": 0.3493194250095408, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.62822961807251, + "learning_rate": 1e-06, + "loss": 0.4126, + "mean_token_accuracy": 0.8652708530426025, + "num_tokens": 106719577.0, + "step": 2746 + }, + { + "epoch": 0.3494466352881313, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.1205673217773438e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.538283824920654, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8760159015655518, + "num_tokens": 106762489.0, + "step": 2747 + }, + { + "epoch": 0.3495738455667218, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.635157585144043, + "learning_rate": 1e-06, + "loss": 0.396, + "mean_token_accuracy": 0.8653280735015869, + "num_tokens": 106804345.0, + "step": 2748 + }, + { + "epoch": 0.3497010558453123, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.567002296447754, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.87491375207901, + "num_tokens": 106848970.0, + "step": 2749 + }, + { + "epoch": 0.34982826612390283, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.583902359008789, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8814484477043152, + "num_tokens": 106885143.0, + "step": 2750 + }, + { + "epoch": 0.3499554764024933, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.668729305267334, + "learning_rate": 1e-06, + "loss": 0.4023, + "mean_token_accuracy": 0.8652276396751404, + "num_tokens": 106922702.0, + "step": 2751 + }, + { + "epoch": 0.35008268668108383, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.608010768890381, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8845532536506653, + "num_tokens": 106956595.0, + "step": 2752 + }, + { + "epoch": 0.35020989695967436, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.584586143493652, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8800413608551025, + "num_tokens": 106991517.0, + "step": 2753 + }, + { + "epoch": 0.35033710723826483, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.613125801086426, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8746089935302734, + "num_tokens": 107029022.0, + "step": 2754 + }, + { + "epoch": 0.35046431751685536, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.523791790008545, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8744176626205444, + "num_tokens": 107073870.0, + "step": 2755 + }, + { + "epoch": 0.3505915277954459, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.654158115386963, + "learning_rate": 1e-06, + "loss": 0.4004, + "mean_token_accuracy": 0.8667402267456055, + "num_tokens": 107111902.0, + "step": 2756 + }, + { + "epoch": 0.35071873807403636, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.6702375411987305, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8757129907608032, + "num_tokens": 107148920.0, + "step": 2757 + }, + { + "epoch": 0.3508459483526269, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.589191436767578, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8772857189178467, + "num_tokens": 107184837.0, + "step": 2758 + }, + { + "epoch": 0.3509731586312174, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.664380073547363, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8779077529907227, + "num_tokens": 107223076.0, + "step": 2759 + }, + { + "epoch": 0.3511003689098079, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.486944198608398, + "learning_rate": 1e-06, + "loss": 0.379, + "mean_token_accuracy": 0.8723552227020264, + "num_tokens": 107271230.0, + "step": 2760 + }, + { + "epoch": 0.3512275791883984, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.689511299133301, + "learning_rate": 1e-06, + "loss": 0.4569, + "mean_token_accuracy": 0.8531535863876343, + "num_tokens": 107311731.0, + "step": 2761 + }, + { + "epoch": 0.35135478946698895, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.644305229187012, + "learning_rate": 1e-06, + "loss": 0.3965, + "mean_token_accuracy": 0.8611530065536499, + "num_tokens": 107349199.0, + "step": 2762 + }, + { + "epoch": 0.3514819997455794, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.571982383728027, + "learning_rate": 1e-06, + "loss": 0.3714, + "mean_token_accuracy": 0.8760972023010254, + "num_tokens": 107390526.0, + "step": 2763 + }, + { + "epoch": 0.35160921002416995, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.716817378997803, + "learning_rate": 1e-06, + "loss": 0.3832, + "mean_token_accuracy": 0.8724535703659058, + "num_tokens": 107421374.0, + "step": 2764 + }, + { + "epoch": 0.3517364203027605, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.775362491607666, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8761769533157349, + "num_tokens": 107450619.0, + "step": 2765 + }, + { + "epoch": 0.35186363058135095, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.610628604888916, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8743191957473755, + "num_tokens": 107491501.0, + "step": 2766 + }, + { + "epoch": 0.3519908408599415, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.559145927429199, + "learning_rate": 1e-06, + "loss": 0.3938, + "mean_token_accuracy": 0.8665041327476501, + "num_tokens": 107530814.0, + "step": 2767 + }, + { + "epoch": 0.352118051138532, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.482522010803223, + "learning_rate": 1e-06, + "loss": 0.3948, + "mean_token_accuracy": 0.8676912784576416, + "num_tokens": 107575648.0, + "step": 2768 + }, + { + "epoch": 0.3522452614171225, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.5681657791137695, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.8667100667953491, + "num_tokens": 107615721.0, + "step": 2769 + }, + { + "epoch": 0.352372471695713, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.655458927154541, + "learning_rate": 1e-06, + "loss": 0.404, + "mean_token_accuracy": 0.8690618276596069, + "num_tokens": 107652378.0, + "step": 2770 + }, + { + "epoch": 0.35249968197430354, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.6005353927612305, + "learning_rate": 1e-06, + "loss": 0.432, + "mean_token_accuracy": 0.8590344190597534, + "num_tokens": 107696023.0, + "step": 2771 + }, + { + "epoch": 0.352626892252894, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.476212501525879, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.886481761932373, + "num_tokens": 107740324.0, + "step": 2772 + }, + { + "epoch": 0.35275410253148454, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.590925216674805, + "learning_rate": 1e-06, + "loss": 0.3956, + "mean_token_accuracy": 0.8703871965408325, + "num_tokens": 107779145.0, + "step": 2773 + }, + { + "epoch": 0.35288131281007507, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.601181983947754, + "learning_rate": 1e-06, + "loss": 0.4246, + "mean_token_accuracy": 0.8638745546340942, + "num_tokens": 107818226.0, + "step": 2774 + }, + { + "epoch": 0.35300852308866554, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.602911949157715, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8765290975570679, + "num_tokens": 107855293.0, + "step": 2775 + }, + { + "epoch": 0.35313573336725607, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.650007724761963, + "learning_rate": 1e-06, + "loss": 0.4159, + "mean_token_accuracy": 0.8590864539146423, + "num_tokens": 107891895.0, + "step": 2776 + }, + { + "epoch": 0.3532629436458466, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.6786675453186035, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.8711292147636414, + "num_tokens": 107925332.0, + "step": 2777 + }, + { + "epoch": 0.35339015392443707, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.528019428253174, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8835475444793701, + "num_tokens": 107963871.0, + "step": 2778 + }, + { + "epoch": 0.3535173642030276, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.695658206939697, + "learning_rate": 1e-06, + "loss": 0.3865, + "mean_token_accuracy": 0.8739036321640015, + "num_tokens": 107995942.0, + "step": 2779 + }, + { + "epoch": 0.3536445744816181, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1324882507324219e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.632225036621094, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8823542594909668, + "num_tokens": 108031079.0, + "step": 2780 + }, + { + "epoch": 0.35377178476020865, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1324882507324219e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.627948760986328, + "learning_rate": 1e-06, + "loss": 0.3938, + "mean_token_accuracy": 0.8681057095527649, + "num_tokens": 108069167.0, + "step": 2781 + }, + { + "epoch": 0.3538989950387991, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1324882507324219e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.61842679977417, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8750646114349365, + "num_tokens": 108109800.0, + "step": 2782 + }, + { + "epoch": 0.35402620531738965, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1324882507324219e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.834088325500488, + "learning_rate": 1e-06, + "loss": 0.4356, + "mean_token_accuracy": 0.8573538661003113, + "num_tokens": 108142265.0, + "step": 2783 + }, + { + "epoch": 0.3541534155959802, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1324882507324219e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.535482883453369, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8796149492263794, + "num_tokens": 108185301.0, + "step": 2784 + }, + { + "epoch": 0.35428062587457065, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1324882507324219e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.628084182739258, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.8708827495574951, + "num_tokens": 108222558.0, + "step": 2785 + }, + { + "epoch": 0.3544078361531612, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1324882507324219e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.579325199127197, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8794766664505005, + "num_tokens": 108261119.0, + "step": 2786 + }, + { + "epoch": 0.3545350464317517, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.664487361907959, + "learning_rate": 1e-06, + "loss": 0.392, + "mean_token_accuracy": 0.867811381816864, + "num_tokens": 108294819.0, + "step": 2787 + }, + { + "epoch": 0.3546622567103422, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1324882507324219e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.641583442687988, + "learning_rate": 1e-06, + "loss": 0.4199, + "mean_token_accuracy": 0.8583675622940063, + "num_tokens": 108336492.0, + "step": 2788 + }, + { + "epoch": 0.3547894669889327, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.679685592651367, + "learning_rate": 1e-06, + "loss": 0.3793, + "mean_token_accuracy": 0.8706795573234558, + "num_tokens": 108371266.0, + "step": 2789 + }, + { + "epoch": 0.35491667726752324, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.656957626342773, + "learning_rate": 1e-06, + "loss": 0.3972, + "mean_token_accuracy": 0.8667190074920654, + "num_tokens": 108407046.0, + "step": 2790 + }, + { + "epoch": 0.3550438875461137, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.723813533782959, + "learning_rate": 1e-06, + "loss": 0.464, + "mean_token_accuracy": 0.8485027551651001, + "num_tokens": 108445234.0, + "step": 2791 + }, + { + "epoch": 0.35517109782470424, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6983489990234375, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8773850202560425, + "num_tokens": 108482786.0, + "step": 2792 + }, + { + "epoch": 0.35529830810329477, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.594953536987305, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8806232810020447, + "num_tokens": 108521136.0, + "step": 2793 + }, + { + "epoch": 0.35542551838188524, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.607008457183838, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.871703028678894, + "num_tokens": 108563095.0, + "step": 2794 + }, + { + "epoch": 0.35555272866047577, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.582339763641357, + "learning_rate": 1e-06, + "loss": 0.4086, + "mean_token_accuracy": 0.8653110265731812, + "num_tokens": 108605129.0, + "step": 2795 + }, + { + "epoch": 0.3556799389390663, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.677000522613525, + "learning_rate": 1e-06, + "loss": 0.3834, + "mean_token_accuracy": 0.8725951910018921, + "num_tokens": 108642079.0, + "step": 2796 + }, + { + "epoch": 0.35580714921765677, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.670837879180908, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8799792528152466, + "num_tokens": 108676269.0, + "step": 2797 + }, + { + "epoch": 0.3559343594962473, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.662148475646973, + "learning_rate": 1e-06, + "loss": 0.4074, + "mean_token_accuracy": 0.8641982674598694, + "num_tokens": 108714198.0, + "step": 2798 + }, + { + "epoch": 0.3560615697748378, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1265277862548828e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.684591770172119, + "learning_rate": 1e-06, + "loss": 0.4235, + "mean_token_accuracy": 0.8617422580718994, + "num_tokens": 108747723.0, + "step": 2799 + }, + { + "epoch": 0.3561887800534283, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1324882507324219e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.577568531036377, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8764544129371643, + "num_tokens": 108786721.0, + "step": 2800 + }, + { + "epoch": 0.35631599033201883, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1324882507324219e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.680907726287842, + "learning_rate": 1e-06, + "loss": 0.3906, + "mean_token_accuracy": 0.8709489703178406, + "num_tokens": 108822700.0, + "step": 2801 + }, + { + "epoch": 0.35644320061060936, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1324882507324219e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.569866180419922, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8856375217437744, + "num_tokens": 108863369.0, + "step": 2802 + }, + { + "epoch": 0.35657041088919983, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.138448715209961e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.689818859100342, + "learning_rate": 1e-06, + "loss": 0.4781, + "mean_token_accuracy": 0.8473823070526123, + "num_tokens": 108903478.0, + "step": 2803 + }, + { + "epoch": 0.35669762116779036, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.138448715209961e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.5570831298828125, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.875205397605896, + "num_tokens": 108944204.0, + "step": 2804 + }, + { + "epoch": 0.3568248314463809, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.138448715209961e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.604921817779541, + "learning_rate": 1e-06, + "loss": 0.4059, + "mean_token_accuracy": 0.8620654344558716, + "num_tokens": 108986982.0, + "step": 2805 + }, + { + "epoch": 0.35695204172497136, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.138448715209961e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.612919807434082, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.8721623420715332, + "num_tokens": 109026271.0, + "step": 2806 + }, + { + "epoch": 0.3570792520035619, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1444091796875e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.767420291900635, + "learning_rate": 1e-06, + "loss": 0.446, + "mean_token_accuracy": 0.8483924269676208, + "num_tokens": 109063566.0, + "step": 2807 + }, + { + "epoch": 0.3572064622821524, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1444091796875e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.524240493774414, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8841264843940735, + "num_tokens": 109104196.0, + "step": 2808 + }, + { + "epoch": 0.3573336725607429, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1444091796875e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.650453090667725, + "learning_rate": 1e-06, + "loss": 0.3943, + "mean_token_accuracy": 0.866133451461792, + "num_tokens": 109140290.0, + "step": 2809 + }, + { + "epoch": 0.3574608828393334, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1444091796875e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6474175453186035, + "learning_rate": 1e-06, + "loss": 0.3892, + "mean_token_accuracy": 0.8699554204940796, + "num_tokens": 109182999.0, + "step": 2810 + }, + { + "epoch": 0.35758809311792394, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1444091796875e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.679919242858887, + "learning_rate": 1e-06, + "loss": 0.418, + "mean_token_accuracy": 0.8596417903900146, + "num_tokens": 109221839.0, + "step": 2811 + }, + { + "epoch": 0.3577153033965144, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1444091796875e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.612062931060791, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8846930861473083, + "num_tokens": 109255986.0, + "step": 2812 + }, + { + "epoch": 0.35784251367510495, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1444091796875e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.566787242889404, + "learning_rate": 1e-06, + "loss": 0.4111, + "mean_token_accuracy": 0.8626055717468262, + "num_tokens": 109298019.0, + "step": 2813 + }, + { + "epoch": 0.3579697239536955, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1444091796875e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.69745397567749, + "learning_rate": 1e-06, + "loss": 0.3816, + "mean_token_accuracy": 0.8640196323394775, + "num_tokens": 109330947.0, + "step": 2814 + }, + { + "epoch": 0.35809693423228595, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1444091796875e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.544764041900635, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8814412951469421, + "num_tokens": 109371376.0, + "step": 2815 + }, + { + "epoch": 0.3582241445108765, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1444091796875e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.635594367980957, + "learning_rate": 1e-06, + "loss": 0.4118, + "mean_token_accuracy": 0.864181399345398, + "num_tokens": 109412353.0, + "step": 2816 + }, + { + "epoch": 0.358351354789467, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1444091796875e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.601653575897217, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.8748286962509155, + "num_tokens": 109453631.0, + "step": 2817 + }, + { + "epoch": 0.3584785650680575, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.138448715209961e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.692965507507324, + "learning_rate": 1e-06, + "loss": 0.4006, + "mean_token_accuracy": 0.8663070201873779, + "num_tokens": 109494666.0, + "step": 2818 + }, + { + "epoch": 0.358605775346648, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.138448715209961e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.6486592292785645, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.874188244342804, + "num_tokens": 109529860.0, + "step": 2819 + }, + { + "epoch": 0.35873298562523853, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.138448715209961e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.573151111602783, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.887877345085144, + "num_tokens": 109569542.0, + "step": 2820 + }, + { + "epoch": 0.358860195903829, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.138448715209961e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.750197410583496, + "learning_rate": 1e-06, + "loss": 0.3985, + "mean_token_accuracy": 0.867188572883606, + "num_tokens": 109608194.0, + "step": 2821 + }, + { + "epoch": 0.35898740618241953, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.578440189361572, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8819345235824585, + "num_tokens": 109647670.0, + "step": 2822 + }, + { + "epoch": 0.35911461646101006, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.564089775085449, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.877610445022583, + "num_tokens": 109688517.0, + "step": 2823 + }, + { + "epoch": 0.35924182673960053, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.514883518218994, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8868837356567383, + "num_tokens": 109727570.0, + "step": 2824 + }, + { + "epoch": 0.35936903701819106, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.7619123458862305, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.867830753326416, + "num_tokens": 109761081.0, + "step": 2825 + }, + { + "epoch": 0.3594962472967816, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.504560947418213, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8884084224700928, + "num_tokens": 109802290.0, + "step": 2826 + }, + { + "epoch": 0.35962345757537206, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.604220867156982, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8783353567123413, + "num_tokens": 109840169.0, + "step": 2827 + }, + { + "epoch": 0.3597506678539626, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.747170448303223, + "learning_rate": 1e-06, + "loss": 0.4452, + "mean_token_accuracy": 0.8527918457984924, + "num_tokens": 109875516.0, + "step": 2828 + }, + { + "epoch": 0.3598778781325531, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.596899032592773, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8711713552474976, + "num_tokens": 109914854.0, + "step": 2829 + }, + { + "epoch": 0.36000508841114365, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.563377857208252, + "learning_rate": 1e-06, + "loss": 0.3975, + "mean_token_accuracy": 0.8633260130882263, + "num_tokens": 109957014.0, + "step": 2830 + }, + { + "epoch": 0.3601322986897341, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.67763614654541, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8801969885826111, + "num_tokens": 109990786.0, + "step": 2831 + }, + { + "epoch": 0.36025950896832465, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.580109596252441, + "learning_rate": 1e-06, + "loss": 0.3942, + "mean_token_accuracy": 0.8668700456619263, + "num_tokens": 110030212.0, + "step": 2832 + }, + { + "epoch": 0.3603867192469152, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.595489025115967, + "learning_rate": 1e-06, + "loss": 0.3915, + "mean_token_accuracy": 0.8690638542175293, + "num_tokens": 110069215.0, + "step": 2833 + }, + { + "epoch": 0.36051392952550565, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.574591636657715, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8822202682495117, + "num_tokens": 110107808.0, + "step": 2834 + }, + { + "epoch": 0.3606411398040962, + "ewc_loss": 0.0089111328125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.573021411895752, + "learning_rate": 1e-06, + "loss": 0.3917, + "mean_token_accuracy": 0.8697396516799927, + "num_tokens": 110149038.0, + "step": 2835 + }, + { + "epoch": 0.3607683500826867, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.655282020568848, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.8788394927978516, + "num_tokens": 110182042.0, + "step": 2836 + }, + { + "epoch": 0.3608955603612772, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.489989280700684, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8841695189476013, + "num_tokens": 110225724.0, + "step": 2837 + }, + { + "epoch": 0.3610227706398677, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.56964111328125, + "learning_rate": 1e-06, + "loss": 0.3853, + "mean_token_accuracy": 0.8701072931289673, + "num_tokens": 110270701.0, + "step": 2838 + }, + { + "epoch": 0.36114998091845824, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.6702880859375, + "learning_rate": 1e-06, + "loss": 0.4263, + "mean_token_accuracy": 0.8587969541549683, + "num_tokens": 110315419.0, + "step": 2839 + }, + { + "epoch": 0.3612771911970487, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.674751281738281, + "learning_rate": 1e-06, + "loss": 0.4395, + "mean_token_accuracy": 0.8567330837249756, + "num_tokens": 110353318.0, + "step": 2840 + }, + { + "epoch": 0.36140440147563924, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.536535739898682, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.8698259592056274, + "num_tokens": 110392517.0, + "step": 2841 + }, + { + "epoch": 0.36153161175422976, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.7218451499938965, + "learning_rate": 1e-06, + "loss": 0.3886, + "mean_token_accuracy": 0.870682418346405, + "num_tokens": 110434539.0, + "step": 2842 + }, + { + "epoch": 0.36165882203282024, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.638164043426514, + "learning_rate": 1e-06, + "loss": 0.4127, + "mean_token_accuracy": 0.8580353260040283, + "num_tokens": 110471296.0, + "step": 2843 + }, + { + "epoch": 0.36178603231141077, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.641229629516602, + "learning_rate": 1e-06, + "loss": 0.3993, + "mean_token_accuracy": 0.8673550486564636, + "num_tokens": 110511883.0, + "step": 2844 + }, + { + "epoch": 0.3619132425900013, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.629744529724121, + "learning_rate": 1e-06, + "loss": 0.4007, + "mean_token_accuracy": 0.8643297553062439, + "num_tokens": 110553183.0, + "step": 2845 + }, + { + "epoch": 0.36204045286859177, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.645707607269287, + "learning_rate": 1e-06, + "loss": 0.3944, + "mean_token_accuracy": 0.8670326471328735, + "num_tokens": 110586593.0, + "step": 2846 + }, + { + "epoch": 0.3621676631471823, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.558310031890869, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8873286843299866, + "num_tokens": 110624827.0, + "step": 2847 + }, + { + "epoch": 0.3622948734257728, + "ewc_loss": 0.00897216796875, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.808208465576172e-06, + "grad_norm": 7.545122146606445, + "learning_rate": 1e-06, + "loss": 0.3757, + "mean_token_accuracy": 0.875070333480835, + "num_tokens": 110664014.0, + "step": 2848 + }, + { + "epoch": 0.3624220837043633, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.690011501312256, + "learning_rate": 1e-06, + "loss": 0.3921, + "mean_token_accuracy": 0.8659024834632874, + "num_tokens": 110699305.0, + "step": 2849 + }, + { + "epoch": 0.3625492939829538, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.150369644165039e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.674518585205078, + "learning_rate": 1e-06, + "loss": 0.3736, + "mean_token_accuracy": 0.8740056157112122, + "num_tokens": 110731377.0, + "step": 2850 + }, + { + "epoch": 0.36267650426154435, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1563301086425781e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.696672439575195, + "learning_rate": 1e-06, + "loss": 0.4283, + "mean_token_accuracy": 0.8554105758666992, + "num_tokens": 110767379.0, + "step": 2851 + }, + { + "epoch": 0.3628037145401348, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1563301086425781e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.626826286315918, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.882430911064148, + "num_tokens": 110803174.0, + "step": 2852 + }, + { + "epoch": 0.36293092481872535, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1563301086425781e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.720182418823242, + "learning_rate": 1e-06, + "loss": 0.4234, + "mean_token_accuracy": 0.8563446998596191, + "num_tokens": 110838069.0, + "step": 2853 + }, + { + "epoch": 0.3630581350973159, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1622905731201172e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.660823822021484, + "learning_rate": 1e-06, + "loss": 0.4319, + "mean_token_accuracy": 0.8565673828125, + "num_tokens": 110880015.0, + "step": 2854 + }, + { + "epoch": 0.36318534537590635, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1622905731201172e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.631680965423584, + "learning_rate": 1e-06, + "loss": 0.457, + "mean_token_accuracy": 0.8504595160484314, + "num_tokens": 110923404.0, + "step": 2855 + }, + { + "epoch": 0.3633125556544969, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1622905731201172e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.551023006439209, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.8728357553482056, + "num_tokens": 110969716.0, + "step": 2856 + }, + { + "epoch": 0.3634397659330874, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1622905731201172e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.560501575469971, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8719208240509033, + "num_tokens": 111013081.0, + "step": 2857 + }, + { + "epoch": 0.3635669762116779, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1622905731201172e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.641323566436768, + "learning_rate": 1e-06, + "loss": 0.4097, + "mean_token_accuracy": 0.8640076518058777, + "num_tokens": 111053711.0, + "step": 2858 + }, + { + "epoch": 0.3636941864902684, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1622905731201172e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.571655750274658, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8685857653617859, + "num_tokens": 111093615.0, + "step": 2859 + }, + { + "epoch": 0.36382139676885894, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1622905731201172e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.572427272796631, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8850637078285217, + "num_tokens": 111130687.0, + "step": 2860 + }, + { + "epoch": 0.3639486070474494, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1622905731201172e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.572589874267578, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8814075589179993, + "num_tokens": 111168493.0, + "step": 2861 + }, + { + "epoch": 0.36407581732603994, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1622905731201172e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.683535575866699, + "learning_rate": 1e-06, + "loss": 0.3916, + "mean_token_accuracy": 0.8682057857513428, + "num_tokens": 111206773.0, + "step": 2862 + }, + { + "epoch": 0.36420302760463047, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1622905731201172e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.731323719024658, + "learning_rate": 1e-06, + "loss": 0.4376, + "mean_token_accuracy": 0.8579226732254028, + "num_tokens": 111240794.0, + "step": 2863 + }, + { + "epoch": 0.36433023788322094, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1682510375976562e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.724361896514893, + "learning_rate": 1e-06, + "loss": 0.4076, + "mean_token_accuracy": 0.8607511520385742, + "num_tokens": 111272642.0, + "step": 2864 + }, + { + "epoch": 0.36445744816181147, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1682510375976562e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6574602127075195, + "learning_rate": 1e-06, + "loss": 0.4174, + "mean_token_accuracy": 0.8610423803329468, + "num_tokens": 111312963.0, + "step": 2865 + }, + { + "epoch": 0.364584658440402, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1682510375976562e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.558175563812256, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8824855089187622, + "num_tokens": 111349149.0, + "step": 2866 + }, + { + "epoch": 0.36471186871899247, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1682510375976562e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.553276538848877, + "learning_rate": 1e-06, + "loss": 0.4013, + "mean_token_accuracy": 0.8659210801124573, + "num_tokens": 111398859.0, + "step": 2867 + }, + { + "epoch": 0.364839078997583, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1682510375976562e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.554001331329346, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8812781572341919, + "num_tokens": 111439809.0, + "step": 2868 + }, + { + "epoch": 0.36496628927617353, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1742115020751953e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.664636135101318, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8789242506027222, + "num_tokens": 111474166.0, + "step": 2869 + }, + { + "epoch": 0.365093499554764, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1742115020751953e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.798537731170654, + "learning_rate": 1e-06, + "loss": 0.4155, + "mean_token_accuracy": 0.8601905107498169, + "num_tokens": 111507755.0, + "step": 2870 + }, + { + "epoch": 0.36522070983335453, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1801719665527344e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.601798057556152, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8826615810394287, + "num_tokens": 111545330.0, + "step": 2871 + }, + { + "epoch": 0.36534792011194506, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1801719665527344e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.58036470413208, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8850562572479248, + "num_tokens": 111586821.0, + "step": 2872 + }, + { + "epoch": 0.36547513039053553, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1801719665527344e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.70501708984375, + "learning_rate": 1e-06, + "loss": 0.4038, + "mean_token_accuracy": 0.8633860349655151, + "num_tokens": 111621099.0, + "step": 2873 + }, + { + "epoch": 0.36560234066912606, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1801719665527344e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.605413436889648, + "learning_rate": 1e-06, + "loss": 0.4053, + "mean_token_accuracy": 0.8637315630912781, + "num_tokens": 111665522.0, + "step": 2874 + }, + { + "epoch": 0.3657295509477166, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1801719665527344e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.659689903259277, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8678049445152283, + "num_tokens": 111701475.0, + "step": 2875 + }, + { + "epoch": 0.36585676122630706, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1861324310302734e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.752313613891602, + "learning_rate": 1e-06, + "loss": 0.4034, + "mean_token_accuracy": 0.8658024668693542, + "num_tokens": 111737063.0, + "step": 2876 + }, + { + "epoch": 0.3659839715048976, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1861324310302734e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.610045433044434, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8790911436080933, + "num_tokens": 111777863.0, + "step": 2877 + }, + { + "epoch": 0.3661111817834881, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1861324310302734e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.609838008880615, + "learning_rate": 1e-06, + "loss": 0.3955, + "mean_token_accuracy": 0.8695794343948364, + "num_tokens": 111821471.0, + "step": 2878 + }, + { + "epoch": 0.3662383920620786, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1861324310302734e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.760069847106934, + "learning_rate": 1e-06, + "loss": 0.4275, + "mean_token_accuracy": 0.8561770915985107, + "num_tokens": 111857087.0, + "step": 2879 + }, + { + "epoch": 0.3663656023406691, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1861324310302734e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.574450492858887, + "learning_rate": 1e-06, + "loss": 0.4379, + "mean_token_accuracy": 0.8586944341659546, + "num_tokens": 111903820.0, + "step": 2880 + }, + { + "epoch": 0.36649281261925964, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1861324310302734e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.596333980560303, + "learning_rate": 1e-06, + "loss": 0.413, + "mean_token_accuracy": 0.8609476685523987, + "num_tokens": 111945909.0, + "step": 2881 + }, + { + "epoch": 0.3666200228978502, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1861324310302734e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.577069282531738, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8794065713882446, + "num_tokens": 111987382.0, + "step": 2882 + }, + { + "epoch": 0.36674723317644065, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.1861324310302734e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.663503170013428, + "learning_rate": 1e-06, + "loss": 0.3944, + "mean_token_accuracy": 0.8638448715209961, + "num_tokens": 112023555.0, + "step": 2883 + }, + { + "epoch": 0.3668744434550312, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1861324310302734e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.656877517700195, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8752526640892029, + "num_tokens": 112059084.0, + "step": 2884 + }, + { + "epoch": 0.3670016537336217, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1861324310302734e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.558858394622803, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8782051801681519, + "num_tokens": 112101002.0, + "step": 2885 + }, + { + "epoch": 0.3671288640122122, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1920928955078125e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6371378898620605, + "learning_rate": 1e-06, + "loss": 0.4311, + "mean_token_accuracy": 0.8586834669113159, + "num_tokens": 112145174.0, + "step": 2886 + }, + { + "epoch": 0.3672560742908027, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1920928955078125e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.62213659286499, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8805485963821411, + "num_tokens": 112183219.0, + "step": 2887 + }, + { + "epoch": 0.36738328456939323, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1920928955078125e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.535702228546143, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8827186226844788, + "num_tokens": 112221938.0, + "step": 2888 + }, + { + "epoch": 0.3675104948479837, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1920928955078125e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.569094657897949, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8878507614135742, + "num_tokens": 112265186.0, + "step": 2889 + }, + { + "epoch": 0.36763770512657423, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1920928955078125e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.701849937438965, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8803402185440063, + "num_tokens": 112298784.0, + "step": 2890 + }, + { + "epoch": 0.36776491540516476, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1980533599853516e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.595427513122559, + "learning_rate": 1e-06, + "loss": 0.4191, + "mean_token_accuracy": 0.8580719828605652, + "num_tokens": 112343091.0, + "step": 2891 + }, + { + "epoch": 0.36789212568375523, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1980533599853516e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.761335372924805, + "learning_rate": 1e-06, + "loss": 0.3917, + "mean_token_accuracy": 0.8695571422576904, + "num_tokens": 112379396.0, + "step": 2892 + }, + { + "epoch": 0.36801933596234576, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1980533599853516e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.5765461921691895, + "learning_rate": 1e-06, + "loss": 0.4072, + "mean_token_accuracy": 0.865595281124115, + "num_tokens": 112420952.0, + "step": 2893 + }, + { + "epoch": 0.3681465462409363, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1980533599853516e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.637209892272949, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8764523863792419, + "num_tokens": 112456879.0, + "step": 2894 + }, + { + "epoch": 0.36827375651952676, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1980533599853516e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.600617408752441, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8796260356903076, + "num_tokens": 112493282.0, + "step": 2895 + }, + { + "epoch": 0.3684009667981173, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1980533599853516e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.689296722412109, + "learning_rate": 1e-06, + "loss": 0.3913, + "mean_token_accuracy": 0.866487979888916, + "num_tokens": 112531020.0, + "step": 2896 + }, + { + "epoch": 0.3685281770767078, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1980533599853516e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6093597412109375, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8720544576644897, + "num_tokens": 112570600.0, + "step": 2897 + }, + { + "epoch": 0.3686553873552983, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1980533599853516e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.624821186065674, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.879128098487854, + "num_tokens": 112606964.0, + "step": 2898 + }, + { + "epoch": 0.3687825976338888, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.1980533599853516e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.665271759033203, + "learning_rate": 1e-06, + "loss": 0.4004, + "mean_token_accuracy": 0.8699080944061279, + "num_tokens": 112645535.0, + "step": 2899 + }, + { + "epoch": 0.36890980791247935, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.1980533599853516e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.664924621582031, + "learning_rate": 1e-06, + "loss": 0.4369, + "mean_token_accuracy": 0.8516658544540405, + "num_tokens": 112686177.0, + "step": 2900 + }, + { + "epoch": 0.3690370181910698, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.1980533599853516e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.734933853149414, + "learning_rate": 1e-06, + "loss": 0.421, + "mean_token_accuracy": 0.8609535098075867, + "num_tokens": 112725965.0, + "step": 2901 + }, + { + "epoch": 0.36916422846966035, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.585515975952148, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8811032772064209, + "num_tokens": 112765542.0, + "step": 2902 + }, + { + "epoch": 0.3692914387482509, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.5600409507751465, + "learning_rate": 1e-06, + "loss": 0.3987, + "mean_token_accuracy": 0.864820122718811, + "num_tokens": 112809628.0, + "step": 2903 + }, + { + "epoch": 0.36941864902684135, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.562614440917969, + "learning_rate": 1e-06, + "loss": 0.4021, + "mean_token_accuracy": 0.8666552901268005, + "num_tokens": 112853693.0, + "step": 2904 + }, + { + "epoch": 0.3695458593054319, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.716890811920166, + "learning_rate": 1e-06, + "loss": 0.4428, + "mean_token_accuracy": 0.85298091173172, + "num_tokens": 112894919.0, + "step": 2905 + }, + { + "epoch": 0.3696730695840224, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6154375076293945, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8764167428016663, + "num_tokens": 112933020.0, + "step": 2906 + }, + { + "epoch": 0.3698002798626129, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.605372428894043, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8918232917785645, + "num_tokens": 112967575.0, + "step": 2907 + }, + { + "epoch": 0.3699274901412034, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.73674201965332, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.8663992285728455, + "num_tokens": 113001580.0, + "step": 2908 + }, + { + "epoch": 0.37005470041979394, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.629983425140381, + "learning_rate": 1e-06, + "loss": 0.4081, + "mean_token_accuracy": 0.8631467819213867, + "num_tokens": 113043837.0, + "step": 2909 + }, + { + "epoch": 0.3701819106983844, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.624003887176514, + "learning_rate": 1e-06, + "loss": 0.4129, + "mean_token_accuracy": 0.8642562031745911, + "num_tokens": 113086513.0, + "step": 2910 + }, + { + "epoch": 0.37030912097697494, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.61765193939209, + "learning_rate": 1e-06, + "loss": 0.4156, + "mean_token_accuracy": 0.8619815111160278, + "num_tokens": 113124560.0, + "step": 2911 + }, + { + "epoch": 0.37043633125556547, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.635346412658691, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.8727813959121704, + "num_tokens": 113165183.0, + "step": 2912 + }, + { + "epoch": 0.37056354153415594, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6417317390441895, + "learning_rate": 1e-06, + "loss": 0.4448, + "mean_token_accuracy": 0.8558149337768555, + "num_tokens": 113208458.0, + "step": 2913 + }, + { + "epoch": 0.37069075181274647, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.573179244995117, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8757117986679077, + "num_tokens": 113249307.0, + "step": 2914 + }, + { + "epoch": 0.370817962091337, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.659458637237549, + "learning_rate": 1e-06, + "loss": 0.3878, + "mean_token_accuracy": 0.8689481019973755, + "num_tokens": 113284784.0, + "step": 2915 + }, + { + "epoch": 0.37094517236992747, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.5326409339904785, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8781962990760803, + "num_tokens": 113325725.0, + "step": 2916 + }, + { + "epoch": 0.371072382648518, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6990742683410645, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8743258118629456, + "num_tokens": 113364540.0, + "step": 2917 + }, + { + "epoch": 0.3711995929271085, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.577004909515381, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8711531758308411, + "num_tokens": 113406759.0, + "step": 2918 + }, + { + "epoch": 0.371326803205699, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.563574314117432, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8745055794715881, + "num_tokens": 113452364.0, + "step": 2919 + }, + { + "epoch": 0.3714540134842895, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.647467613220215, + "learning_rate": 1e-06, + "loss": 0.3798, + "mean_token_accuracy": 0.8729171752929688, + "num_tokens": 113491633.0, + "step": 2920 + }, + { + "epoch": 0.37158122376288005, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6895670890808105, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8786091804504395, + "num_tokens": 113526972.0, + "step": 2921 + }, + { + "epoch": 0.3717084340414705, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.5828094482421875, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8724257946014404, + "num_tokens": 113569623.0, + "step": 2922 + }, + { + "epoch": 0.37183564432006105, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.606709957122803, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.8727552890777588, + "num_tokens": 113616346.0, + "step": 2923 + }, + { + "epoch": 0.3719628545986516, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.671751976013184, + "learning_rate": 1e-06, + "loss": 0.3953, + "mean_token_accuracy": 0.8698790073394775, + "num_tokens": 113652468.0, + "step": 2924 + }, + { + "epoch": 0.37209006487724205, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.651275157928467, + "learning_rate": 1e-06, + "loss": 0.4065, + "mean_token_accuracy": 0.8658429980278015, + "num_tokens": 113692993.0, + "step": 2925 + }, + { + "epoch": 0.3722172751558326, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.650763034820557, + "learning_rate": 1e-06, + "loss": 0.3861, + "mean_token_accuracy": 0.8692846298217773, + "num_tokens": 113734030.0, + "step": 2926 + }, + { + "epoch": 0.3723444854344231, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.67697286605835, + "learning_rate": 1e-06, + "loss": 0.4664, + "mean_token_accuracy": 0.844167172908783, + "num_tokens": 113775930.0, + "step": 2927 + }, + { + "epoch": 0.3724716957130136, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2040138244628906e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.663967609405518, + "learning_rate": 1e-06, + "loss": 0.4225, + "mean_token_accuracy": 0.8565758466720581, + "num_tokens": 113813307.0, + "step": 2928 + }, + { + "epoch": 0.3725989059916041, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2099742889404297e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.641882419586182, + "learning_rate": 1e-06, + "loss": 0.4124, + "mean_token_accuracy": 0.8610258102416992, + "num_tokens": 113855921.0, + "step": 2929 + }, + { + "epoch": 0.37272611627019464, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2099742889404297e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.645381927490234, + "learning_rate": 1e-06, + "loss": 0.4235, + "mean_token_accuracy": 0.8634233474731445, + "num_tokens": 113899023.0, + "step": 2930 + }, + { + "epoch": 0.37285332654878517, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.2099742889404297e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.723769187927246, + "learning_rate": 1e-06, + "loss": 0.418, + "mean_token_accuracy": 0.8610399961471558, + "num_tokens": 113933137.0, + "step": 2931 + }, + { + "epoch": 0.37298053682737564, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.692422389984131, + "learning_rate": 1e-06, + "loss": 0.4266, + "mean_token_accuracy": 0.8571502566337585, + "num_tokens": 113971933.0, + "step": 2932 + }, + { + "epoch": 0.37310774710596617, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.2099742889404297e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.69273567199707, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8767527937889099, + "num_tokens": 114006664.0, + "step": 2933 + }, + { + "epoch": 0.3732349573845567, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2099742889404297e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.548925876617432, + "learning_rate": 1e-06, + "loss": 0.3887, + "mean_token_accuracy": 0.8679110407829285, + "num_tokens": 114050212.0, + "step": 2934 + }, + { + "epoch": 0.37336216766314717, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2099742889404297e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6168131828308105, + "learning_rate": 1e-06, + "loss": 0.4136, + "mean_token_accuracy": 0.8608882427215576, + "num_tokens": 114092901.0, + "step": 2935 + }, + { + "epoch": 0.3734893779417377, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2099742889404297e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.584288597106934, + "learning_rate": 1e-06, + "loss": 0.4366, + "mean_token_accuracy": 0.8528991937637329, + "num_tokens": 114141329.0, + "step": 2936 + }, + { + "epoch": 0.3736165882203282, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.678764820098877, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8775051236152649, + "num_tokens": 114176542.0, + "step": 2937 + }, + { + "epoch": 0.3737437984989187, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.708798408508301, + "learning_rate": 1e-06, + "loss": 0.3916, + "mean_token_accuracy": 0.8694069385528564, + "num_tokens": 114211699.0, + "step": 2938 + }, + { + "epoch": 0.37387100877750923, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.710817337036133, + "learning_rate": 1e-06, + "loss": 0.4078, + "mean_token_accuracy": 0.8626949787139893, + "num_tokens": 114247971.0, + "step": 2939 + }, + { + "epoch": 0.37399821905609976, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.693964958190918, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8775722980499268, + "num_tokens": 114282957.0, + "step": 2940 + }, + { + "epoch": 0.37412542933469023, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.573275566101074, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8744677901268005, + "num_tokens": 114323193.0, + "step": 2941 + }, + { + "epoch": 0.37425263961328076, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.675976276397705, + "learning_rate": 1e-06, + "loss": 0.4051, + "mean_token_accuracy": 0.8631749153137207, + "num_tokens": 114360507.0, + "step": 2942 + }, + { + "epoch": 0.3743798498918713, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.5828471183776855, + "learning_rate": 1e-06, + "loss": 0.3946, + "mean_token_accuracy": 0.865964412689209, + "num_tokens": 114403033.0, + "step": 2943 + }, + { + "epoch": 0.37450706017046176, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.638120174407959, + "learning_rate": 1e-06, + "loss": 0.4012, + "mean_token_accuracy": 0.8645413517951965, + "num_tokens": 114442109.0, + "step": 2944 + }, + { + "epoch": 0.3746342704490523, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.526061534881592, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8816750049591064, + "num_tokens": 114483317.0, + "step": 2945 + }, + { + "epoch": 0.3747614807276428, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.66365909576416, + "learning_rate": 1e-06, + "loss": 0.4171, + "mean_token_accuracy": 0.859927237033844, + "num_tokens": 114524441.0, + "step": 2946 + }, + { + "epoch": 0.3748886910062333, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.677236080169678, + "learning_rate": 1e-06, + "loss": 0.4029, + "mean_token_accuracy": 0.8678038120269775, + "num_tokens": 114561076.0, + "step": 2947 + }, + { + "epoch": 0.3750159012848238, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.813807487487793, + "learning_rate": 1e-06, + "loss": 0.4178, + "mean_token_accuracy": 0.8589650392532349, + "num_tokens": 114594556.0, + "step": 2948 + }, + { + "epoch": 0.37514311156341434, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.713715076446533, + "learning_rate": 1e-06, + "loss": 0.3906, + "mean_token_accuracy": 0.8679391145706177, + "num_tokens": 114628888.0, + "step": 2949 + }, + { + "epoch": 0.3752703218420048, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.665170669555664, + "learning_rate": 1e-06, + "loss": 0.453, + "mean_token_accuracy": 0.8501795530319214, + "num_tokens": 114665578.0, + "step": 2950 + }, + { + "epoch": 0.37539753212059535, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.7155303955078125, + "learning_rate": 1e-06, + "loss": 0.3974, + "mean_token_accuracy": 0.8688409328460693, + "num_tokens": 114705385.0, + "step": 2951 + }, + { + "epoch": 0.3755247423991859, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2159347534179688e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.507209777832031, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8767326474189758, + "num_tokens": 114750953.0, + "step": 2952 + }, + { + "epoch": 0.37565195267777635, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.642053127288818, + "learning_rate": 1e-06, + "loss": 0.3826, + "mean_token_accuracy": 0.8727067708969116, + "num_tokens": 114792401.0, + "step": 2953 + }, + { + "epoch": 0.3757791629563669, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.607600212097168, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8799526691436768, + "num_tokens": 114834379.0, + "step": 2954 + }, + { + "epoch": 0.3759063732349574, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.730227947235107, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.8679152727127075, + "num_tokens": 114867827.0, + "step": 2955 + }, + { + "epoch": 0.3760335835135479, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6089348793029785, + "learning_rate": 1e-06, + "loss": 0.3826, + "mean_token_accuracy": 0.8723050355911255, + "num_tokens": 114911547.0, + "step": 2956 + }, + { + "epoch": 0.3761607937921384, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.710238456726074, + "learning_rate": 1e-06, + "loss": 0.3774, + "mean_token_accuracy": 0.8701148629188538, + "num_tokens": 114947458.0, + "step": 2957 + }, + { + "epoch": 0.37628800407072893, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.614533424377441, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8827400207519531, + "num_tokens": 114987474.0, + "step": 2958 + }, + { + "epoch": 0.3764152143493194, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.580244064331055, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8921551704406738, + "num_tokens": 115027148.0, + "step": 2959 + }, + { + "epoch": 0.37654242462790993, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.7261881828308105, + "learning_rate": 1e-06, + "loss": 0.3952, + "mean_token_accuracy": 0.8684679269790649, + "num_tokens": 115064773.0, + "step": 2960 + }, + { + "epoch": 0.37666963490650046, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.745368480682373, + "learning_rate": 1e-06, + "loss": 0.3969, + "mean_token_accuracy": 0.865503191947937, + "num_tokens": 115101960.0, + "step": 2961 + }, + { + "epoch": 0.37679684518509093, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6732282638549805, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8798904418945312, + "num_tokens": 115136595.0, + "step": 2962 + }, + { + "epoch": 0.37692405546368146, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.631564617156982, + "learning_rate": 1e-06, + "loss": 0.3874, + "mean_token_accuracy": 0.8679360747337341, + "num_tokens": 115175346.0, + "step": 2963 + }, + { + "epoch": 0.377051265742272, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.745902061462402, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.882997989654541, + "num_tokens": 115207991.0, + "step": 2964 + }, + { + "epoch": 0.37717847602086246, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.670948028564453, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8770813345909119, + "num_tokens": 115246258.0, + "step": 2965 + }, + { + "epoch": 0.377305686299453, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2218952178955078e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.783087730407715, + "learning_rate": 1e-06, + "loss": 0.388, + "mean_token_accuracy": 0.8727173805236816, + "num_tokens": 115284674.0, + "step": 2966 + }, + { + "epoch": 0.3774328965780435, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.684995651245117, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8945144414901733, + "num_tokens": 115321627.0, + "step": 2967 + }, + { + "epoch": 0.377560106856634, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.826747894287109, + "learning_rate": 1e-06, + "loss": 0.4385, + "mean_token_accuracy": 0.8515721559524536, + "num_tokens": 115352063.0, + "step": 2968 + }, + { + "epoch": 0.3776873171352245, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.717207908630371, + "learning_rate": 1e-06, + "loss": 0.3733, + "mean_token_accuracy": 0.8742240071296692, + "num_tokens": 115396954.0, + "step": 2969 + }, + { + "epoch": 0.37781452741381505, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.562178611755371, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.870617151260376, + "num_tokens": 115438778.0, + "step": 2970 + }, + { + "epoch": 0.3779417376924055, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.616560935974121, + "learning_rate": 1e-06, + "loss": 0.4383, + "mean_token_accuracy": 0.8530658483505249, + "num_tokens": 115480851.0, + "step": 2971 + }, + { + "epoch": 0.37806894797099605, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.802524566650391, + "learning_rate": 1e-06, + "loss": 0.4544, + "mean_token_accuracy": 0.8485379219055176, + "num_tokens": 115515851.0, + "step": 2972 + }, + { + "epoch": 0.3781961582495866, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.531825542449951, + "learning_rate": 1e-06, + "loss": 0.3846, + "mean_token_accuracy": 0.8708899021148682, + "num_tokens": 115560391.0, + "step": 2973 + }, + { + "epoch": 0.37832336852817705, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.748661041259766, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8762757778167725, + "num_tokens": 115595684.0, + "step": 2974 + }, + { + "epoch": 0.3784505788067676, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.625783443450928, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8854267001152039, + "num_tokens": 115633767.0, + "step": 2975 + }, + { + "epoch": 0.3785777890853581, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.8051629066467285, + "learning_rate": 1e-06, + "loss": 0.455, + "mean_token_accuracy": 0.8503868579864502, + "num_tokens": 115669125.0, + "step": 2976 + }, + { + "epoch": 0.3787049993639486, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.570417404174805, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8785631060600281, + "num_tokens": 115709781.0, + "step": 2977 + }, + { + "epoch": 0.3788322096425391, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.660007476806641, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8788729906082153, + "num_tokens": 115750644.0, + "step": 2978 + }, + { + "epoch": 0.37895941992112964, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.703814506530762, + "learning_rate": 1e-06, + "loss": 0.4029, + "mean_token_accuracy": 0.8645898103713989, + "num_tokens": 115793706.0, + "step": 2979 + }, + { + "epoch": 0.37908663019972016, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.669036388397217, + "learning_rate": 1e-06, + "loss": 0.4155, + "mean_token_accuracy": 0.8598989248275757, + "num_tokens": 115836692.0, + "step": 2980 + }, + { + "epoch": 0.37921384047831064, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2278556823730469e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.718140125274658, + "learning_rate": 1e-06, + "loss": 0.3884, + "mean_token_accuracy": 0.8678423166275024, + "num_tokens": 115872756.0, + "step": 2981 + }, + { + "epoch": 0.37934105075690117, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.621744155883789, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8749313354492188, + "num_tokens": 115914737.0, + "step": 2982 + }, + { + "epoch": 0.3794682610354917, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.676031112670898, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8783160448074341, + "num_tokens": 115951887.0, + "step": 2983 + }, + { + "epoch": 0.37959547131408217, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.5792927742004395, + "learning_rate": 1e-06, + "loss": 0.4055, + "mean_token_accuracy": 0.8641520738601685, + "num_tokens": 115998886.0, + "step": 2984 + }, + { + "epoch": 0.3797226815926727, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.233816146850586e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.853474140167236, + "learning_rate": 1e-06, + "loss": 0.3905, + "mean_token_accuracy": 0.865783154964447, + "num_tokens": 116032427.0, + "step": 2985 + }, + { + "epoch": 0.3798498918712632, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.239776611328125e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.587637901306152, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8871796131134033, + "num_tokens": 116071785.0, + "step": 2986 + }, + { + "epoch": 0.3799771021498537, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.245737075805664e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.790181636810303, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8852254152297974, + "num_tokens": 116103788.0, + "step": 2987 + }, + { + "epoch": 0.3801043124284442, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.245737075805664e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 8.002815246582031, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8885824680328369, + "num_tokens": 116140681.0, + "step": 2988 + }, + { + "epoch": 0.38023152270703475, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.245737075805664e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.69070291519165, + "learning_rate": 1e-06, + "loss": 0.4155, + "mean_token_accuracy": 0.8578495383262634, + "num_tokens": 116177610.0, + "step": 2989 + }, + { + "epoch": 0.3803587329856252, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.245737075805664e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.760714530944824, + "learning_rate": 1e-06, + "loss": 0.4005, + "mean_token_accuracy": 0.8666059970855713, + "num_tokens": 116215840.0, + "step": 2990 + }, + { + "epoch": 0.38048594326421575, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.245737075805664e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.647894382476807, + "learning_rate": 1e-06, + "loss": 0.4379, + "mean_token_accuracy": 0.8513288497924805, + "num_tokens": 116259998.0, + "step": 2991 + }, + { + "epoch": 0.3806131535428063, + "ewc_loss": 0.00909423828125, + "ewc_loss_diag": 1.245737075805664e-05, + "ewc_loss_parallel": 7.867813110351562e-06, + "grad_norm": 7.7143144607543945, + "learning_rate": 1e-06, + "loss": 0.3894, + "mean_token_accuracy": 0.8698335886001587, + "num_tokens": 116296363.0, + "step": 2992 + }, + { + "epoch": 0.38074036382139675, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.245737075805664e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.68490743637085, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8805160522460938, + "num_tokens": 116334815.0, + "step": 2993 + }, + { + "epoch": 0.3808675740999873, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.685402870178223, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8741840124130249, + "num_tokens": 116369466.0, + "step": 2994 + }, + { + "epoch": 0.3809947843785778, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.245737075805664e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.716200828552246, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.8797577023506165, + "num_tokens": 116406031.0, + "step": 2995 + }, + { + "epoch": 0.3811219946571683, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.245737075805664e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.72510290145874, + "learning_rate": 1e-06, + "loss": 0.4109, + "mean_token_accuracy": 0.8616693019866943, + "num_tokens": 116443706.0, + "step": 2996 + }, + { + "epoch": 0.3812492049357588, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.245737075805664e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.715473651885986, + "learning_rate": 1e-06, + "loss": 0.4004, + "mean_token_accuracy": 0.8657366037368774, + "num_tokens": 116479304.0, + "step": 2997 + }, + { + "epoch": 0.38137641521434934, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.239776611328125e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.789643287658691, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8721628189086914, + "num_tokens": 116509811.0, + "step": 2998 + }, + { + "epoch": 0.3815036254929398, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.6029276847839355, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8788701891899109, + "num_tokens": 116553445.0, + "step": 2999 + }, + { + "epoch": 0.38163083577153034, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.239776611328125e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.651825428009033, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.880128800868988, + "num_tokens": 116589921.0, + "step": 3000 + }, + { + "epoch": 0.38175804605012087, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.627824783325195, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8776966333389282, + "num_tokens": 116632465.0, + "step": 3001 + }, + { + "epoch": 0.38188525632871134, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.725225925445557, + "learning_rate": 1e-06, + "loss": 0.4182, + "mean_token_accuracy": 0.8624953031539917, + "num_tokens": 116671417.0, + "step": 3002 + }, + { + "epoch": 0.38201246660730187, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.691423416137695, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8722690939903259, + "num_tokens": 116710019.0, + "step": 3003 + }, + { + "epoch": 0.3821396768858924, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.686656475067139, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.8687572479248047, + "num_tokens": 116746789.0, + "step": 3004 + }, + { + "epoch": 0.38226688716448287, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 8.878157615661621, + "learning_rate": 1e-06, + "loss": 0.4385, + "mean_token_accuracy": 0.8526607751846313, + "num_tokens": 116788563.0, + "step": 3005 + }, + { + "epoch": 0.3823940974430734, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.687461853027344, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8737455606460571, + "num_tokens": 116826212.0, + "step": 3006 + }, + { + "epoch": 0.38252130772166393, + "ewc_loss": 0.009033203125, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.748603820800781e-06, + "grad_norm": 7.483292102813721, + "learning_rate": 1e-06, + "loss": 0.3994, + "mean_token_accuracy": 0.8658350706100464, + "num_tokens": 116875441.0, + "step": 3007 + }, + { + "epoch": 0.3826485180002544, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.824818134307861, + "learning_rate": 1e-06, + "loss": 0.435, + "mean_token_accuracy": 0.8510317802429199, + "num_tokens": 116910463.0, + "step": 3008 + }, + { + "epoch": 0.38277572827884493, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.694647789001465, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8806912302970886, + "num_tokens": 116945788.0, + "step": 3009 + }, + { + "epoch": 0.38290293855743546, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.696439266204834, + "learning_rate": 1e-06, + "loss": 0.4126, + "mean_token_accuracy": 0.8621721267700195, + "num_tokens": 116988698.0, + "step": 3010 + }, + { + "epoch": 0.38303014883602593, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.760817050933838, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8751365542411804, + "num_tokens": 117021554.0, + "step": 3011 + }, + { + "epoch": 0.38315735911461646, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.679426193237305, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8773200511932373, + "num_tokens": 117057348.0, + "step": 3012 + }, + { + "epoch": 0.383284569393207, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.636961936950684, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8858683109283447, + "num_tokens": 117096744.0, + "step": 3013 + }, + { + "epoch": 0.38341177967179746, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.758026599884033, + "learning_rate": 1e-06, + "loss": 0.3957, + "mean_token_accuracy": 0.8660567998886108, + "num_tokens": 117130975.0, + "step": 3014 + }, + { + "epoch": 0.383538989950388, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.630538463592529, + "learning_rate": 1e-06, + "loss": 0.4316, + "mean_token_accuracy": 0.8535600304603577, + "num_tokens": 117176856.0, + "step": 3015 + }, + { + "epoch": 0.3836662002289785, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.868696689605713, + "learning_rate": 1e-06, + "loss": 0.4351, + "mean_token_accuracy": 0.8561729192733765, + "num_tokens": 117212706.0, + "step": 3016 + }, + { + "epoch": 0.383793410507569, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.706469535827637, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8788381218910217, + "num_tokens": 117245399.0, + "step": 3017 + }, + { + "epoch": 0.3839206207861595, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.74267578125, + "learning_rate": 1e-06, + "loss": 0.3815, + "mean_token_accuracy": 0.8729106783866882, + "num_tokens": 117282580.0, + "step": 3018 + }, + { + "epoch": 0.38404783106475004, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.722680568695068, + "learning_rate": 1e-06, + "loss": 0.3732, + "mean_token_accuracy": 0.8743519186973572, + "num_tokens": 117321819.0, + "step": 3019 + }, + { + "epoch": 0.3841750413433405, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.606198310852051, + "learning_rate": 1e-06, + "loss": 0.4019, + "mean_token_accuracy": 0.8661187291145325, + "num_tokens": 117365509.0, + "step": 3020 + }, + { + "epoch": 0.38430225162193105, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.669139862060547, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8770538568496704, + "num_tokens": 117408542.0, + "step": 3021 + }, + { + "epoch": 0.3844294619005216, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.723381996154785, + "learning_rate": 1e-06, + "loss": 0.4571, + "mean_token_accuracy": 0.849345862865448, + "num_tokens": 117453156.0, + "step": 3022 + }, + { + "epoch": 0.38455667217911205, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.867856502532959, + "learning_rate": 1e-06, + "loss": 0.4071, + "mean_token_accuracy": 0.8645331263542175, + "num_tokens": 117488573.0, + "step": 3023 + }, + { + "epoch": 0.3846838824577026, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.697791576385498, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8753626942634583, + "num_tokens": 117530090.0, + "step": 3024 + }, + { + "epoch": 0.3848110927362931, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.640251636505127, + "learning_rate": 1e-06, + "loss": 0.4121, + "mean_token_accuracy": 0.8625291585922241, + "num_tokens": 117575424.0, + "step": 3025 + }, + { + "epoch": 0.3849383030148836, + "ewc_loss": 0.0091552734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.839057445526123, + "learning_rate": 1e-06, + "loss": 0.399, + "mean_token_accuracy": 0.8664042353630066, + "num_tokens": 117612428.0, + "step": 3026 + }, + { + "epoch": 0.3850655132934741, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.709969997406006, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8771497011184692, + "num_tokens": 117649511.0, + "step": 3027 + }, + { + "epoch": 0.38519272357206463, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.703883171081543, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8772789239883423, + "num_tokens": 117684939.0, + "step": 3028 + }, + { + "epoch": 0.3853199338506551, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2516975402832031e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.701167583465576, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8844399452209473, + "num_tokens": 117722912.0, + "step": 3029 + }, + { + "epoch": 0.38544714412924563, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.634013652801514, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8852878212928772, + "num_tokens": 117766869.0, + "step": 3030 + }, + { + "epoch": 0.38557435440783616, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.761697292327881, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8858485221862793, + "num_tokens": 117800261.0, + "step": 3031 + }, + { + "epoch": 0.3857015646864267, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.7010393142700195, + "learning_rate": 1e-06, + "loss": 0.4544, + "mean_token_accuracy": 0.8495820760726929, + "num_tokens": 117843963.0, + "step": 3032 + }, + { + "epoch": 0.38582877496501716, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.7322797775268555, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8782989978790283, + "num_tokens": 117879386.0, + "step": 3033 + }, + { + "epoch": 0.3859559852436077, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.814686298370361, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8738842010498047, + "num_tokens": 117912570.0, + "step": 3034 + }, + { + "epoch": 0.3860831955221982, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.652197360992432, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8782297372817993, + "num_tokens": 117954119.0, + "step": 3035 + }, + { + "epoch": 0.3862104058007887, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.764330863952637, + "learning_rate": 1e-06, + "loss": 0.4536, + "mean_token_accuracy": 0.8501943349838257, + "num_tokens": 117993745.0, + "step": 3036 + }, + { + "epoch": 0.3863376160793792, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.733062744140625, + "learning_rate": 1e-06, + "loss": 0.3576, + "mean_token_accuracy": 0.8760886192321777, + "num_tokens": 118027375.0, + "step": 3037 + }, + { + "epoch": 0.38646482635796975, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.666974067687988, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8841429948806763, + "num_tokens": 118068984.0, + "step": 3038 + }, + { + "epoch": 0.3865920366365602, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.754394054412842, + "learning_rate": 1e-06, + "loss": 0.4434, + "mean_token_accuracy": 0.8520079255104065, + "num_tokens": 118110085.0, + "step": 3039 + }, + { + "epoch": 0.38671924691515075, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.678550720214844, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8812199234962463, + "num_tokens": 118149637.0, + "step": 3040 + }, + { + "epoch": 0.3868464571937413, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.80114221572876, + "learning_rate": 1e-06, + "loss": 0.3971, + "mean_token_accuracy": 0.8609211444854736, + "num_tokens": 118183227.0, + "step": 3041 + }, + { + "epoch": 0.38697366747233175, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.778454780578613, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.8776432275772095, + "num_tokens": 118218381.0, + "step": 3042 + }, + { + "epoch": 0.3871008777509223, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6772050857543945, + "learning_rate": 1e-06, + "loss": 0.3982, + "mean_token_accuracy": 0.8691576719284058, + "num_tokens": 118258817.0, + "step": 3043 + }, + { + "epoch": 0.3872280880295128, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.633610248565674, + "learning_rate": 1e-06, + "loss": 0.4253, + "mean_token_accuracy": 0.8619092106819153, + "num_tokens": 118303693.0, + "step": 3044 + }, + { + "epoch": 0.3873552983081033, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.894065856933594, + "learning_rate": 1e-06, + "loss": 0.3947, + "mean_token_accuracy": 0.8684951066970825, + "num_tokens": 118335702.0, + "step": 3045 + }, + { + "epoch": 0.3874825085866938, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.696977138519287, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.8696742653846741, + "num_tokens": 118374582.0, + "step": 3046 + }, + { + "epoch": 0.38760971886528434, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.799322605133057, + "learning_rate": 1e-06, + "loss": 0.4087, + "mean_token_accuracy": 0.8627756834030151, + "num_tokens": 118409664.0, + "step": 3047 + }, + { + "epoch": 0.3877369291438748, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.735133171081543, + "learning_rate": 1e-06, + "loss": 0.3919, + "mean_token_accuracy": 0.8686975240707397, + "num_tokens": 118445318.0, + "step": 3048 + }, + { + "epoch": 0.38786413942246534, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.7381181716918945, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8798617124557495, + "num_tokens": 118482313.0, + "step": 3049 + }, + { + "epoch": 0.38799134970105587, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.766099452972412, + "learning_rate": 1e-06, + "loss": 0.4287, + "mean_token_accuracy": 0.8570433855056763, + "num_tokens": 118522559.0, + "step": 3050 + }, + { + "epoch": 0.38811855997964634, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.661626815795898, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8787432312965393, + "num_tokens": 118564185.0, + "step": 3051 + }, + { + "epoch": 0.38824577025823687, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.780425548553467, + "learning_rate": 1e-06, + "loss": 0.3966, + "mean_token_accuracy": 0.866179883480072, + "num_tokens": 118600374.0, + "step": 3052 + }, + { + "epoch": 0.3883729805368274, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2636184692382812e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.690141201019287, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8805749416351318, + "num_tokens": 118633476.0, + "step": 3053 + }, + { + "epoch": 0.38850019081541787, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.662791728973389, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8771562576293945, + "num_tokens": 118670989.0, + "step": 3054 + }, + { + "epoch": 0.3886274010940084, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.689455509185791, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8809830546379089, + "num_tokens": 118707277.0, + "step": 3055 + }, + { + "epoch": 0.3887546113725989, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.740049839019775, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8846014738082886, + "num_tokens": 118741466.0, + "step": 3056 + }, + { + "epoch": 0.3888818216511894, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.603949546813965, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8734985589981079, + "num_tokens": 118783340.0, + "step": 3057 + }, + { + "epoch": 0.3890090319297799, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.740853786468506, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8740552663803101, + "num_tokens": 118820366.0, + "step": 3058 + }, + { + "epoch": 0.38913624220837045, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.582080841064453, + "learning_rate": 1e-06, + "loss": 0.4113, + "mean_token_accuracy": 0.8663783073425293, + "num_tokens": 118868272.0, + "step": 3059 + }, + { + "epoch": 0.3892634524869609, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.62034797668457, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8730883598327637, + "num_tokens": 118910953.0, + "step": 3060 + }, + { + "epoch": 0.38939066276555145, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.8933916091918945, + "learning_rate": 1e-06, + "loss": 0.4596, + "mean_token_accuracy": 0.8493423461914062, + "num_tokens": 118945755.0, + "step": 3061 + }, + { + "epoch": 0.389517873044142, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.623161315917969, + "learning_rate": 1e-06, + "loss": 0.3769, + "mean_token_accuracy": 0.8770096302032471, + "num_tokens": 118987480.0, + "step": 3062 + }, + { + "epoch": 0.38964508332273246, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.657928943634033, + "learning_rate": 1e-06, + "loss": 0.3971, + "mean_token_accuracy": 0.8650412559509277, + "num_tokens": 119029749.0, + "step": 3063 + }, + { + "epoch": 0.389772293601323, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.669583797454834, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8693791627883911, + "num_tokens": 119069598.0, + "step": 3064 + }, + { + "epoch": 0.3898995038799135, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.776120662689209, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8821211457252502, + "num_tokens": 119106100.0, + "step": 3065 + }, + { + "epoch": 0.390026714158504, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2814998626708984e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.692966938018799, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8791989088058472, + "num_tokens": 119144362.0, + "step": 3066 + }, + { + "epoch": 0.3901539244370945, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.700430870056152, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8672798871994019, + "num_tokens": 119182049.0, + "step": 3067 + }, + { + "epoch": 0.39028113471568504, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.692441940307617, + "learning_rate": 1e-06, + "loss": 0.4128, + "mean_token_accuracy": 0.8646929264068604, + "num_tokens": 119220987.0, + "step": 3068 + }, + { + "epoch": 0.3904083449942755, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.663855075836182, + "learning_rate": 1e-06, + "loss": 0.3729, + "mean_token_accuracy": 0.8714096546173096, + "num_tokens": 119258418.0, + "step": 3069 + }, + { + "epoch": 0.39053555527286604, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.757687091827393, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.882958173751831, + "num_tokens": 119290701.0, + "step": 3070 + }, + { + "epoch": 0.39066276555145657, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2814998626708984e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.70020055770874, + "learning_rate": 1e-06, + "loss": 0.4079, + "mean_token_accuracy": 0.8652557134628296, + "num_tokens": 119330406.0, + "step": 3071 + }, + { + "epoch": 0.39078997583004704, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2755393981933594e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.601339340209961, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8886120319366455, + "num_tokens": 119367475.0, + "step": 3072 + }, + { + "epoch": 0.39091718610863757, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2814998626708984e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.784637451171875, + "learning_rate": 1e-06, + "loss": 0.4344, + "mean_token_accuracy": 0.8566871285438538, + "num_tokens": 119408487.0, + "step": 3073 + }, + { + "epoch": 0.3910443963872281, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2814998626708984e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.71016263961792, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8793915510177612, + "num_tokens": 119448166.0, + "step": 3074 + }, + { + "epoch": 0.39117160666581857, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2814998626708984e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.672917366027832, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8808963894844055, + "num_tokens": 119484607.0, + "step": 3075 + }, + { + "epoch": 0.3912988169444091, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2814998626708984e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.800172805786133, + "learning_rate": 1e-06, + "loss": 0.3981, + "mean_token_accuracy": 0.8643651604652405, + "num_tokens": 119520930.0, + "step": 3076 + }, + { + "epoch": 0.39142602722299963, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2814998626708984e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.655576229095459, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.8774173259735107, + "num_tokens": 119561828.0, + "step": 3077 + }, + { + "epoch": 0.3915532375015901, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2814998626708984e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.706502437591553, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8713022470474243, + "num_tokens": 119598441.0, + "step": 3078 + }, + { + "epoch": 0.39168044778018063, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2814998626708984e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.730110168457031, + "learning_rate": 1e-06, + "loss": 0.4444, + "mean_token_accuracy": 0.8526878952980042, + "num_tokens": 119640844.0, + "step": 3079 + }, + { + "epoch": 0.39180765805877116, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2814998626708984e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.586318016052246, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.876141369342804, + "num_tokens": 119679854.0, + "step": 3080 + }, + { + "epoch": 0.3919348683373617, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2814998626708984e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6759138107299805, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.8738781809806824, + "num_tokens": 119719052.0, + "step": 3081 + }, + { + "epoch": 0.39206207861595216, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2814998626708984e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.736927032470703, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8681668043136597, + "num_tokens": 119756497.0, + "step": 3082 + }, + { + "epoch": 0.3921892888945427, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2814998626708984e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.668334007263184, + "learning_rate": 1e-06, + "loss": 0.4069, + "mean_token_accuracy": 0.863863468170166, + "num_tokens": 119797063.0, + "step": 3083 + }, + { + "epoch": 0.3923164991731332, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2874603271484375e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.694869041442871, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8743313550949097, + "num_tokens": 119832748.0, + "step": 3084 + }, + { + "epoch": 0.3924437094517237, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2874603271484375e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.583644390106201, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.874743640422821, + "num_tokens": 119876726.0, + "step": 3085 + }, + { + "epoch": 0.3925709197303142, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2874603271484375e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 8.189043998718262, + "learning_rate": 1e-06, + "loss": 0.3967, + "mean_token_accuracy": 0.867971658706665, + "num_tokens": 119917095.0, + "step": 3086 + }, + { + "epoch": 0.39269813000890474, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2874603271484375e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.700972080230713, + "learning_rate": 1e-06, + "loss": 0.4066, + "mean_token_accuracy": 0.8634229898452759, + "num_tokens": 119957275.0, + "step": 3087 + }, + { + "epoch": 0.3928253402874952, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2874603271484375e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.661295413970947, + "learning_rate": 1e-06, + "loss": 0.3874, + "mean_token_accuracy": 0.8710142970085144, + "num_tokens": 119996615.0, + "step": 3088 + }, + { + "epoch": 0.39295255056608575, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2874603271484375e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.726644992828369, + "learning_rate": 1e-06, + "loss": 0.436, + "mean_token_accuracy": 0.855154275894165, + "num_tokens": 120036650.0, + "step": 3089 + }, + { + "epoch": 0.3930797608446763, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2874603271484375e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.610525131225586, + "learning_rate": 1e-06, + "loss": 0.3856, + "mean_token_accuracy": 0.8735067844390869, + "num_tokens": 120080864.0, + "step": 3090 + }, + { + "epoch": 0.39320697112326675, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.2874603271484375e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.541064262390137, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8819501399993896, + "num_tokens": 120118548.0, + "step": 3091 + }, + { + "epoch": 0.3933341814018573, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2993812561035156e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.714785099029541, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8816219568252563, + "num_tokens": 120153054.0, + "step": 3092 + }, + { + "epoch": 0.3934613916804478, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2993812561035156e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.703688144683838, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8957096338272095, + "num_tokens": 120182515.0, + "step": 3093 + }, + { + "epoch": 0.3935886019590383, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.2993812561035156e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6030354499816895, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8823235034942627, + "num_tokens": 120222111.0, + "step": 3094 + }, + { + "epoch": 0.3937158122376288, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.3053417205810547e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6868791580200195, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.881904125213623, + "num_tokens": 120262418.0, + "step": 3095 + }, + { + "epoch": 0.39384302251621933, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3053417205810547e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.716637134552002, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8807845115661621, + "num_tokens": 120300003.0, + "step": 3096 + }, + { + "epoch": 0.3939702327948098, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3053417205810547e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.684360027313232, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.884231448173523, + "num_tokens": 120338966.0, + "step": 3097 + }, + { + "epoch": 0.39409744307340033, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.3053417205810547e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.7602128982543945, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.873143196105957, + "num_tokens": 120373893.0, + "step": 3098 + }, + { + "epoch": 0.39422465335199086, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.3053417205810547e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.731720924377441, + "learning_rate": 1e-06, + "loss": 0.3886, + "mean_token_accuracy": 0.8714121580123901, + "num_tokens": 120408401.0, + "step": 3099 + }, + { + "epoch": 0.39435186363058133, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.3053417205810547e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.7704973220825195, + "learning_rate": 1e-06, + "loss": 0.4103, + "mean_token_accuracy": 0.8596973419189453, + "num_tokens": 120445821.0, + "step": 3100 + }, + { + "epoch": 0.39447907390917186, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.3053417205810547e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.642088890075684, + "learning_rate": 1e-06, + "loss": 0.414, + "mean_token_accuracy": 0.8615453839302063, + "num_tokens": 120491778.0, + "step": 3101 + }, + { + "epoch": 0.3946062841877624, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3053417205810547e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.599175453186035, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8859700560569763, + "num_tokens": 120534137.0, + "step": 3102 + }, + { + "epoch": 0.39473349446635286, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3053417205810547e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.700016021728516, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8764334917068481, + "num_tokens": 120573782.0, + "step": 3103 + }, + { + "epoch": 0.3948607047449434, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3053417205810547e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.958343029022217, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8712210655212402, + "num_tokens": 120613713.0, + "step": 3104 + }, + { + "epoch": 0.3949879150235339, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3053417205810547e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.685718059539795, + "learning_rate": 1e-06, + "loss": 0.4013, + "mean_token_accuracy": 0.8670451045036316, + "num_tokens": 120654739.0, + "step": 3105 + }, + { + "epoch": 0.3951151253021244, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3053417205810547e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.633184432983398, + "learning_rate": 1e-06, + "loss": 0.3918, + "mean_token_accuracy": 0.8712332248687744, + "num_tokens": 120698458.0, + "step": 3106 + }, + { + "epoch": 0.3952423355807149, + "ewc_loss": 0.00921630859375, + "ewc_loss_diag": 1.3053417205810547e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.737337589263916, + "learning_rate": 1e-06, + "loss": 0.4007, + "mean_token_accuracy": 0.8639892339706421, + "num_tokens": 120738147.0, + "step": 3107 + }, + { + "epoch": 0.39536954585930545, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3113021850585938e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.853492736816406, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8714007139205933, + "num_tokens": 120767195.0, + "step": 3108 + }, + { + "epoch": 0.3954967561378959, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3113021850585938e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.718747138977051, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8792213201522827, + "num_tokens": 120803017.0, + "step": 3109 + }, + { + "epoch": 0.39562396641648645, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3113021850585938e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6559319496154785, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8778502941131592, + "num_tokens": 120843458.0, + "step": 3110 + }, + { + "epoch": 0.395751176695077, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3113021850585938e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.646659851074219, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8734617233276367, + "num_tokens": 120886468.0, + "step": 3111 + }, + { + "epoch": 0.39587838697366745, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3113021850585938e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.72274112701416, + "learning_rate": 1e-06, + "loss": 0.3904, + "mean_token_accuracy": 0.8655915260314941, + "num_tokens": 120924213.0, + "step": 3112 + }, + { + "epoch": 0.396005597252258, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3113021850585938e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.653572082519531, + "learning_rate": 1e-06, + "loss": 0.388, + "mean_token_accuracy": 0.8725413680076599, + "num_tokens": 120965263.0, + "step": 3113 + }, + { + "epoch": 0.3961328075308485, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3113021850585938e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.713701248168945, + "learning_rate": 1e-06, + "loss": 0.3842, + "mean_token_accuracy": 0.8727690577507019, + "num_tokens": 121004541.0, + "step": 3114 + }, + { + "epoch": 0.396260017809439, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3113021850585938e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6933979988098145, + "learning_rate": 1e-06, + "loss": 0.3772, + "mean_token_accuracy": 0.8745031356811523, + "num_tokens": 121044868.0, + "step": 3115 + }, + { + "epoch": 0.3963872280880295, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3172626495361328e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.904717445373535, + "learning_rate": 1e-06, + "loss": 0.3885, + "mean_token_accuracy": 0.8685590028762817, + "num_tokens": 121076490.0, + "step": 3116 + }, + { + "epoch": 0.39651443836662004, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3172626495361328e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.781449317932129, + "learning_rate": 1e-06, + "loss": 0.4271, + "mean_token_accuracy": 0.8594447374343872, + "num_tokens": 121119004.0, + "step": 3117 + }, + { + "epoch": 0.3966416486452105, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3172626495361328e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.743473529815674, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8746787309646606, + "num_tokens": 121156074.0, + "step": 3118 + }, + { + "epoch": 0.39676885892380104, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3172626495361328e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6310296058654785, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8859639167785645, + "num_tokens": 121196208.0, + "step": 3119 + }, + { + "epoch": 0.39689606920239157, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3172626495361328e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.745672225952148, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8759256601333618, + "num_tokens": 121232543.0, + "step": 3120 + }, + { + "epoch": 0.39702327948098204, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3172626495361328e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.815888404846191, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8767601251602173, + "num_tokens": 121264778.0, + "step": 3121 + }, + { + "epoch": 0.39715048975957257, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3172626495361328e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.659511089324951, + "learning_rate": 1e-06, + "loss": 0.4018, + "mean_token_accuracy": 0.8661797046661377, + "num_tokens": 121307064.0, + "step": 3122 + }, + { + "epoch": 0.3972777000381631, + "ewc_loss": 0.00927734375, + "ewc_loss_diag": 1.3172626495361328e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.72232723236084, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8656400442123413, + "num_tokens": 121346991.0, + "step": 3123 + }, + { + "epoch": 0.39740491031675357, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3172626495361328e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.793135643005371, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.8702952265739441, + "num_tokens": 121379367.0, + "step": 3124 + }, + { + "epoch": 0.3975321205953441, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3172626495361328e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.720696449279785, + "learning_rate": 1e-06, + "loss": 0.4015, + "mean_token_accuracy": 0.8606759309768677, + "num_tokens": 121416265.0, + "step": 3125 + }, + { + "epoch": 0.3976593308739346, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3172626495361328e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 8.044695854187012, + "learning_rate": 1e-06, + "loss": 0.4193, + "mean_token_accuracy": 0.8612881898880005, + "num_tokens": 121453971.0, + "step": 3126 + }, + { + "epoch": 0.3977865411525251, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3172626495361328e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.661313533782959, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8614751100540161, + "num_tokens": 121494603.0, + "step": 3127 + }, + { + "epoch": 0.3979137514311156, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3232231140136719e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.69551420211792, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8735241293907166, + "num_tokens": 121530652.0, + "step": 3128 + }, + { + "epoch": 0.39804096170970615, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3232231140136719e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.766704559326172, + "learning_rate": 1e-06, + "loss": 0.4101, + "mean_token_accuracy": 0.8632015585899353, + "num_tokens": 121569849.0, + "step": 3129 + }, + { + "epoch": 0.3981681719882967, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3232231140136719e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.7020111083984375, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8835176229476929, + "num_tokens": 121601491.0, + "step": 3130 + }, + { + "epoch": 0.39829538226688715, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3232231140136719e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.74924373626709, + "learning_rate": 1e-06, + "loss": 0.4323, + "mean_token_accuracy": 0.8536419868469238, + "num_tokens": 121641702.0, + "step": 3131 + }, + { + "epoch": 0.3984225925454777, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3232231140136719e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.619556427001953, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8809480667114258, + "num_tokens": 121685902.0, + "step": 3132 + }, + { + "epoch": 0.3985498028240682, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3232231140136719e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.853942394256592, + "learning_rate": 1e-06, + "loss": 0.4224, + "mean_token_accuracy": 0.8567142486572266, + "num_tokens": 121720909.0, + "step": 3133 + }, + { + "epoch": 0.3986770131026587, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3232231140136719e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.770961284637451, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8710017800331116, + "num_tokens": 121756109.0, + "step": 3134 + }, + { + "epoch": 0.3988042233812492, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3232231140136719e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.643868446350098, + "learning_rate": 1e-06, + "loss": 0.4122, + "mean_token_accuracy": 0.864181399345398, + "num_tokens": 121798883.0, + "step": 3135 + }, + { + "epoch": 0.39893143365983974, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3232231140136719e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.704436302185059, + "learning_rate": 1e-06, + "loss": 0.4015, + "mean_token_accuracy": 0.8624091744422913, + "num_tokens": 121836084.0, + "step": 3136 + }, + { + "epoch": 0.3990586439384302, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3232231140136719e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.610185146331787, + "learning_rate": 1e-06, + "loss": 0.4021, + "mean_token_accuracy": 0.8633126020431519, + "num_tokens": 121881809.0, + "step": 3137 + }, + { + "epoch": 0.39918585421702074, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3232231140136719e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.737844467163086, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.892788290977478, + "num_tokens": 121916668.0, + "step": 3138 + }, + { + "epoch": 0.39931306449561127, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3232231140136719e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.71936559677124, + "learning_rate": 1e-06, + "loss": 0.4189, + "mean_token_accuracy": 0.8629364967346191, + "num_tokens": 121955581.0, + "step": 3139 + }, + { + "epoch": 0.39944027477420174, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.329183578491211e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.759566783905029, + "learning_rate": 1e-06, + "loss": 0.3992, + "mean_token_accuracy": 0.8672130107879639, + "num_tokens": 121993365.0, + "step": 3140 + }, + { + "epoch": 0.39956748505279227, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.329183578491211e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.709515571594238, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.872622013092041, + "num_tokens": 122031846.0, + "step": 3141 + }, + { + "epoch": 0.3996946953313828, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.329183578491211e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.803227424621582, + "learning_rate": 1e-06, + "loss": 0.417, + "mean_token_accuracy": 0.861477255821228, + "num_tokens": 122063710.0, + "step": 3142 + }, + { + "epoch": 0.39982190560997327, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.329183578491211e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.685929298400879, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8713760375976562, + "num_tokens": 122100135.0, + "step": 3143 + }, + { + "epoch": 0.3999491158885638, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.329183578491211e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.6146674156188965, + "learning_rate": 1e-06, + "loss": 0.3998, + "mean_token_accuracy": 0.8684097528457642, + "num_tokens": 122142212.0, + "step": 3144 + }, + { + "epoch": 0.40007632616715433, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.329183578491211e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.847437381744385, + "learning_rate": 1e-06, + "loss": 0.4032, + "mean_token_accuracy": 0.8652157783508301, + "num_tokens": 122174658.0, + "step": 3145 + }, + { + "epoch": 0.4002035364457448, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.329183578491211e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.736194133758545, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8866147994995117, + "num_tokens": 122209995.0, + "step": 3146 + }, + { + "epoch": 0.40033074672433533, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.329183578491211e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.731478214263916, + "learning_rate": 1e-06, + "loss": 0.3915, + "mean_token_accuracy": 0.8661514520645142, + "num_tokens": 122245869.0, + "step": 3147 + }, + { + "epoch": 0.40045795700292586, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.33514404296875e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.620733737945557, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8914295434951782, + "num_tokens": 122287838.0, + "step": 3148 + }, + { + "epoch": 0.40058516728151633, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.33514404296875e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.938651084899902, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8793537616729736, + "num_tokens": 122324289.0, + "step": 3149 + }, + { + "epoch": 0.40071237756010686, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.33514404296875e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.677181243896484, + "learning_rate": 1e-06, + "loss": 0.4141, + "mean_token_accuracy": 0.8608394265174866, + "num_tokens": 122369349.0, + "step": 3150 + }, + { + "epoch": 0.4008395878386974, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.33514404296875e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.674875259399414, + "learning_rate": 1e-06, + "loss": 0.4075, + "mean_token_accuracy": 0.8632887601852417, + "num_tokens": 122416066.0, + "step": 3151 + }, + { + "epoch": 0.40096679811728786, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.758518218994141, + "learning_rate": 1e-06, + "loss": 0.3776, + "mean_token_accuracy": 0.8733959794044495, + "num_tokens": 122452893.0, + "step": 3152 + }, + { + "epoch": 0.4010940083958784, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.696352958679199, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.871963381767273, + "num_tokens": 122492350.0, + "step": 3153 + }, + { + "epoch": 0.4012212186744689, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.783981800079346, + "learning_rate": 1e-06, + "loss": 0.4098, + "mean_token_accuracy": 0.862594723701477, + "num_tokens": 122528007.0, + "step": 3154 + }, + { + "epoch": 0.4013484289530594, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.647502899169922, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8760496377944946, + "num_tokens": 122571295.0, + "step": 3155 + }, + { + "epoch": 0.4014756392316499, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.6518707275390625, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8764382600784302, + "num_tokens": 122615908.0, + "step": 3156 + }, + { + "epoch": 0.40160284951024044, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.709282875061035, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8790818452835083, + "num_tokens": 122653243.0, + "step": 3157 + }, + { + "epoch": 0.4017300597888309, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.738772392272949, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8771423101425171, + "num_tokens": 122691038.0, + "step": 3158 + }, + { + "epoch": 0.40185727006742145, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.679442882537842, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8771467208862305, + "num_tokens": 122727425.0, + "step": 3159 + }, + { + "epoch": 0.401984480346012, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.845695495605469, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.867974042892456, + "num_tokens": 122759133.0, + "step": 3160 + }, + { + "epoch": 0.40211169062460245, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.711937427520752, + "learning_rate": 1e-06, + "loss": 0.389, + "mean_token_accuracy": 0.8667308688163757, + "num_tokens": 122795664.0, + "step": 3161 + }, + { + "epoch": 0.402238900903193, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.630388259887695, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8811789751052856, + "num_tokens": 122836705.0, + "step": 3162 + }, + { + "epoch": 0.4023661111817835, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.784163475036621, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8795500993728638, + "num_tokens": 122871821.0, + "step": 3163 + }, + { + "epoch": 0.402493321460374, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.7678937911987305, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.8725569248199463, + "num_tokens": 122910247.0, + "step": 3164 + }, + { + "epoch": 0.4026205317389645, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.797423839569092, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8784812688827515, + "num_tokens": 122941124.0, + "step": 3165 + }, + { + "epoch": 0.40274774201755503, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.854308128356934, + "learning_rate": 1e-06, + "loss": 0.3779, + "mean_token_accuracy": 0.8741908669471741, + "num_tokens": 122973009.0, + "step": 3166 + }, + { + "epoch": 0.4028749522961455, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.678111553192139, + "learning_rate": 1e-06, + "loss": 0.3874, + "mean_token_accuracy": 0.8693327903747559, + "num_tokens": 123011675.0, + "step": 3167 + }, + { + "epoch": 0.40300216257473603, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.341104507446289e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.800445556640625, + "learning_rate": 1e-06, + "loss": 0.3838, + "mean_token_accuracy": 0.8694465756416321, + "num_tokens": 123053133.0, + "step": 3168 + }, + { + "epoch": 0.40312937285332656, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3470649719238281e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.722986221313477, + "learning_rate": 1e-06, + "loss": 0.4235, + "mean_token_accuracy": 0.8566194772720337, + "num_tokens": 123090785.0, + "step": 3169 + }, + { + "epoch": 0.40325658313191703, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3470649719238281e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.772861480712891, + "learning_rate": 1e-06, + "loss": 0.3868, + "mean_token_accuracy": 0.8727854490280151, + "num_tokens": 123124106.0, + "step": 3170 + }, + { + "epoch": 0.40338379341050756, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3530254364013672e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.755272388458252, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8753278851509094, + "num_tokens": 123163836.0, + "step": 3171 + }, + { + "epoch": 0.4035110036890981, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3530254364013672e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.644599914550781, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8855164051055908, + "num_tokens": 123200151.0, + "step": 3172 + }, + { + "epoch": 0.40363821396768856, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3530254364013672e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.638824939727783, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8662248253822327, + "num_tokens": 123241402.0, + "step": 3173 + }, + { + "epoch": 0.4037654242462791, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3470649719238281e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.724236488342285, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8815546631813049, + "num_tokens": 123275472.0, + "step": 3174 + }, + { + "epoch": 0.4038926345248696, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3470649719238281e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.698710918426514, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.879393458366394, + "num_tokens": 123312717.0, + "step": 3175 + }, + { + "epoch": 0.4040198448034601, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3470649719238281e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.6289496421813965, + "learning_rate": 1e-06, + "loss": 0.402, + "mean_token_accuracy": 0.872958779335022, + "num_tokens": 123360838.0, + "step": 3176 + }, + { + "epoch": 0.4041470550820506, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3470649719238281e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.690860748291016, + "learning_rate": 1e-06, + "loss": 0.3911, + "mean_token_accuracy": 0.8697324991226196, + "num_tokens": 123400686.0, + "step": 3177 + }, + { + "epoch": 0.40427426536064115, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3470649719238281e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.740927696228027, + "learning_rate": 1e-06, + "loss": 0.4449, + "mean_token_accuracy": 0.8490606546401978, + "num_tokens": 123440549.0, + "step": 3178 + }, + { + "epoch": 0.4044014756392316, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3470649719238281e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.761771202087402, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8851525187492371, + "num_tokens": 123472682.0, + "step": 3179 + }, + { + "epoch": 0.40452868591782215, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3470649719238281e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.635446548461914, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.883514404296875, + "num_tokens": 123508534.0, + "step": 3180 + }, + { + "epoch": 0.4046558961964127, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3589859008789062e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.715795516967773, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.8695764541625977, + "num_tokens": 123546561.0, + "step": 3181 + }, + { + "epoch": 0.4047831064750032, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3589859008789062e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.642167568206787, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8719242811203003, + "num_tokens": 123584596.0, + "step": 3182 + }, + { + "epoch": 0.4049103167535937, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3589859008789062e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.693467140197754, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.8710196018218994, + "num_tokens": 123627782.0, + "step": 3183 + }, + { + "epoch": 0.4050375270321842, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3589859008789062e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.665849685668945, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8783038854598999, + "num_tokens": 123665122.0, + "step": 3184 + }, + { + "epoch": 0.40516473731077474, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3589859008789062e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.740371227264404, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.8708139657974243, + "num_tokens": 123700435.0, + "step": 3185 + }, + { + "epoch": 0.4052919475893652, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3589859008789062e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.713714122772217, + "learning_rate": 1e-06, + "loss": 0.392, + "mean_token_accuracy": 0.869400143623352, + "num_tokens": 123736059.0, + "step": 3186 + }, + { + "epoch": 0.40541915786795574, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3589859008789062e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.78961181640625, + "learning_rate": 1e-06, + "loss": 0.4302, + "mean_token_accuracy": 0.8551905155181885, + "num_tokens": 123769118.0, + "step": 3187 + }, + { + "epoch": 0.40554636814654627, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3589859008789062e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.765363693237305, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8748624324798584, + "num_tokens": 123807087.0, + "step": 3188 + }, + { + "epoch": 0.40567357842513674, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3589859008789062e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.736077308654785, + "learning_rate": 1e-06, + "loss": 0.4258, + "mean_token_accuracy": 0.859316885471344, + "num_tokens": 123841845.0, + "step": 3189 + }, + { + "epoch": 0.40580078870372727, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3589859008789062e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.879588603973389, + "learning_rate": 1e-06, + "loss": 0.3946, + "mean_token_accuracy": 0.8647505044937134, + "num_tokens": 123873924.0, + "step": 3190 + }, + { + "epoch": 0.4059279989823178, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3589859008789062e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.653599739074707, + "learning_rate": 1e-06, + "loss": 0.3858, + "mean_token_accuracy": 0.8719764947891235, + "num_tokens": 123915702.0, + "step": 3191 + }, + { + "epoch": 0.40605520926090827, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3709068298339844e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.754917144775391, + "learning_rate": 1e-06, + "loss": 0.4159, + "mean_token_accuracy": 0.8617808222770691, + "num_tokens": 123955030.0, + "step": 3192 + }, + { + "epoch": 0.4061824195394988, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3709068298339844e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.788777828216553, + "learning_rate": 1e-06, + "loss": 0.421, + "mean_token_accuracy": 0.8601059317588806, + "num_tokens": 123993925.0, + "step": 3193 + }, + { + "epoch": 0.4063096298180893, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3709068298339844e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.717000961303711, + "learning_rate": 1e-06, + "loss": 0.4025, + "mean_token_accuracy": 0.8699109554290771, + "num_tokens": 124034899.0, + "step": 3194 + }, + { + "epoch": 0.4064368400966798, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3709068298339844e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.772982120513916, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8749898672103882, + "num_tokens": 124074500.0, + "step": 3195 + }, + { + "epoch": 0.4065640503752703, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3709068298339844e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.672375679016113, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8786469101905823, + "num_tokens": 124113756.0, + "step": 3196 + }, + { + "epoch": 0.40669126065386085, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3709068298339844e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.764630317687988, + "learning_rate": 1e-06, + "loss": 0.4371, + "mean_token_accuracy": 0.8530705571174622, + "num_tokens": 124153171.0, + "step": 3197 + }, + { + "epoch": 0.4068184709324513, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.3709068298339844e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.674885272979736, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8835156559944153, + "num_tokens": 124193145.0, + "step": 3198 + }, + { + "epoch": 0.40694568121104185, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3709068298339844e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.567685604095459, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8800168037414551, + "num_tokens": 124239940.0, + "step": 3199 + }, + { + "epoch": 0.4070728914896324, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3768672943115234e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.716878890991211, + "learning_rate": 1e-06, + "loss": 0.3576, + "mean_token_accuracy": 0.8789811134338379, + "num_tokens": 124282848.0, + "step": 3200 + }, + { + "epoch": 0.40720010176822286, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3768672943115234e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.719679832458496, + "learning_rate": 1e-06, + "loss": 0.386, + "mean_token_accuracy": 0.869044303894043, + "num_tokens": 124325649.0, + "step": 3201 + }, + { + "epoch": 0.4073273120468134, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3768672943115234e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.812105655670166, + "learning_rate": 1e-06, + "loss": 0.39, + "mean_token_accuracy": 0.870378851890564, + "num_tokens": 124359170.0, + "step": 3202 + }, + { + "epoch": 0.4074545223254039, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3828277587890625e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.738000869750977, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8827114701271057, + "num_tokens": 124391511.0, + "step": 3203 + }, + { + "epoch": 0.4075817326039944, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3828277587890625e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.906161785125732, + "learning_rate": 1e-06, + "loss": 0.392, + "mean_token_accuracy": 0.8667324781417847, + "num_tokens": 124423248.0, + "step": 3204 + }, + { + "epoch": 0.4077089428825849, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.3828277587890625e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.738396644592285, + "learning_rate": 1e-06, + "loss": 0.3733, + "mean_token_accuracy": 0.8748890161514282, + "num_tokens": 124461339.0, + "step": 3205 + }, + { + "epoch": 0.40783615316117544, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3828277587890625e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.591280460357666, + "learning_rate": 1e-06, + "loss": 0.4028, + "mean_token_accuracy": 0.8646025657653809, + "num_tokens": 124506811.0, + "step": 3206 + }, + { + "epoch": 0.4079633634397659, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.3828277587890625e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.674996852874756, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8777810335159302, + "num_tokens": 124544255.0, + "step": 3207 + }, + { + "epoch": 0.40809057371835644, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3828277587890625e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.764257431030273, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8822619915008545, + "num_tokens": 124577788.0, + "step": 3208 + }, + { + "epoch": 0.40821778399694697, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3828277587890625e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.682801723480225, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8779065608978271, + "num_tokens": 124614982.0, + "step": 3209 + }, + { + "epoch": 0.40834499427553744, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3828277587890625e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.756289005279541, + "learning_rate": 1e-06, + "loss": 0.4011, + "mean_token_accuracy": 0.8636101484298706, + "num_tokens": 124653428.0, + "step": 3210 + }, + { + "epoch": 0.40847220455412797, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3828277587890625e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.657827377319336, + "learning_rate": 1e-06, + "loss": 0.4255, + "mean_token_accuracy": 0.8563690185546875, + "num_tokens": 124693732.0, + "step": 3211 + }, + { + "epoch": 0.4085994148327185, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.3828277587890625e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.702146530151367, + "learning_rate": 1e-06, + "loss": 0.3967, + "mean_token_accuracy": 0.8654531240463257, + "num_tokens": 124735203.0, + "step": 3212 + }, + { + "epoch": 0.40872662511130897, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3828277587890625e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.719346523284912, + "learning_rate": 1e-06, + "loss": 0.4231, + "mean_token_accuracy": 0.858189582824707, + "num_tokens": 124774663.0, + "step": 3213 + }, + { + "epoch": 0.4088538353898995, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3828277587890625e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.6880292892456055, + "learning_rate": 1e-06, + "loss": 0.3843, + "mean_token_accuracy": 0.8708609342575073, + "num_tokens": 124819352.0, + "step": 3214 + }, + { + "epoch": 0.40898104566849003, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3828277587890625e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.741932392120361, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8745250701904297, + "num_tokens": 124856739.0, + "step": 3215 + }, + { + "epoch": 0.4091082559470805, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.687890529632568, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8806113600730896, + "num_tokens": 124897944.0, + "step": 3216 + }, + { + "epoch": 0.40923546622567103, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.855223178863525, + "learning_rate": 1e-06, + "loss": 0.3832, + "mean_token_accuracy": 0.8714690208435059, + "num_tokens": 124929607.0, + "step": 3217 + }, + { + "epoch": 0.40936267650426156, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.7046895027160645, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8690997362136841, + "num_tokens": 124970624.0, + "step": 3218 + }, + { + "epoch": 0.40948988678285203, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.673835754394531, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8771489262580872, + "num_tokens": 125010064.0, + "step": 3219 + }, + { + "epoch": 0.40961709706144256, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.729246139526367, + "learning_rate": 1e-06, + "loss": 0.4189, + "mean_token_accuracy": 0.8573650121688843, + "num_tokens": 125050655.0, + "step": 3220 + }, + { + "epoch": 0.4097443073400331, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.705826759338379, + "learning_rate": 1e-06, + "loss": 0.3913, + "mean_token_accuracy": 0.8662851452827454, + "num_tokens": 125092960.0, + "step": 3221 + }, + { + "epoch": 0.40987151761862356, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.733424663543701, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8797613382339478, + "num_tokens": 125129844.0, + "step": 3222 + }, + { + "epoch": 0.4099987278972141, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.735720157623291, + "learning_rate": 1e-06, + "loss": 0.4416, + "mean_token_accuracy": 0.850427508354187, + "num_tokens": 125172421.0, + "step": 3223 + }, + { + "epoch": 0.4101259381758046, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.705574989318848, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8764785528182983, + "num_tokens": 125213064.0, + "step": 3224 + }, + { + "epoch": 0.4102531484543951, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.6837873458862305, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8749555349349976, + "num_tokens": 125252782.0, + "step": 3225 + }, + { + "epoch": 0.4103803587329856, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.771235466003418, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.8689647316932678, + "num_tokens": 125292432.0, + "step": 3226 + }, + { + "epoch": 0.41050756901157615, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.665942192077637, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8715625405311584, + "num_tokens": 125334481.0, + "step": 3227 + }, + { + "epoch": 0.4106347792901666, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.816829681396484, + "learning_rate": 1e-06, + "loss": 0.4694, + "mean_token_accuracy": 0.8405201435089111, + "num_tokens": 125375184.0, + "step": 3228 + }, + { + "epoch": 0.41076198956875715, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.8228840827941895, + "learning_rate": 1e-06, + "loss": 0.4034, + "mean_token_accuracy": 0.8639787435531616, + "num_tokens": 125410311.0, + "step": 3229 + }, + { + "epoch": 0.4108891998473477, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.3887882232666016e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.823843002319336, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8879587650299072, + "num_tokens": 125442479.0, + "step": 3230 + }, + { + "epoch": 0.4110164101259382, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3947486877441406e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.747286796569824, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8750220537185669, + "num_tokens": 125481754.0, + "step": 3231 + }, + { + "epoch": 0.4111436204045287, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3947486877441406e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.765932083129883, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8775305151939392, + "num_tokens": 125514930.0, + "step": 3232 + }, + { + "epoch": 0.4112708306831192, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3947486877441406e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.762213230133057, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.8686710596084595, + "num_tokens": 125550526.0, + "step": 3233 + }, + { + "epoch": 0.41139804096170973, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3947486877441406e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.649645805358887, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.879780650138855, + "num_tokens": 125594073.0, + "step": 3234 + }, + { + "epoch": 0.4115252512403002, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3947486877441406e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.869334697723389, + "learning_rate": 1e-06, + "loss": 0.3984, + "mean_token_accuracy": 0.8683335781097412, + "num_tokens": 125629583.0, + "step": 3235 + }, + { + "epoch": 0.41165246151889073, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.3947486877441406e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.689202308654785, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8748350143432617, + "num_tokens": 125669052.0, + "step": 3236 + }, + { + "epoch": 0.41177967179748126, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.3947486877441406e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.71973991394043, + "learning_rate": 1e-06, + "loss": 0.3843, + "mean_token_accuracy": 0.8726929426193237, + "num_tokens": 125707789.0, + "step": 3237 + }, + { + "epoch": 0.41190688207607173, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3947486877441406e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.6681671142578125, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.8764762878417969, + "num_tokens": 125751715.0, + "step": 3238 + }, + { + "epoch": 0.41203409235466226, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.3947486877441406e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.861009120941162, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.8692227005958557, + "num_tokens": 125782762.0, + "step": 3239 + }, + { + "epoch": 0.4121613026332528, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4007091522216797e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.6662187576293945, + "learning_rate": 1e-06, + "loss": 0.3902, + "mean_token_accuracy": 0.8672680854797363, + "num_tokens": 125827362.0, + "step": 3240 + }, + { + "epoch": 0.41228851291184326, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4007091522216797e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.667232990264893, + "learning_rate": 1e-06, + "loss": 0.4032, + "mean_token_accuracy": 0.864235520362854, + "num_tokens": 125869785.0, + "step": 3241 + }, + { + "epoch": 0.4124157231904338, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4007091522216797e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.692651748657227, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8783100843429565, + "num_tokens": 125909384.0, + "step": 3242 + }, + { + "epoch": 0.4125429334690243, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4007091522216797e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.689728736877441, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.8715893030166626, + "num_tokens": 125953547.0, + "step": 3243 + }, + { + "epoch": 0.4126701437476148, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4007091522216797e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.778454780578613, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8804430961608887, + "num_tokens": 125991712.0, + "step": 3244 + }, + { + "epoch": 0.4127973540262053, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4066696166992188e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.692091464996338, + "learning_rate": 1e-06, + "loss": 0.403, + "mean_token_accuracy": 0.8675812482833862, + "num_tokens": 126037038.0, + "step": 3245 + }, + { + "epoch": 0.41292456430479585, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.705173015594482, + "learning_rate": 1e-06, + "loss": 0.3738, + "mean_token_accuracy": 0.8733969926834106, + "num_tokens": 126078925.0, + "step": 3246 + }, + { + "epoch": 0.4130517745833863, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.801122665405273, + "learning_rate": 1e-06, + "loss": 0.4168, + "mean_token_accuracy": 0.8595833778381348, + "num_tokens": 126124133.0, + "step": 3247 + }, + { + "epoch": 0.41317898486197685, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.693375110626221, + "learning_rate": 1e-06, + "loss": 0.3999, + "mean_token_accuracy": 0.8654047846794128, + "num_tokens": 126165988.0, + "step": 3248 + }, + { + "epoch": 0.4133061951405674, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.775949954986572, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8738653659820557, + "num_tokens": 126203662.0, + "step": 3249 + }, + { + "epoch": 0.41343340541915785, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.7810540199279785, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8833067417144775, + "num_tokens": 126235657.0, + "step": 3250 + }, + { + "epoch": 0.4135606156977484, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.727162837982178, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8910858035087585, + "num_tokens": 126270716.0, + "step": 3251 + }, + { + "epoch": 0.4136878259763389, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.6577677726745605, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8765504360198975, + "num_tokens": 126310854.0, + "step": 3252 + }, + { + "epoch": 0.4138150362549294, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.784050941467285, + "learning_rate": 1e-06, + "loss": 0.3999, + "mean_token_accuracy": 0.870632529258728, + "num_tokens": 126352046.0, + "step": 3253 + }, + { + "epoch": 0.4139422465335199, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.724307537078857, + "learning_rate": 1e-06, + "loss": 0.3952, + "mean_token_accuracy": 0.8660860061645508, + "num_tokens": 126397503.0, + "step": 3254 + }, + { + "epoch": 0.41406945681211044, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.728774547576904, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8784009218215942, + "num_tokens": 126435598.0, + "step": 3255 + }, + { + "epoch": 0.4141966670907009, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.667296886444092, + "learning_rate": 1e-06, + "loss": 0.3818, + "mean_token_accuracy": 0.8728948831558228, + "num_tokens": 126481735.0, + "step": 3256 + }, + { + "epoch": 0.41432387736929144, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.877628803253174, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8784782290458679, + "num_tokens": 126515519.0, + "step": 3257 + }, + { + "epoch": 0.41445108764788197, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.743746280670166, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8726966381072998, + "num_tokens": 126555702.0, + "step": 3258 + }, + { + "epoch": 0.41457829792647244, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.711353778839111, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8931093215942383, + "num_tokens": 126593279.0, + "step": 3259 + }, + { + "epoch": 0.41470550820506297, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.645970821380615, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8795956373214722, + "num_tokens": 126636175.0, + "step": 3260 + }, + { + "epoch": 0.4148327184836535, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.430511474609375e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.810831069946289, + "learning_rate": 1e-06, + "loss": 0.4022, + "mean_token_accuracy": 0.8699337244033813, + "num_tokens": 126671872.0, + "step": 3261 + }, + { + "epoch": 0.41495992876224397, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.791737079620361, + "learning_rate": 1e-06, + "loss": 0.4133, + "mean_token_accuracy": 0.8611372113227844, + "num_tokens": 126712868.0, + "step": 3262 + }, + { + "epoch": 0.4150871390408345, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.726594924926758, + "learning_rate": 1e-06, + "loss": 0.4063, + "mean_token_accuracy": 0.864001989364624, + "num_tokens": 126752348.0, + "step": 3263 + }, + { + "epoch": 0.415214349319425, + "ewc_loss": 0.00933837890625, + "ewc_loss_diag": 1.4185905456542969e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.874521732330322, + "learning_rate": 1e-06, + "loss": 0.3867, + "mean_token_accuracy": 0.8667620420455933, + "num_tokens": 126783853.0, + "step": 3264 + }, + { + "epoch": 0.4153415595980155, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.436471939086914e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.722901344299316, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8732614517211914, + "num_tokens": 126821107.0, + "step": 3265 + }, + { + "epoch": 0.415468769876606, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.436471939086914e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.807326316833496, + "learning_rate": 1e-06, + "loss": 0.4157, + "mean_token_accuracy": 0.863103985786438, + "num_tokens": 126858620.0, + "step": 3266 + }, + { + "epoch": 0.41559598015519655, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.430511474609375e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.680367946624756, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8761271238327026, + "num_tokens": 126902283.0, + "step": 3267 + }, + { + "epoch": 0.415723190433787, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.436471939086914e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.753475666046143, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.87223219871521, + "num_tokens": 126941632.0, + "step": 3268 + }, + { + "epoch": 0.41585040071237755, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.436471939086914e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.855360984802246, + "learning_rate": 1e-06, + "loss": 0.4154, + "mean_token_accuracy": 0.8646553158760071, + "num_tokens": 126980350.0, + "step": 3269 + }, + { + "epoch": 0.4159776109909681, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.436471939086914e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.727187156677246, + "learning_rate": 1e-06, + "loss": 0.398, + "mean_token_accuracy": 0.868039071559906, + "num_tokens": 127019089.0, + "step": 3270 + }, + { + "epoch": 0.41610482126955856, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.436471939086914e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.759121417999268, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8857330083847046, + "num_tokens": 127056134.0, + "step": 3271 + }, + { + "epoch": 0.4162320315481491, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.436471939086914e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.9381818771362305, + "learning_rate": 1e-06, + "loss": 0.4298, + "mean_token_accuracy": 0.8542389869689941, + "num_tokens": 127087965.0, + "step": 3272 + }, + { + "epoch": 0.4163592418267396, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.436471939086914e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.699200630187988, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8813925385475159, + "num_tokens": 127130304.0, + "step": 3273 + }, + { + "epoch": 0.4164864521053301, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4424324035644531e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.7739787101745605, + "learning_rate": 1e-06, + "loss": 0.3898, + "mean_token_accuracy": 0.8686458468437195, + "num_tokens": 127169097.0, + "step": 3274 + }, + { + "epoch": 0.4166136623839206, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.83171272277832, + "learning_rate": 1e-06, + "loss": 0.4684, + "mean_token_accuracy": 0.8472872972488403, + "num_tokens": 127206823.0, + "step": 3275 + }, + { + "epoch": 0.41674087266251114, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.665141582489014, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8866804242134094, + "num_tokens": 127246517.0, + "step": 3276 + }, + { + "epoch": 0.4168680829411016, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.761937618255615, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8739891052246094, + "num_tokens": 127287861.0, + "step": 3277 + }, + { + "epoch": 0.41699529321969214, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.723146915435791, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8833029270172119, + "num_tokens": 127329192.0, + "step": 3278 + }, + { + "epoch": 0.41712250349828267, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.798123359680176, + "learning_rate": 1e-06, + "loss": 0.4009, + "mean_token_accuracy": 0.8671257495880127, + "num_tokens": 127366095.0, + "step": 3279 + }, + { + "epoch": 0.4172497137768732, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.862368583679199, + "learning_rate": 1e-06, + "loss": 0.4296, + "mean_token_accuracy": 0.8552044630050659, + "num_tokens": 127403879.0, + "step": 3280 + }, + { + "epoch": 0.41737692405546367, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.79529333114624, + "learning_rate": 1e-06, + "loss": 0.4372, + "mean_token_accuracy": 0.8546833395957947, + "num_tokens": 127443926.0, + "step": 3281 + }, + { + "epoch": 0.4175041343340542, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.655828475952148, + "learning_rate": 1e-06, + "loss": 0.3902, + "mean_token_accuracy": 0.8719679117202759, + "num_tokens": 127488076.0, + "step": 3282 + }, + { + "epoch": 0.41763134461264473, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.7744550704956055, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8750532865524292, + "num_tokens": 127522034.0, + "step": 3283 + }, + { + "epoch": 0.4177585548912352, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.752864360809326, + "learning_rate": 1e-06, + "loss": 0.395, + "mean_token_accuracy": 0.8647741079330444, + "num_tokens": 127560405.0, + "step": 3284 + }, + { + "epoch": 0.41788576516982573, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.688532829284668, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8744430541992188, + "num_tokens": 127602935.0, + "step": 3285 + }, + { + "epoch": 0.41801297544841626, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.760162353515625, + "learning_rate": 1e-06, + "loss": 0.4042, + "mean_token_accuracy": 0.8662421107292175, + "num_tokens": 127647647.0, + "step": 3286 + }, + { + "epoch": 0.41814018572700673, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.707984447479248, + "learning_rate": 1e-06, + "loss": 0.3876, + "mean_token_accuracy": 0.8691062927246094, + "num_tokens": 127687914.0, + "step": 3287 + }, + { + "epoch": 0.41826739600559726, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.689770221710205, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8809894323348999, + "num_tokens": 127727507.0, + "step": 3288 + }, + { + "epoch": 0.4183946062841878, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.706578731536865, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.8661397695541382, + "num_tokens": 127773040.0, + "step": 3289 + }, + { + "epoch": 0.41852181656277826, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.755312442779541, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.871804416179657, + "num_tokens": 127811992.0, + "step": 3290 + }, + { + "epoch": 0.4186490268413688, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.789444446563721, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.8721320629119873, + "num_tokens": 127849366.0, + "step": 3291 + }, + { + "epoch": 0.4187762371199593, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.769809246063232, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.876296877861023, + "num_tokens": 127884000.0, + "step": 3292 + }, + { + "epoch": 0.4189034473985498, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.751802444458008, + "learning_rate": 1e-06, + "loss": 0.3902, + "mean_token_accuracy": 0.8681390285491943, + "num_tokens": 127922754.0, + "step": 3293 + }, + { + "epoch": 0.4190306576771403, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.850499153137207, + "learning_rate": 1e-06, + "loss": 0.4078, + "mean_token_accuracy": 0.8625002503395081, + "num_tokens": 127956219.0, + "step": 3294 + }, + { + "epoch": 0.41915786795573085, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.839211463928223, + "learning_rate": 1e-06, + "loss": 0.4246, + "mean_token_accuracy": 0.8621861934661865, + "num_tokens": 127991720.0, + "step": 3295 + }, + { + "epoch": 0.4192850782343213, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.750709533691406, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8783563375473022, + "num_tokens": 128023490.0, + "step": 3296 + }, + { + "epoch": 0.41941228851291185, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 8.509408950805664, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8798625469207764, + "num_tokens": 128064348.0, + "step": 3297 + }, + { + "epoch": 0.4195394987915024, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.698256969451904, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8838443756103516, + "num_tokens": 128100726.0, + "step": 3298 + }, + { + "epoch": 0.41966670907009285, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.639150619506836, + "learning_rate": 1e-06, + "loss": 0.4024, + "mean_token_accuracy": 0.8658944368362427, + "num_tokens": 128144595.0, + "step": 3299 + }, + { + "epoch": 0.4197939193486834, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.79159688949585, + "learning_rate": 1e-06, + "loss": 0.411, + "mean_token_accuracy": 0.8601188659667969, + "num_tokens": 128188839.0, + "step": 3300 + }, + { + "epoch": 0.4199211296272739, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.837813854217529, + "learning_rate": 1e-06, + "loss": 0.3899, + "mean_token_accuracy": 0.8674972057342529, + "num_tokens": 128228788.0, + "step": 3301 + }, + { + "epoch": 0.4200483399058644, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.743875503540039, + "learning_rate": 1e-06, + "loss": 0.4335, + "mean_token_accuracy": 0.8533092141151428, + "num_tokens": 128269504.0, + "step": 3302 + }, + { + "epoch": 0.4201755501844549, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.796227931976318, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8794395923614502, + "num_tokens": 128303528.0, + "step": 3303 + }, + { + "epoch": 0.42030276046304543, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.719208717346191, + "learning_rate": 1e-06, + "loss": 0.434, + "mean_token_accuracy": 0.8553210496902466, + "num_tokens": 128344081.0, + "step": 3304 + }, + { + "epoch": 0.4204299707416359, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.696756839752197, + "learning_rate": 1e-06, + "loss": 0.4204, + "mean_token_accuracy": 0.8578381538391113, + "num_tokens": 128391961.0, + "step": 3305 + }, + { + "epoch": 0.42055718102022643, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.841235637664795, + "learning_rate": 1e-06, + "loss": 0.3964, + "mean_token_accuracy": 0.8714137077331543, + "num_tokens": 128430389.0, + "step": 3306 + }, + { + "epoch": 0.42068439129881696, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.834349632263184, + "learning_rate": 1e-06, + "loss": 0.3898, + "mean_token_accuracy": 0.8645427823066711, + "num_tokens": 128464369.0, + "step": 3307 + }, + { + "epoch": 0.42081160157740743, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.817844867706299, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.8732220530509949, + "num_tokens": 128497761.0, + "step": 3308 + }, + { + "epoch": 0.42093881185599796, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4483928680419922e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.717865943908691, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8819658756256104, + "num_tokens": 128534489.0, + "step": 3309 + }, + { + "epoch": 0.4210660221345885, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4603137969970703e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.813479900360107, + "learning_rate": 1e-06, + "loss": 0.3877, + "mean_token_accuracy": 0.8716287612915039, + "num_tokens": 128571734.0, + "step": 3310 + }, + { + "epoch": 0.42119323241317896, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4603137969970703e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.8134846687316895, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8790445923805237, + "num_tokens": 128605033.0, + "step": 3311 + }, + { + "epoch": 0.4213204426917695, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4603137969970703e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.787982940673828, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.86937415599823, + "num_tokens": 128638669.0, + "step": 3312 + }, + { + "epoch": 0.42144765297036, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4603137969970703e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.735845565795898, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8753824234008789, + "num_tokens": 128680109.0, + "step": 3313 + }, + { + "epoch": 0.4215748632489505, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4603137969970703e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.727938175201416, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8794248104095459, + "num_tokens": 128715874.0, + "step": 3314 + }, + { + "epoch": 0.421702073527541, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.8239030838012695, + "learning_rate": 1e-06, + "loss": 0.4086, + "mean_token_accuracy": 0.8628624677658081, + "num_tokens": 128752955.0, + "step": 3315 + }, + { + "epoch": 0.42182928380613155, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.741924285888672, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.874569296836853, + "num_tokens": 128791694.0, + "step": 3316 + }, + { + "epoch": 0.421956494084722, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.774548530578613, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8757190704345703, + "num_tokens": 128824288.0, + "step": 3317 + }, + { + "epoch": 0.42208370436331255, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.760205268859863, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8799889087677002, + "num_tokens": 128862079.0, + "step": 3318 + }, + { + "epoch": 0.4222109146419031, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.678378582000732, + "learning_rate": 1e-06, + "loss": 0.3933, + "mean_token_accuracy": 0.867626428604126, + "num_tokens": 128905581.0, + "step": 3319 + }, + { + "epoch": 0.42233812492049355, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.670835494995117, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8879611492156982, + "num_tokens": 128944768.0, + "step": 3320 + }, + { + "epoch": 0.4224653351990841, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.634803295135498, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8840383887290955, + "num_tokens": 128986795.0, + "step": 3321 + }, + { + "epoch": 0.4225925454776746, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.739683151245117, + "learning_rate": 1e-06, + "loss": 0.3768, + "mean_token_accuracy": 0.8759678602218628, + "num_tokens": 129027145.0, + "step": 3322 + }, + { + "epoch": 0.4227197557562651, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.904729843139648, + "learning_rate": 1e-06, + "loss": 0.3876, + "mean_token_accuracy": 0.8694087266921997, + "num_tokens": 129058826.0, + "step": 3323 + }, + { + "epoch": 0.4228469660348556, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.665733337402344, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8720397353172302, + "num_tokens": 129101771.0, + "step": 3324 + }, + { + "epoch": 0.42297417631344614, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.743463516235352, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8757836818695068, + "num_tokens": 129137862.0, + "step": 3325 + }, + { + "epoch": 0.4231013865920366, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.71586799621582, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8837841749191284, + "num_tokens": 129177133.0, + "step": 3326 + }, + { + "epoch": 0.42322859687062714, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.752658843994141, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8735185861587524, + "num_tokens": 129216195.0, + "step": 3327 + }, + { + "epoch": 0.42335580714921767, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.700838088989258, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8854258060455322, + "num_tokens": 129257073.0, + "step": 3328 + }, + { + "epoch": 0.42348301742780814, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.828714847564697, + "learning_rate": 1e-06, + "loss": 0.4388, + "mean_token_accuracy": 0.8525033593177795, + "num_tokens": 129298095.0, + "step": 3329 + }, + { + "epoch": 0.42361022770639867, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.825071334838867, + "learning_rate": 1e-06, + "loss": 0.3925, + "mean_token_accuracy": 0.8673815131187439, + "num_tokens": 129333338.0, + "step": 3330 + }, + { + "epoch": 0.4237374379849892, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.78255033493042, + "learning_rate": 1e-06, + "loss": 0.4195, + "mean_token_accuracy": 0.8624382615089417, + "num_tokens": 129371843.0, + "step": 3331 + }, + { + "epoch": 0.4238646482635797, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.790522575378418, + "learning_rate": 1e-06, + "loss": 0.4129, + "mean_token_accuracy": 0.863140881061554, + "num_tokens": 129411992.0, + "step": 3332 + }, + { + "epoch": 0.4239918585421702, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.835028171539307, + "learning_rate": 1e-06, + "loss": 0.4092, + "mean_token_accuracy": 0.8677866458892822, + "num_tokens": 129448677.0, + "step": 3333 + }, + { + "epoch": 0.4241190688207607, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.635540962219238, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8926558494567871, + "num_tokens": 129492151.0, + "step": 3334 + }, + { + "epoch": 0.42424627909935125, + "ewc_loss": 0.0093994140625, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.927417755126953e-06, + "grad_norm": 7.90808629989624, + "learning_rate": 1e-06, + "loss": 0.4099, + "mean_token_accuracy": 0.8611040115356445, + "num_tokens": 129525850.0, + "step": 3335 + }, + { + "epoch": 0.4243734893779417, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4662742614746094e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.76547384262085, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8838022351264954, + "num_tokens": 129564960.0, + "step": 3336 + }, + { + "epoch": 0.42450069965653225, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.681135654449463, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8778767585754395, + "num_tokens": 129607532.0, + "step": 3337 + }, + { + "epoch": 0.4246279099351228, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.797023773193359, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8826717138290405, + "num_tokens": 129644746.0, + "step": 3338 + }, + { + "epoch": 0.42475512021371326, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.707484245300293, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8764550685882568, + "num_tokens": 129687548.0, + "step": 3339 + }, + { + "epoch": 0.4248823304923038, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.902201175689697, + "learning_rate": 1e-06, + "loss": 0.401, + "mean_token_accuracy": 0.8670024871826172, + "num_tokens": 129723150.0, + "step": 3340 + }, + { + "epoch": 0.4250095407708943, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.679506778717041, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8927470445632935, + "num_tokens": 129764647.0, + "step": 3341 + }, + { + "epoch": 0.4251367510494848, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.670534610748291, + "learning_rate": 1e-06, + "loss": 0.4086, + "mean_token_accuracy": 0.8634839057922363, + "num_tokens": 129809653.0, + "step": 3342 + }, + { + "epoch": 0.4252639613280753, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.8796844482421875, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8806734085083008, + "num_tokens": 129843136.0, + "step": 3343 + }, + { + "epoch": 0.42539117160666584, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.648048400878906, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8896999359130859, + "num_tokens": 129881483.0, + "step": 3344 + }, + { + "epoch": 0.4255183818852563, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.764640808105469, + "learning_rate": 1e-06, + "loss": 0.4106, + "mean_token_accuracy": 0.8634339570999146, + "num_tokens": 129919596.0, + "step": 3345 + }, + { + "epoch": 0.42564559216384684, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.660331726074219, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8776726126670837, + "num_tokens": 129964080.0, + "step": 3346 + }, + { + "epoch": 0.42577280244243737, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.654186248779297, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8766012191772461, + "num_tokens": 130009552.0, + "step": 3347 + }, + { + "epoch": 0.42590001272102784, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.93125057220459, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8808568716049194, + "num_tokens": 130046493.0, + "step": 3348 + }, + { + "epoch": 0.42602722299961837, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.6594014167785645, + "learning_rate": 1e-06, + "loss": 0.3918, + "mean_token_accuracy": 0.8677799701690674, + "num_tokens": 130092108.0, + "step": 3349 + }, + { + "epoch": 0.4261544332782089, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.810063362121582, + "learning_rate": 1e-06, + "loss": 0.4367, + "mean_token_accuracy": 0.8568236827850342, + "num_tokens": 130128790.0, + "step": 3350 + }, + { + "epoch": 0.4262816435567994, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.761541366577148, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8844766616821289, + "num_tokens": 130166021.0, + "step": 3351 + }, + { + "epoch": 0.4264088538353899, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.565191745758057, + "learning_rate": 1e-06, + "loss": 0.4169, + "mean_token_accuracy": 0.8604400753974915, + "num_tokens": 130212356.0, + "step": 3352 + }, + { + "epoch": 0.42653606411398043, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.7078166007995605, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8799198865890503, + "num_tokens": 130254326.0, + "step": 3353 + }, + { + "epoch": 0.4266632743925709, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.676976680755615, + "learning_rate": 1e-06, + "loss": 0.4006, + "mean_token_accuracy": 0.8698300123214722, + "num_tokens": 130293238.0, + "step": 3354 + }, + { + "epoch": 0.42679048467116143, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.762207508087158, + "learning_rate": 1e-06, + "loss": 0.4107, + "mean_token_accuracy": 0.8625826835632324, + "num_tokens": 130337648.0, + "step": 3355 + }, + { + "epoch": 0.42691769494975196, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.693560600280762, + "learning_rate": 1e-06, + "loss": 0.3946, + "mean_token_accuracy": 0.8702572584152222, + "num_tokens": 130378338.0, + "step": 3356 + }, + { + "epoch": 0.42704490522834243, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.674420356750488, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8736575841903687, + "num_tokens": 130419500.0, + "step": 3357 + }, + { + "epoch": 0.42717211550693296, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.638271331787109, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8825129270553589, + "num_tokens": 130462459.0, + "step": 3358 + }, + { + "epoch": 0.4272993257855235, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.644505500793457, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8712273836135864, + "num_tokens": 130504689.0, + "step": 3359 + }, + { + "epoch": 0.42742653606411396, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.707190036773682, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8818342685699463, + "num_tokens": 130546486.0, + "step": 3360 + }, + { + "epoch": 0.4275537463427045, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.790260314941406, + "learning_rate": 1e-06, + "loss": 0.3865, + "mean_token_accuracy": 0.8701820373535156, + "num_tokens": 130588131.0, + "step": 3361 + }, + { + "epoch": 0.427680956621295, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 8.03786563873291, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8816474676132202, + "num_tokens": 130624056.0, + "step": 3362 + }, + { + "epoch": 0.4278081668998855, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.675020217895508, + "learning_rate": 1e-06, + "loss": 0.3947, + "mean_token_accuracy": 0.8698539137840271, + "num_tokens": 130664604.0, + "step": 3363 + }, + { + "epoch": 0.427935377178476, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.71884298324585, + "learning_rate": 1e-06, + "loss": 0.4232, + "mean_token_accuracy": 0.8554823398590088, + "num_tokens": 130708922.0, + "step": 3364 + }, + { + "epoch": 0.42806258745706655, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.748016834259033, + "learning_rate": 1e-06, + "loss": 0.4281, + "mean_token_accuracy": 0.853701114654541, + "num_tokens": 130749644.0, + "step": 3365 + }, + { + "epoch": 0.428189797735657, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.746309280395508, + "learning_rate": 1e-06, + "loss": 0.4169, + "mean_token_accuracy": 0.8634346723556519, + "num_tokens": 130793345.0, + "step": 3366 + }, + { + "epoch": 0.42831700801424755, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4781951904296875e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.764594554901123, + "learning_rate": 1e-06, + "loss": 0.4155, + "mean_token_accuracy": 0.8597149848937988, + "num_tokens": 130833782.0, + "step": 3367 + }, + { + "epoch": 0.4284442182928381, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4722347259521484e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.666825294494629, + "learning_rate": 1e-06, + "loss": 0.3884, + "mean_token_accuracy": 0.8674193620681763, + "num_tokens": 130875450.0, + "step": 3368 + }, + { + "epoch": 0.42857142857142855, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4841556549072266e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.759033679962158, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.878020703792572, + "num_tokens": 130911146.0, + "step": 3369 + }, + { + "epoch": 0.4286986388500191, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4841556549072266e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.731853008270264, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8808690309524536, + "num_tokens": 130948793.0, + "step": 3370 + }, + { + "epoch": 0.4288258491286096, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4841556549072266e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.904191017150879, + "learning_rate": 1e-06, + "loss": 0.3907, + "mean_token_accuracy": 0.864443838596344, + "num_tokens": 130983412.0, + "step": 3371 + }, + { + "epoch": 0.4289530594072001, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4841556549072266e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.763864994049072, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.864621639251709, + "num_tokens": 131018995.0, + "step": 3372 + }, + { + "epoch": 0.4290802696857906, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4841556549072266e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.748810768127441, + "learning_rate": 1e-06, + "loss": 0.3854, + "mean_token_accuracy": 0.8693987131118774, + "num_tokens": 131058880.0, + "step": 3373 + }, + { + "epoch": 0.42920747996438113, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4901161193847656e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.769557476043701, + "learning_rate": 1e-06, + "loss": 0.3892, + "mean_token_accuracy": 0.8689192533493042, + "num_tokens": 131099039.0, + "step": 3374 + }, + { + "epoch": 0.4293346902429716, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4901161193847656e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.789971828460693, + "learning_rate": 1e-06, + "loss": 0.3987, + "mean_token_accuracy": 0.8634381294250488, + "num_tokens": 131134754.0, + "step": 3375 + }, + { + "epoch": 0.42946190052156213, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4901161193847656e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.775839328765869, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8839830160140991, + "num_tokens": 131170450.0, + "step": 3376 + }, + { + "epoch": 0.42958911080015266, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4901161193847656e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.763456344604492, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8734555244445801, + "num_tokens": 131206448.0, + "step": 3377 + }, + { + "epoch": 0.42971632107874314, + "ewc_loss": 0.00946044921875, + "ewc_loss_diag": 1.4901161193847656e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.785201549530029, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8725276589393616, + "num_tokens": 131239034.0, + "step": 3378 + }, + { + "epoch": 0.42984353135733366, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4901161193847656e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.724536895751953, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.877081036567688, + "num_tokens": 131281203.0, + "step": 3379 + }, + { + "epoch": 0.4299707416359242, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.4901161193847656e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.829838752746582, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.887597918510437, + "num_tokens": 131316302.0, + "step": 3380 + }, + { + "epoch": 0.4300979519145147, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.4901161193847656e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.708883285522461, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8747843503952026, + "num_tokens": 131356898.0, + "step": 3381 + }, + { + "epoch": 0.4302251621931052, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4901161193847656e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.877638816833496, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8758121728897095, + "num_tokens": 131387999.0, + "step": 3382 + }, + { + "epoch": 0.4303523724716957, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.791030406951904, + "learning_rate": 1e-06, + "loss": 0.4396, + "mean_token_accuracy": 0.8506584763526917, + "num_tokens": 131427794.0, + "step": 3383 + }, + { + "epoch": 0.43047958275028625, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.729188442230225, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8712878227233887, + "num_tokens": 131469077.0, + "step": 3384 + }, + { + "epoch": 0.4306067930288767, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.676037311553955, + "learning_rate": 1e-06, + "loss": 0.3937, + "mean_token_accuracy": 0.8657694458961487, + "num_tokens": 131512019.0, + "step": 3385 + }, + { + "epoch": 0.43073400330746725, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.728110313415527, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8712877631187439, + "num_tokens": 131552008.0, + "step": 3386 + }, + { + "epoch": 0.4308612135860578, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.906041622161865, + "learning_rate": 1e-06, + "loss": 0.437, + "mean_token_accuracy": 0.8566670417785645, + "num_tokens": 131586887.0, + "step": 3387 + }, + { + "epoch": 0.43098842386464825, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.802429676055908, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8778659105300903, + "num_tokens": 131620337.0, + "step": 3388 + }, + { + "epoch": 0.4311156341432388, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.859964370727539, + "learning_rate": 1e-06, + "loss": 0.3925, + "mean_token_accuracy": 0.8662741184234619, + "num_tokens": 131654633.0, + "step": 3389 + }, + { + "epoch": 0.4312428444218293, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.7211713790893555, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.8706695437431335, + "num_tokens": 131693275.0, + "step": 3390 + }, + { + "epoch": 0.4313700547004198, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.793748378753662, + "learning_rate": 1e-06, + "loss": 0.3834, + "mean_token_accuracy": 0.8684449195861816, + "num_tokens": 131728929.0, + "step": 3391 + }, + { + "epoch": 0.4314972649790103, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.963558197021484, + "learning_rate": 1e-06, + "loss": 0.4486, + "mean_token_accuracy": 0.8511525988578796, + "num_tokens": 131765648.0, + "step": 3392 + }, + { + "epoch": 0.43162447525760084, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.745884418487549, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.8734645843505859, + "num_tokens": 131804506.0, + "step": 3393 + }, + { + "epoch": 0.4317516855361913, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.84243106842041, + "learning_rate": 1e-06, + "loss": 0.4216, + "mean_token_accuracy": 0.8565759062767029, + "num_tokens": 131838436.0, + "step": 3394 + }, + { + "epoch": 0.43187889581478184, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.713245868682861, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.8734309077262878, + "num_tokens": 131877991.0, + "step": 3395 + }, + { + "epoch": 0.43200610609337237, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.77711296081543, + "learning_rate": 1e-06, + "loss": 0.4035, + "mean_token_accuracy": 0.8643374443054199, + "num_tokens": 131916880.0, + "step": 3396 + }, + { + "epoch": 0.43213331637196284, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.4960765838623047e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.781161308288574, + "learning_rate": 1e-06, + "loss": 0.3966, + "mean_token_accuracy": 0.8655058145523071, + "num_tokens": 131955523.0, + "step": 3397 + }, + { + "epoch": 0.43226052665055337, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.753544807434082, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8729356527328491, + "num_tokens": 131993405.0, + "step": 3398 + }, + { + "epoch": 0.4323877369291439, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.7981367111206055, + "learning_rate": 1e-06, + "loss": 0.3966, + "mean_token_accuracy": 0.8691056966781616, + "num_tokens": 132032212.0, + "step": 3399 + }, + { + "epoch": 0.43251494720773437, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.755785942077637, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8845663070678711, + "num_tokens": 132071980.0, + "step": 3400 + }, + { + "epoch": 0.4326421574863249, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.670114994049072, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8817543983459473, + "num_tokens": 132115792.0, + "step": 3401 + }, + { + "epoch": 0.4327693677649154, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.789580821990967, + "learning_rate": 1e-06, + "loss": 0.3866, + "mean_token_accuracy": 0.8707410097122192, + "num_tokens": 132155666.0, + "step": 3402 + }, + { + "epoch": 0.4328965780435059, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.773135185241699, + "learning_rate": 1e-06, + "loss": 0.3888, + "mean_token_accuracy": 0.8686985969543457, + "num_tokens": 132190860.0, + "step": 3403 + }, + { + "epoch": 0.4330237883220964, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.808253765106201, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8780153393745422, + "num_tokens": 132227673.0, + "step": 3404 + }, + { + "epoch": 0.43315099860068695, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.792690753936768, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8769083023071289, + "num_tokens": 132263919.0, + "step": 3405 + }, + { + "epoch": 0.4332782088792774, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.721608638763428, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8767367601394653, + "num_tokens": 132303992.0, + "step": 3406 + }, + { + "epoch": 0.43340541915786795, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.718270778656006, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8847734928131104, + "num_tokens": 132344427.0, + "step": 3407 + }, + { + "epoch": 0.4335326294364585, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.680431842803955, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8704221248626709, + "num_tokens": 132386026.0, + "step": 3408 + }, + { + "epoch": 0.43365983971504896, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.818068504333496, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8848704099655151, + "num_tokens": 132418647.0, + "step": 3409 + }, + { + "epoch": 0.4337870499936395, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.665261268615723, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8756676316261292, + "num_tokens": 132460824.0, + "step": 3410 + }, + { + "epoch": 0.43391426027223, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.699316024780273, + "learning_rate": 1e-06, + "loss": 0.3966, + "mean_token_accuracy": 0.8696374893188477, + "num_tokens": 132503066.0, + "step": 3411 + }, + { + "epoch": 0.4340414705508205, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.821699619293213, + "learning_rate": 1e-06, + "loss": 0.3949, + "mean_token_accuracy": 0.8667301535606384, + "num_tokens": 132540399.0, + "step": 3412 + }, + { + "epoch": 0.434168680829411, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.701572418212891, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.890789270401001, + "num_tokens": 132582946.0, + "step": 3413 + }, + { + "epoch": 0.43429589110800154, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.927181243896484, + "learning_rate": 1e-06, + "loss": 0.4314, + "mean_token_accuracy": 0.853670597076416, + "num_tokens": 132619587.0, + "step": 3414 + }, + { + "epoch": 0.434423101386592, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.710111141204834, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8693458437919617, + "num_tokens": 132660096.0, + "step": 3415 + }, + { + "epoch": 0.43455031166518254, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.688475131988525, + "learning_rate": 1e-06, + "loss": 0.3884, + "mean_token_accuracy": 0.8701273202896118, + "num_tokens": 132701366.0, + "step": 3416 + }, + { + "epoch": 0.43467752194377307, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.763303279876709, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8750159740447998, + "num_tokens": 132741229.0, + "step": 3417 + }, + { + "epoch": 0.43480473222236354, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5020370483398438e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.73262882232666, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.878368616104126, + "num_tokens": 132775868.0, + "step": 3418 + }, + { + "epoch": 0.43493194250095407, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5079975128173828e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.769266128540039, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8890484571456909, + "num_tokens": 132807623.0, + "step": 3419 + }, + { + "epoch": 0.4350591527795446, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.5079975128173828e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.765439987182617, + "learning_rate": 1e-06, + "loss": 0.3796, + "mean_token_accuracy": 0.8680681586265564, + "num_tokens": 132846203.0, + "step": 3420 + }, + { + "epoch": 0.4351863630581351, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5079975128173828e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.788147449493408, + "learning_rate": 1e-06, + "loss": 0.4128, + "mean_token_accuracy": 0.8616900444030762, + "num_tokens": 132884915.0, + "step": 3421 + }, + { + "epoch": 0.4353135733367256, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5079975128173828e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.803459644317627, + "learning_rate": 1e-06, + "loss": 0.4329, + "mean_token_accuracy": 0.8587013483047485, + "num_tokens": 132924398.0, + "step": 3422 + }, + { + "epoch": 0.43544078361531613, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5079975128173828e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.745960712432861, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8766624927520752, + "num_tokens": 132959829.0, + "step": 3423 + }, + { + "epoch": 0.4355679938939066, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5079975128173828e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.763757705688477, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8705816864967346, + "num_tokens": 132999114.0, + "step": 3424 + }, + { + "epoch": 0.43569520417249713, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5079975128173828e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.689924240112305, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8721539378166199, + "num_tokens": 133042441.0, + "step": 3425 + }, + { + "epoch": 0.43582241445108766, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.5079975128173828e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.700270652770996, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8859806060791016, + "num_tokens": 133080941.0, + "step": 3426 + }, + { + "epoch": 0.43594962472967813, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5079975128173828e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.703896999359131, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8719000816345215, + "num_tokens": 133127040.0, + "step": 3427 + }, + { + "epoch": 0.43607683500826866, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5079975128173828e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.723933219909668, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8818629384040833, + "num_tokens": 133165740.0, + "step": 3428 + }, + { + "epoch": 0.4362040452868592, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5079975128173828e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.671609878540039, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8745859265327454, + "num_tokens": 133210052.0, + "step": 3429 + }, + { + "epoch": 0.4363312555654497, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5079975128173828e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 8.04141902923584, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.8716081380844116, + "num_tokens": 133251836.0, + "step": 3430 + }, + { + "epoch": 0.4364584658440402, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5079975128173828e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.696084976196289, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.883851170539856, + "num_tokens": 133290685.0, + "step": 3431 + }, + { + "epoch": 0.4365856761226307, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.696560382843018, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8772302865982056, + "num_tokens": 133330726.0, + "step": 3432 + }, + { + "epoch": 0.43671288640122125, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.828343391418457, + "learning_rate": 1e-06, + "loss": 0.3849, + "mean_token_accuracy": 0.8709633350372314, + "num_tokens": 133366165.0, + "step": 3433 + }, + { + "epoch": 0.4368400966798117, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.630406856536865, + "learning_rate": 1e-06, + "loss": 0.4187, + "mean_token_accuracy": 0.8606455326080322, + "num_tokens": 133412414.0, + "step": 3434 + }, + { + "epoch": 0.43696730695840225, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.734785556793213, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8896468877792358, + "num_tokens": 133451181.0, + "step": 3435 + }, + { + "epoch": 0.4370945172369928, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.8127007484436035, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8741245269775391, + "num_tokens": 133484564.0, + "step": 3436 + }, + { + "epoch": 0.43722172751558325, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.65245246887207, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8887062668800354, + "num_tokens": 133524871.0, + "step": 3437 + }, + { + "epoch": 0.4373489377941738, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.812901020050049, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.8723890781402588, + "num_tokens": 133561955.0, + "step": 3438 + }, + { + "epoch": 0.4374761480727643, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.676092624664307, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8876698613166809, + "num_tokens": 133598564.0, + "step": 3439 + }, + { + "epoch": 0.4376033583513548, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.734556674957275, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.8794302940368652, + "num_tokens": 133636852.0, + "step": 3440 + }, + { + "epoch": 0.4377305686299453, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.6964240074157715, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8845832347869873, + "num_tokens": 133679287.0, + "step": 3441 + }, + { + "epoch": 0.43785777890853583, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.816657543182373, + "learning_rate": 1e-06, + "loss": 0.4045, + "mean_token_accuracy": 0.862398087978363, + "num_tokens": 133717214.0, + "step": 3442 + }, + { + "epoch": 0.4379849891871263, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.633368015289307, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8765790462493896, + "num_tokens": 133762767.0, + "step": 3443 + }, + { + "epoch": 0.43811219946571683, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.876312732696533, + "learning_rate": 1e-06, + "loss": 0.3912, + "mean_token_accuracy": 0.8691304326057434, + "num_tokens": 133796769.0, + "step": 3444 + }, + { + "epoch": 0.43823940974430736, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.780450820922852, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8748786449432373, + "num_tokens": 133836572.0, + "step": 3445 + }, + { + "epoch": 0.43836662002289783, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.8512749671936035, + "learning_rate": 1e-06, + "loss": 0.4396, + "mean_token_accuracy": 0.8559818267822266, + "num_tokens": 133871933.0, + "step": 3446 + }, + { + "epoch": 0.43849383030148836, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.922036170959473, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8799689412117004, + "num_tokens": 133905630.0, + "step": 3447 + }, + { + "epoch": 0.4386210405800789, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5139579772949219e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.852082252502441, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8781551122665405, + "num_tokens": 133939847.0, + "step": 3448 + }, + { + "epoch": 0.43874825085866936, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.702403545379639, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8778912425041199, + "num_tokens": 133980631.0, + "step": 3449 + }, + { + "epoch": 0.4388754611372599, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.826663017272949, + "learning_rate": 1e-06, + "loss": 0.3761, + "mean_token_accuracy": 0.8700810670852661, + "num_tokens": 134013633.0, + "step": 3450 + }, + { + "epoch": 0.4390026714158504, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.782864570617676, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8802494406700134, + "num_tokens": 134051748.0, + "step": 3451 + }, + { + "epoch": 0.4391298816944409, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.843483924865723, + "learning_rate": 1e-06, + "loss": 0.4471, + "mean_token_accuracy": 0.8523590564727783, + "num_tokens": 134090464.0, + "step": 3452 + }, + { + "epoch": 0.4392570919730314, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.982706069946289, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.8777822256088257, + "num_tokens": 134132301.0, + "step": 3453 + }, + { + "epoch": 0.43938430225162195, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.723550796508789, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8742738366127014, + "num_tokens": 134173175.0, + "step": 3454 + }, + { + "epoch": 0.4395115125302124, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.765493392944336, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.872081458568573, + "num_tokens": 134209881.0, + "step": 3455 + }, + { + "epoch": 0.43963872280880295, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.716054916381836, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8754473328590393, + "num_tokens": 134250301.0, + "step": 3456 + }, + { + "epoch": 0.4397659330873935, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.759720325469971, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8797037601470947, + "num_tokens": 134287513.0, + "step": 3457 + }, + { + "epoch": 0.43989314336598395, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.837371826171875, + "learning_rate": 1e-06, + "loss": 0.4133, + "mean_token_accuracy": 0.8641791343688965, + "num_tokens": 134325508.0, + "step": 3458 + }, + { + "epoch": 0.4400203536445745, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.836404800415039, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8852922916412354, + "num_tokens": 134356871.0, + "step": 3459 + }, + { + "epoch": 0.440147563923165, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.823041915893555, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8745821118354797, + "num_tokens": 134391063.0, + "step": 3460 + }, + { + "epoch": 0.4402747742017555, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.766732215881348, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8751882314682007, + "num_tokens": 134427196.0, + "step": 3461 + }, + { + "epoch": 0.440401984480346, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.813251972198486, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8735223412513733, + "num_tokens": 134465373.0, + "step": 3462 + }, + { + "epoch": 0.44052919475893654, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.78400993347168, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8833644986152649, + "num_tokens": 134498399.0, + "step": 3463 + }, + { + "epoch": 0.440656405037527, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.797472953796387, + "learning_rate": 1e-06, + "loss": 0.4389, + "mean_token_accuracy": 0.8532042503356934, + "num_tokens": 134541763.0, + "step": 3464 + }, + { + "epoch": 0.44078361531611754, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.739904403686523, + "learning_rate": 1e-06, + "loss": 0.3854, + "mean_token_accuracy": 0.8695591688156128, + "num_tokens": 134584581.0, + "step": 3465 + }, + { + "epoch": 0.44091082559470807, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.768664836883545, + "learning_rate": 1e-06, + "loss": 0.3875, + "mean_token_accuracy": 0.8693990707397461, + "num_tokens": 134625125.0, + "step": 3466 + }, + { + "epoch": 0.44103803587329854, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.822818279266357, + "learning_rate": 1e-06, + "loss": 0.4087, + "mean_token_accuracy": 0.8624656200408936, + "num_tokens": 134667950.0, + "step": 3467 + }, + { + "epoch": 0.44116524615188907, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.735260486602783, + "learning_rate": 1e-06, + "loss": 0.3814, + "mean_token_accuracy": 0.8729939460754395, + "num_tokens": 134706226.0, + "step": 3468 + }, + { + "epoch": 0.4412924564304796, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.519918441772461e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.8608527183532715, + "learning_rate": 1e-06, + "loss": 0.3896, + "mean_token_accuracy": 0.8692550659179688, + "num_tokens": 134744221.0, + "step": 3469 + }, + { + "epoch": 0.44141966670907007, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.52587890625e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.901131629943848, + "learning_rate": 1e-06, + "loss": 0.3834, + "mean_token_accuracy": 0.8680837154388428, + "num_tokens": 134776361.0, + "step": 3470 + }, + { + "epoch": 0.4415468769876606, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.52587890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.804179668426514, + "learning_rate": 1e-06, + "loss": 0.4427, + "mean_token_accuracy": 0.8560744524002075, + "num_tokens": 134818727.0, + "step": 3471 + }, + { + "epoch": 0.4416740872662511, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.52587890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.805469512939453, + "learning_rate": 1e-06, + "loss": 0.4404, + "mean_token_accuracy": 0.8586345911026001, + "num_tokens": 134857489.0, + "step": 3472 + }, + { + "epoch": 0.4418012975448416, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.52587890625e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.733983993530273, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8751427531242371, + "num_tokens": 134895900.0, + "step": 3473 + }, + { + "epoch": 0.4419285078234321, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.52587890625e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.727664470672607, + "learning_rate": 1e-06, + "loss": 0.3908, + "mean_token_accuracy": 0.8674929738044739, + "num_tokens": 134936312.0, + "step": 3474 + }, + { + "epoch": 0.44205571810202265, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.802613258361816, + "learning_rate": 1e-06, + "loss": 0.4064, + "mean_token_accuracy": 0.8631609678268433, + "num_tokens": 134975718.0, + "step": 3475 + }, + { + "epoch": 0.4421829283806131, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.770754814147949, + "learning_rate": 1e-06, + "loss": 0.3717, + "mean_token_accuracy": 0.8754194974899292, + "num_tokens": 135016662.0, + "step": 3476 + }, + { + "epoch": 0.44231013865920366, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.822824478149414, + "learning_rate": 1e-06, + "loss": 0.3671, + "mean_token_accuracy": 0.8767428994178772, + "num_tokens": 135058128.0, + "step": 3477 + }, + { + "epoch": 0.4424373489377942, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.824285984039307, + "learning_rate": 1e-06, + "loss": 0.4117, + "mean_token_accuracy": 0.8606839179992676, + "num_tokens": 135098658.0, + "step": 3478 + }, + { + "epoch": 0.44256455921638466, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.52587890625e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.703586578369141, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8837364912033081, + "num_tokens": 135138895.0, + "step": 3479 + }, + { + "epoch": 0.4426917694949752, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.52587890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.899877071380615, + "learning_rate": 1e-06, + "loss": 0.3877, + "mean_token_accuracy": 0.8699965476989746, + "num_tokens": 135171891.0, + "step": 3480 + }, + { + "epoch": 0.4428189797735657, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.52587890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.7017951011657715, + "learning_rate": 1e-06, + "loss": 0.3825, + "mean_token_accuracy": 0.8691441416740417, + "num_tokens": 135213812.0, + "step": 3481 + }, + { + "epoch": 0.44294619005215624, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.726836204528809, + "learning_rate": 1e-06, + "loss": 0.3659, + "mean_token_accuracy": 0.8752963542938232, + "num_tokens": 135255464.0, + "step": 3482 + }, + { + "epoch": 0.4430734003307467, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.52587890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.814401149749756, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8766024112701416, + "num_tokens": 135296699.0, + "step": 3483 + }, + { + "epoch": 0.44320061060933724, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.74937105178833, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.8771065473556519, + "num_tokens": 135335082.0, + "step": 3484 + }, + { + "epoch": 0.44332782088792777, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.799976825714111, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.8716668486595154, + "num_tokens": 135379282.0, + "step": 3485 + }, + { + "epoch": 0.44345503116651824, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.764937400817871, + "learning_rate": 1e-06, + "loss": 0.3798, + "mean_token_accuracy": 0.8702496290206909, + "num_tokens": 135420884.0, + "step": 3486 + }, + { + "epoch": 0.44358224144510877, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.85391902923584, + "learning_rate": 1e-06, + "loss": 0.3877, + "mean_token_accuracy": 0.8694718480110168, + "num_tokens": 135460996.0, + "step": 3487 + }, + { + "epoch": 0.4437094517236993, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.789385795593262, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8693079352378845, + "num_tokens": 135506289.0, + "step": 3488 + }, + { + "epoch": 0.4438366620022898, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.8245320320129395, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8710415959358215, + "num_tokens": 135544048.0, + "step": 3489 + }, + { + "epoch": 0.4439638722808803, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.935725212097168, + "learning_rate": 1e-06, + "loss": 0.4424, + "mean_token_accuracy": 0.8509556651115417, + "num_tokens": 135582079.0, + "step": 3490 + }, + { + "epoch": 0.44409108255947083, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.90083122253418, + "learning_rate": 1e-06, + "loss": 0.4185, + "mean_token_accuracy": 0.8577760457992554, + "num_tokens": 135617854.0, + "step": 3491 + }, + { + "epoch": 0.4442182928380613, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.771236419677734, + "learning_rate": 1e-06, + "loss": 0.4506, + "mean_token_accuracy": 0.8492924571037292, + "num_tokens": 135662449.0, + "step": 3492 + }, + { + "epoch": 0.44434550311665183, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.79779577255249, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.8774523735046387, + "num_tokens": 135696913.0, + "step": 3493 + }, + { + "epoch": 0.44447271339524236, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.761561870574951, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.873159646987915, + "num_tokens": 135735744.0, + "step": 3494 + }, + { + "epoch": 0.44459992367383283, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.78678560256958, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.878646194934845, + "num_tokens": 135769435.0, + "step": 3495 + }, + { + "epoch": 0.44472713395242336, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.760477066040039, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8963605761528015, + "num_tokens": 135804865.0, + "step": 3496 + }, + { + "epoch": 0.4448543442310139, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.905878067016602, + "learning_rate": 1e-06, + "loss": 0.3879, + "mean_token_accuracy": 0.8686633706092834, + "num_tokens": 135839672.0, + "step": 3497 + }, + { + "epoch": 0.44498155450960436, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.728952407836914, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8838568925857544, + "num_tokens": 135883195.0, + "step": 3498 + }, + { + "epoch": 0.4451087647881949, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.964039325714111, + "learning_rate": 1e-06, + "loss": 0.4453, + "mean_token_accuracy": 0.8530906438827515, + "num_tokens": 135914149.0, + "step": 3499 + }, + { + "epoch": 0.4452359750667854, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.866952419281006, + "learning_rate": 1e-06, + "loss": 0.41, + "mean_token_accuracy": 0.8660753965377808, + "num_tokens": 135952176.0, + "step": 3500 + }, + { + "epoch": 0.4453631853453759, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.808343410491943, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8716994524002075, + "num_tokens": 135986428.0, + "step": 3501 + }, + { + "epoch": 0.4454903956239664, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.79105281829834, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8816101551055908, + "num_tokens": 136022478.0, + "step": 3502 + }, + { + "epoch": 0.44561760590255695, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.785492897033691, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8756878972053528, + "num_tokens": 136060231.0, + "step": 3503 + }, + { + "epoch": 0.4457448161811474, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.802979469299316, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8834079504013062, + "num_tokens": 136095570.0, + "step": 3504 + }, + { + "epoch": 0.44587202645973795, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.704325199127197, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8782394528388977, + "num_tokens": 136140696.0, + "step": 3505 + }, + { + "epoch": 0.4459992367383285, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.790918827056885, + "learning_rate": 1e-06, + "loss": 0.3868, + "mean_token_accuracy": 0.8685265779495239, + "num_tokens": 136181039.0, + "step": 3506 + }, + { + "epoch": 0.44612644701691895, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.822845935821533, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8739485740661621, + "num_tokens": 136218426.0, + "step": 3507 + }, + { + "epoch": 0.4462536572955095, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.791029930114746, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8760370016098022, + "num_tokens": 136260206.0, + "step": 3508 + }, + { + "epoch": 0.4463808675741, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.537799835205078e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.778222560882568, + "learning_rate": 1e-06, + "loss": 0.3899, + "mean_token_accuracy": 0.8697687387466431, + "num_tokens": 136303329.0, + "step": 3509 + }, + { + "epoch": 0.4465080778526905, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.819904804229736, + "learning_rate": 1e-06, + "loss": 0.422, + "mean_token_accuracy": 0.857135534286499, + "num_tokens": 136343455.0, + "step": 3510 + }, + { + "epoch": 0.446635288131281, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.789327144622803, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8764655590057373, + "num_tokens": 136383574.0, + "step": 3511 + }, + { + "epoch": 0.44676249840987153, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.874359607696533, + "learning_rate": 1e-06, + "loss": 0.443, + "mean_token_accuracy": 0.8537055253982544, + "num_tokens": 136422569.0, + "step": 3512 + }, + { + "epoch": 0.446889708688462, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.782998085021973, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8734606504440308, + "num_tokens": 136465130.0, + "step": 3513 + }, + { + "epoch": 0.44701691896705253, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.823280334472656, + "learning_rate": 1e-06, + "loss": 0.394, + "mean_token_accuracy": 0.8734017014503479, + "num_tokens": 136503493.0, + "step": 3514 + }, + { + "epoch": 0.44714412924564306, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.803293228149414, + "learning_rate": 1e-06, + "loss": 0.3981, + "mean_token_accuracy": 0.8692513108253479, + "num_tokens": 136543926.0, + "step": 3515 + }, + { + "epoch": 0.44727133952423354, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.765122890472412, + "learning_rate": 1e-06, + "loss": 0.3771, + "mean_token_accuracy": 0.8732230067253113, + "num_tokens": 136583709.0, + "step": 3516 + }, + { + "epoch": 0.44739854980282406, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.84466552734375, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8659100532531738, + "num_tokens": 136619646.0, + "step": 3517 + }, + { + "epoch": 0.4475257600814146, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.8356122970581055, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8747185468673706, + "num_tokens": 136652771.0, + "step": 3518 + }, + { + "epoch": 0.44765297036000506, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.7133049964904785, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8909124135971069, + "num_tokens": 136693253.0, + "step": 3519 + }, + { + "epoch": 0.4477801806385956, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.7616705894470215, + "learning_rate": 1e-06, + "loss": 0.4118, + "mean_token_accuracy": 0.8623354434967041, + "num_tokens": 136732995.0, + "step": 3520 + }, + { + "epoch": 0.4479073909171861, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.872546195983887, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8773185014724731, + "num_tokens": 136768129.0, + "step": 3521 + }, + { + "epoch": 0.4480346011957766, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.677253246307373, + "learning_rate": 1e-06, + "loss": 0.4056, + "mean_token_accuracy": 0.8629298806190491, + "num_tokens": 136820073.0, + "step": 3522 + }, + { + "epoch": 0.4481618114743671, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.870689392089844, + "learning_rate": 1e-06, + "loss": 0.3918, + "mean_token_accuracy": 0.8699809908866882, + "num_tokens": 136855222.0, + "step": 3523 + }, + { + "epoch": 0.44828902175295765, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.833619594573975, + "learning_rate": 1e-06, + "loss": 0.4135, + "mean_token_accuracy": 0.861302375793457, + "num_tokens": 136896063.0, + "step": 3524 + }, + { + "epoch": 0.4484162320315481, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.8008036613464355, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8795719742774963, + "num_tokens": 136932901.0, + "step": 3525 + }, + { + "epoch": 0.44854344231013865, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.761459827423096, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8712059855461121, + "num_tokens": 136975064.0, + "step": 3526 + }, + { + "epoch": 0.4486706525887292, + "ewc_loss": 0.009521484375, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.963705062866211, + "learning_rate": 1e-06, + "loss": 0.4112, + "mean_token_accuracy": 0.8644863963127136, + "num_tokens": 137008016.0, + "step": 3527 + }, + { + "epoch": 0.44879786286731965, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.789147853851318, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8791611790657043, + "num_tokens": 137042579.0, + "step": 3528 + }, + { + "epoch": 0.4489250731459102, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.769875526428223, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8766396641731262, + "num_tokens": 137081910.0, + "step": 3529 + }, + { + "epoch": 0.4490522834245007, + "ewc_loss": 0.00958251953125, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.827220439910889, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.8743656873703003, + "num_tokens": 137121310.0, + "step": 3530 + }, + { + "epoch": 0.44917949370309124, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.8439483642578125, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8688551783561707, + "num_tokens": 137160478.0, + "step": 3531 + }, + { + "epoch": 0.4493067039816817, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.830543041229248, + "learning_rate": 1e-06, + "loss": 0.3924, + "mean_token_accuracy": 0.8686915040016174, + "num_tokens": 137201524.0, + "step": 3532 + }, + { + "epoch": 0.44943391426027224, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.756833076477051, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8807822465896606, + "num_tokens": 137238957.0, + "step": 3533 + }, + { + "epoch": 0.44956112453886277, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.859975814819336, + "learning_rate": 1e-06, + "loss": 0.3714, + "mean_token_accuracy": 0.8746048212051392, + "num_tokens": 137276970.0, + "step": 3534 + }, + { + "epoch": 0.44968833481745324, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.794361591339111, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8804259896278381, + "num_tokens": 137313419.0, + "step": 3535 + }, + { + "epoch": 0.44981554509604377, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 8.480314254760742, + "learning_rate": 1e-06, + "loss": 0.3961, + "mean_token_accuracy": 0.8712484240531921, + "num_tokens": 137347524.0, + "step": 3536 + }, + { + "epoch": 0.4499427553746343, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.780793190002441, + "learning_rate": 1e-06, + "loss": 0.3788, + "mean_token_accuracy": 0.8724690079689026, + "num_tokens": 137388022.0, + "step": 3537 + }, + { + "epoch": 0.45006996565322477, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.681667804718018, + "learning_rate": 1e-06, + "loss": 0.3647, + "mean_token_accuracy": 0.877503514289856, + "num_tokens": 137429566.0, + "step": 3538 + }, + { + "epoch": 0.4501971759318153, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.868496894836426, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.8723150491714478, + "num_tokens": 137464857.0, + "step": 3539 + }, + { + "epoch": 0.4503243862104058, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.768237113952637, + "learning_rate": 1e-06, + "loss": 0.3971, + "mean_token_accuracy": 0.866847038269043, + "num_tokens": 137504010.0, + "step": 3540 + }, + { + "epoch": 0.4504515964889963, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.7798051834106445, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8827794194221497, + "num_tokens": 137540681.0, + "step": 3541 + }, + { + "epoch": 0.4505788067675868, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.862738609313965, + "learning_rate": 1e-06, + "loss": 0.4437, + "mean_token_accuracy": 0.8581004738807678, + "num_tokens": 137576270.0, + "step": 3542 + }, + { + "epoch": 0.45070601704617735, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.696542263031006, + "learning_rate": 1e-06, + "loss": 0.3849, + "mean_token_accuracy": 0.8724870681762695, + "num_tokens": 137619991.0, + "step": 3543 + }, + { + "epoch": 0.4508332273247678, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.78262186050415, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8765155076980591, + "num_tokens": 137655450.0, + "step": 3544 + }, + { + "epoch": 0.45096043760335836, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.718025207519531, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8916934728622437, + "num_tokens": 137692704.0, + "step": 3545 + }, + { + "epoch": 0.4510876478819489, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.790152549743652, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.875388503074646, + "num_tokens": 137729965.0, + "step": 3546 + }, + { + "epoch": 0.45121485816053936, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.809317588806152, + "learning_rate": 1e-06, + "loss": 0.3996, + "mean_token_accuracy": 0.8686568737030029, + "num_tokens": 137768941.0, + "step": 3547 + }, + { + "epoch": 0.4513420684391299, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.762310028076172, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.879111647605896, + "num_tokens": 137808978.0, + "step": 3548 + }, + { + "epoch": 0.4514692787177204, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5497207641601562e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.839014053344727, + "learning_rate": 1e-06, + "loss": 0.4335, + "mean_token_accuracy": 0.854956865310669, + "num_tokens": 137849405.0, + "step": 3549 + }, + { + "epoch": 0.4515964889963109, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.882505893707275, + "learning_rate": 1e-06, + "loss": 0.413, + "mean_token_accuracy": 0.8613139390945435, + "num_tokens": 137888046.0, + "step": 3550 + }, + { + "epoch": 0.4517236992749014, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.697451114654541, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8809946775436401, + "num_tokens": 137929776.0, + "step": 3551 + }, + { + "epoch": 0.45185090955349194, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.77605676651001, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8735524415969849, + "num_tokens": 137969356.0, + "step": 3552 + }, + { + "epoch": 0.4519781198320824, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.818996906280518, + "learning_rate": 1e-06, + "loss": 0.3979, + "mean_token_accuracy": 0.8661888837814331, + "num_tokens": 138004684.0, + "step": 3553 + }, + { + "epoch": 0.45210533011067294, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.743443965911865, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.881851315498352, + "num_tokens": 138044665.0, + "step": 3554 + }, + { + "epoch": 0.45223254038926347, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.776525020599365, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8756814002990723, + "num_tokens": 138084299.0, + "step": 3555 + }, + { + "epoch": 0.45235975066785394, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.816629409790039, + "learning_rate": 1e-06, + "loss": 0.3848, + "mean_token_accuracy": 0.872636079788208, + "num_tokens": 138125087.0, + "step": 3556 + }, + { + "epoch": 0.45248696094644447, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.7256927490234375, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.890803337097168, + "num_tokens": 138163771.0, + "step": 3557 + }, + { + "epoch": 0.452614171225035, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.903023719787598, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.866191565990448, + "num_tokens": 138197226.0, + "step": 3558 + }, + { + "epoch": 0.4527413815036255, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.723302841186523, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8768236637115479, + "num_tokens": 138240194.0, + "step": 3559 + }, + { + "epoch": 0.452868591782216, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.813432693481445, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8805676698684692, + "num_tokens": 138279070.0, + "step": 3560 + }, + { + "epoch": 0.45299580206080653, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.902322292327881, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8866181373596191, + "num_tokens": 138312837.0, + "step": 3561 + }, + { + "epoch": 0.453123012339397, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.765019416809082, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8822438716888428, + "num_tokens": 138350017.0, + "step": 3562 + }, + { + "epoch": 0.45325022261798753, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5616416931152344e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.858264923095703, + "learning_rate": 1e-06, + "loss": 0.3988, + "mean_token_accuracy": 0.8646147847175598, + "num_tokens": 138387719.0, + "step": 3563 + }, + { + "epoch": 0.45337743289657806, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5735626220703125e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.910645484924316, + "learning_rate": 1e-06, + "loss": 0.4104, + "mean_token_accuracy": 0.8642080426216125, + "num_tokens": 138424567.0, + "step": 3564 + }, + { + "epoch": 0.45350464317516853, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5735626220703125e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.947263717651367, + "learning_rate": 1e-06, + "loss": 0.4229, + "mean_token_accuracy": 0.8569854497909546, + "num_tokens": 138459602.0, + "step": 3565 + }, + { + "epoch": 0.45363185345375906, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5735626220703125e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.833360195159912, + "learning_rate": 1e-06, + "loss": 0.3953, + "mean_token_accuracy": 0.8641966581344604, + "num_tokens": 138494616.0, + "step": 3566 + }, + { + "epoch": 0.4537590637323496, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5735626220703125e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.70172643661499, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8798388838768005, + "num_tokens": 138537505.0, + "step": 3567 + }, + { + "epoch": 0.45388627401094006, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5735626220703125e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.767991065979004, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8741106390953064, + "num_tokens": 138581259.0, + "step": 3568 + }, + { + "epoch": 0.4540134842895306, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5735626220703125e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.784537315368652, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8768033981323242, + "num_tokens": 138620996.0, + "step": 3569 + }, + { + "epoch": 0.4541406945681211, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5735626220703125e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.797730922698975, + "learning_rate": 1e-06, + "loss": 0.3936, + "mean_token_accuracy": 0.8660089373588562, + "num_tokens": 138661406.0, + "step": 3570 + }, + { + "epoch": 0.4542679048467116, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5735626220703125e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.884028911590576, + "learning_rate": 1e-06, + "loss": 0.436, + "mean_token_accuracy": 0.8548584580421448, + "num_tokens": 138701005.0, + "step": 3571 + }, + { + "epoch": 0.4543951151253021, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.792518138885498, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8723708391189575, + "num_tokens": 138743141.0, + "step": 3572 + }, + { + "epoch": 0.45452232540389265, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.863677978515625, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8748632073402405, + "num_tokens": 138777192.0, + "step": 3573 + }, + { + "epoch": 0.4546495356824831, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.794469833374023, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8839144706726074, + "num_tokens": 138814626.0, + "step": 3574 + }, + { + "epoch": 0.45477674596107365, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.846855163574219, + "learning_rate": 1e-06, + "loss": 0.44, + "mean_token_accuracy": 0.8514459133148193, + "num_tokens": 138855606.0, + "step": 3575 + }, + { + "epoch": 0.4549039562396642, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.8105573654174805, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8684706091880798, + "num_tokens": 138894830.0, + "step": 3576 + }, + { + "epoch": 0.45503116651825465, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.828211784362793, + "learning_rate": 1e-06, + "loss": 0.3689, + "mean_token_accuracy": 0.8766074180603027, + "num_tokens": 138931842.0, + "step": 3577 + }, + { + "epoch": 0.4551583767968452, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.891415596008301, + "learning_rate": 1e-06, + "loss": 0.424, + "mean_token_accuracy": 0.8589234352111816, + "num_tokens": 138969572.0, + "step": 3578 + }, + { + "epoch": 0.4552855870754357, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.883009910583496, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8821839094161987, + "num_tokens": 139000671.0, + "step": 3579 + }, + { + "epoch": 0.4554127973540262, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.807242393493652, + "learning_rate": 1e-06, + "loss": 0.3717, + "mean_token_accuracy": 0.8727126121520996, + "num_tokens": 139039048.0, + "step": 3580 + }, + { + "epoch": 0.4555400076326167, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.8684492111206055, + "learning_rate": 1e-06, + "loss": 0.3919, + "mean_token_accuracy": 0.8678851127624512, + "num_tokens": 139076124.0, + "step": 3581 + }, + { + "epoch": 0.45566721791120723, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.810619831085205, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8799856305122375, + "num_tokens": 139114528.0, + "step": 3582 + }, + { + "epoch": 0.45579442818979776, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.779047966003418, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.873932957649231, + "num_tokens": 139153356.0, + "step": 3583 + }, + { + "epoch": 0.45592163846838824, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.979734420776367, + "learning_rate": 1e-06, + "loss": 0.4046, + "mean_token_accuracy": 0.8686526417732239, + "num_tokens": 139188307.0, + "step": 3584 + }, + { + "epoch": 0.45604884874697876, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.69287109375, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8802890777587891, + "num_tokens": 139231114.0, + "step": 3585 + }, + { + "epoch": 0.4561760590255693, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.9100022315979, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8692562580108643, + "num_tokens": 139265405.0, + "step": 3586 + }, + { + "epoch": 0.45630326930415976, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5854835510253906e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.712224006652832, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8757078647613525, + "num_tokens": 139310024.0, + "step": 3587 + }, + { + "epoch": 0.4564304795827503, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.856405735015869, + "learning_rate": 1e-06, + "loss": 0.3975, + "mean_token_accuracy": 0.8676261901855469, + "num_tokens": 139346080.0, + "step": 3588 + }, + { + "epoch": 0.4565576898613408, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.756608963012695, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.8756996393203735, + "num_tokens": 139386268.0, + "step": 3589 + }, + { + "epoch": 0.4566849001399313, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 9.113179206848145, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8764978647232056, + "num_tokens": 139431201.0, + "step": 3590 + }, + { + "epoch": 0.4568121104185218, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.925229072570801, + "learning_rate": 1e-06, + "loss": 0.4164, + "mean_token_accuracy": 0.8639472723007202, + "num_tokens": 139470874.0, + "step": 3591 + }, + { + "epoch": 0.45693932069711235, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.592149257659912, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8837775588035583, + "num_tokens": 139512278.0, + "step": 3592 + }, + { + "epoch": 0.4570665309757028, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.916069030761719, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8732877969741821, + "num_tokens": 139549760.0, + "step": 3593 + }, + { + "epoch": 0.45719374125429335, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.860291481018066, + "learning_rate": 1e-06, + "loss": 0.4094, + "mean_token_accuracy": 0.8644897937774658, + "num_tokens": 139589424.0, + "step": 3594 + }, + { + "epoch": 0.4573209515328839, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.895118713378906, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.8803383708000183, + "num_tokens": 139628538.0, + "step": 3595 + }, + { + "epoch": 0.45744816181147435, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.8255181312561035, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8849418759346008, + "num_tokens": 139667229.0, + "step": 3596 + }, + { + "epoch": 0.4575753720900649, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.827434539794922, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8779467940330505, + "num_tokens": 139704220.0, + "step": 3597 + }, + { + "epoch": 0.4577025823686554, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.920855522155762, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8776290416717529, + "num_tokens": 139740441.0, + "step": 3598 + }, + { + "epoch": 0.4578297926472459, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.747128486633301, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8692635297775269, + "num_tokens": 139784663.0, + "step": 3599 + }, + { + "epoch": 0.4579570029258364, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.887072563171387, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.8678032159805298, + "num_tokens": 139823725.0, + "step": 3600 + }, + { + "epoch": 0.45808421320442694, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.824853420257568, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8696682453155518, + "num_tokens": 139863288.0, + "step": 3601 + }, + { + "epoch": 0.4582114234830174, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 8.028468132019043, + "learning_rate": 1e-06, + "loss": 0.4289, + "mean_token_accuracy": 0.8597403764724731, + "num_tokens": 139895797.0, + "step": 3602 + }, + { + "epoch": 0.45833863376160794, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.813334941864014, + "learning_rate": 1e-06, + "loss": 0.4009, + "mean_token_accuracy": 0.8641982078552246, + "num_tokens": 139937830.0, + "step": 3603 + }, + { + "epoch": 0.45846584404019847, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.816288471221924, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.8731153607368469, + "num_tokens": 139975032.0, + "step": 3604 + }, + { + "epoch": 0.45859305431878894, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.883237838745117, + "learning_rate": 1e-06, + "loss": 0.4607, + "mean_token_accuracy": 0.8472616076469421, + "num_tokens": 140014450.0, + "step": 3605 + }, + { + "epoch": 0.45872026459737947, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.787362575531006, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.877994954586029, + "num_tokens": 140053267.0, + "step": 3606 + }, + { + "epoch": 0.45884747487597, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.5974044799804688e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.808522701263428, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8730686902999878, + "num_tokens": 140094414.0, + "step": 3607 + }, + { + "epoch": 0.45897468515456047, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.609325408935547e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.832126617431641, + "learning_rate": 1e-06, + "loss": 0.3831, + "mean_token_accuracy": 0.8727527856826782, + "num_tokens": 140134963.0, + "step": 3608 + }, + { + "epoch": 0.459101895433151, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.609325408935547e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.8819050788879395, + "learning_rate": 1e-06, + "loss": 0.4362, + "mean_token_accuracy": 0.8525786399841309, + "num_tokens": 140170562.0, + "step": 3609 + }, + { + "epoch": 0.4592291057117415, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.609325408935547e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.895169734954834, + "learning_rate": 1e-06, + "loss": 0.4017, + "mean_token_accuracy": 0.8646773099899292, + "num_tokens": 140205412.0, + "step": 3610 + }, + { + "epoch": 0.459356315990332, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.784888744354248, + "learning_rate": 1e-06, + "loss": 0.3993, + "mean_token_accuracy": 0.8651116490364075, + "num_tokens": 140250236.0, + "step": 3611 + }, + { + "epoch": 0.4594835262689225, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.789690017700195, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8803107738494873, + "num_tokens": 140285912.0, + "step": 3612 + }, + { + "epoch": 0.45961073654751305, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.609325408935547e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.857548236846924, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8772595524787903, + "num_tokens": 140319663.0, + "step": 3613 + }, + { + "epoch": 0.4597379468261035, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.609325408935547e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.901662826538086, + "learning_rate": 1e-06, + "loss": 0.3859, + "mean_token_accuracy": 0.8726856112480164, + "num_tokens": 140356174.0, + "step": 3614 + }, + { + "epoch": 0.45986515710469406, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.609325408935547e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.719616413116455, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8843541741371155, + "num_tokens": 140399171.0, + "step": 3615 + }, + { + "epoch": 0.4599923673832846, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.609325408935547e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 8.001280784606934, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8849666118621826, + "num_tokens": 140428015.0, + "step": 3616 + }, + { + "epoch": 0.46011957766187506, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.609325408935547e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.861303806304932, + "learning_rate": 1e-06, + "loss": 0.3775, + "mean_token_accuracy": 0.8744734525680542, + "num_tokens": 140467785.0, + "step": 3617 + }, + { + "epoch": 0.4602467879404656, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.609325408935547e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.9407877922058105, + "learning_rate": 1e-06, + "loss": 0.4076, + "mean_token_accuracy": 0.8623608946800232, + "num_tokens": 140502518.0, + "step": 3618 + }, + { + "epoch": 0.4603739982190561, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.609325408935547e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.837072372436523, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.8751429319381714, + "num_tokens": 140541002.0, + "step": 3619 + }, + { + "epoch": 0.4605012084976466, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.609325408935547e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.8523640632629395, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.879318118095398, + "num_tokens": 140578193.0, + "step": 3620 + }, + { + "epoch": 0.4606284187762371, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.804715156555176, + "learning_rate": 1e-06, + "loss": 0.3734, + "mean_token_accuracy": 0.8732238411903381, + "num_tokens": 140618092.0, + "step": 3621 + }, + { + "epoch": 0.46075562905482764, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.846673011779785, + "learning_rate": 1e-06, + "loss": 0.4533, + "mean_token_accuracy": 0.8472900986671448, + "num_tokens": 140659282.0, + "step": 3622 + }, + { + "epoch": 0.4608828393334181, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.776344299316406, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8835222125053406, + "num_tokens": 140695996.0, + "step": 3623 + }, + { + "epoch": 0.46101004961200864, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.7627716064453125, + "learning_rate": 1e-06, + "loss": 0.3945, + "mean_token_accuracy": 0.8675907254219055, + "num_tokens": 140739656.0, + "step": 3624 + }, + { + "epoch": 0.46113725989059917, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.781667232513428, + "learning_rate": 1e-06, + "loss": 0.4026, + "mean_token_accuracy": 0.8706505298614502, + "num_tokens": 140783605.0, + "step": 3625 + }, + { + "epoch": 0.46126447016918964, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.822208881378174, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8894040584564209, + "num_tokens": 140821365.0, + "step": 3626 + }, + { + "epoch": 0.4613916804477802, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.860384941101074, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8806426525115967, + "num_tokens": 140856687.0, + "step": 3627 + }, + { + "epoch": 0.4615188907263707, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.772457122802734, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8802648782730103, + "num_tokens": 140896893.0, + "step": 3628 + }, + { + "epoch": 0.4616461010049612, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.9292120933532715, + "learning_rate": 1e-06, + "loss": 0.4138, + "mean_token_accuracy": 0.858975887298584, + "num_tokens": 140935414.0, + "step": 3629 + }, + { + "epoch": 0.4617733112835517, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.791645526885986, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8741579055786133, + "num_tokens": 140973981.0, + "step": 3630 + }, + { + "epoch": 0.46190052156214223, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.804840564727783, + "learning_rate": 1e-06, + "loss": 0.3922, + "mean_token_accuracy": 0.8665194511413574, + "num_tokens": 141010344.0, + "step": 3631 + }, + { + "epoch": 0.46202773184073276, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.85337495803833, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8862878680229187, + "num_tokens": 141047035.0, + "step": 3632 + }, + { + "epoch": 0.46215494211932323, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 8.023049354553223, + "learning_rate": 1e-06, + "loss": 0.4146, + "mean_token_accuracy": 0.8647449016571045, + "num_tokens": 141079387.0, + "step": 3633 + }, + { + "epoch": 0.46228215239791376, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.989432334899902, + "learning_rate": 1e-06, + "loss": 0.3804, + "mean_token_accuracy": 0.8756598830223083, + "num_tokens": 141111295.0, + "step": 3634 + }, + { + "epoch": 0.4624093626765043, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.672272682189941, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8855462074279785, + "num_tokens": 141152409.0, + "step": 3635 + }, + { + "epoch": 0.46253657295509476, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.893219470977783, + "learning_rate": 1e-06, + "loss": 0.4268, + "mean_token_accuracy": 0.857877254486084, + "num_tokens": 141190435.0, + "step": 3636 + }, + { + "epoch": 0.4626637832336853, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.789970874786377, + "learning_rate": 1e-06, + "loss": 0.3867, + "mean_token_accuracy": 0.8697793483734131, + "num_tokens": 141229628.0, + "step": 3637 + }, + { + "epoch": 0.4627909935122758, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.796812534332275, + "learning_rate": 1e-06, + "loss": 0.3867, + "mean_token_accuracy": 0.8709033727645874, + "num_tokens": 141268241.0, + "step": 3638 + }, + { + "epoch": 0.4629182037908663, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.813101291656494, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.878832221031189, + "num_tokens": 141304902.0, + "step": 3639 + }, + { + "epoch": 0.4630454140694568, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.925466060638428, + "learning_rate": 1e-06, + "loss": 0.3745, + "mean_token_accuracy": 0.8749920725822449, + "num_tokens": 141337826.0, + "step": 3640 + }, + { + "epoch": 0.46317262434804735, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.830562591552734, + "learning_rate": 1e-06, + "loss": 0.3923, + "mean_token_accuracy": 0.871017336845398, + "num_tokens": 141376669.0, + "step": 3641 + }, + { + "epoch": 0.4632998346266378, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.773310661315918, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8805156946182251, + "num_tokens": 141412980.0, + "step": 3642 + }, + { + "epoch": 0.46342704490522835, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.7359819412231445, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8775616884231567, + "num_tokens": 141455056.0, + "step": 3643 + }, + { + "epoch": 0.4635542551838189, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.695283889770508, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8826619386672974, + "num_tokens": 141498992.0, + "step": 3644 + }, + { + "epoch": 0.46368146546240935, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.725275993347168, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8885245323181152, + "num_tokens": 141539861.0, + "step": 3645 + }, + { + "epoch": 0.4638086757409999, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.83616304397583, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8783769607543945, + "num_tokens": 141575493.0, + "step": 3646 + }, + { + "epoch": 0.4639358860195904, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.901788234710693, + "learning_rate": 1e-06, + "loss": 0.4467, + "mean_token_accuracy": 0.8540360927581787, + "num_tokens": 141615790.0, + "step": 3647 + }, + { + "epoch": 0.4640630962981809, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.808488368988037, + "learning_rate": 1e-06, + "loss": 0.3947, + "mean_token_accuracy": 0.8683205246925354, + "num_tokens": 141653715.0, + "step": 3648 + }, + { + "epoch": 0.4641903065767714, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.9267706871032715, + "learning_rate": 1e-06, + "loss": 0.3979, + "mean_token_accuracy": 0.8686600923538208, + "num_tokens": 141687245.0, + "step": 3649 + }, + { + "epoch": 0.46431751685536193, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.759069919586182, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8783621788024902, + "num_tokens": 141728644.0, + "step": 3650 + }, + { + "epoch": 0.4644447271339524, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.716609001159668, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8906548023223877, + "num_tokens": 141771730.0, + "step": 3651 + }, + { + "epoch": 0.46457193741254293, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.968788146972656, + "learning_rate": 1e-06, + "loss": 0.4688, + "mean_token_accuracy": 0.8422913551330566, + "num_tokens": 141814620.0, + "step": 3652 + }, + { + "epoch": 0.46469914769113346, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.828913688659668, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8734959959983826, + "num_tokens": 141851445.0, + "step": 3653 + }, + { + "epoch": 0.46482635796972394, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.864071846008301, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8727130889892578, + "num_tokens": 141886691.0, + "step": 3654 + }, + { + "epoch": 0.46495356824831446, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.925958156585693, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8791699409484863, + "num_tokens": 141923144.0, + "step": 3655 + }, + { + "epoch": 0.465080778526905, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.728919506072998, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8823695182800293, + "num_tokens": 141966578.0, + "step": 3656 + }, + { + "epoch": 0.46520798880549546, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.793769359588623, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8786493539810181, + "num_tokens": 142008749.0, + "step": 3657 + }, + { + "epoch": 0.465335199084086, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.780611038208008, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8746505975723267, + "num_tokens": 142050581.0, + "step": 3658 + }, + { + "epoch": 0.4654624093626765, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.832546710968018, + "learning_rate": 1e-06, + "loss": 0.3738, + "mean_token_accuracy": 0.8734270334243774, + "num_tokens": 142091397.0, + "step": 3659 + }, + { + "epoch": 0.465589619641267, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.867748260498047, + "learning_rate": 1e-06, + "loss": 0.3956, + "mean_token_accuracy": 0.8644598722457886, + "num_tokens": 142129189.0, + "step": 3660 + }, + { + "epoch": 0.4657168299198575, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.788166046142578, + "learning_rate": 1e-06, + "loss": 0.3985, + "mean_token_accuracy": 0.8641284704208374, + "num_tokens": 142172863.0, + "step": 3661 + }, + { + "epoch": 0.46584404019844805, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 7.987022399902344e-06, + "grad_norm": 7.81334114074707, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.8734422922134399, + "num_tokens": 142218609.0, + "step": 3662 + }, + { + "epoch": 0.4659712504770385, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.91056489944458, + "learning_rate": 1e-06, + "loss": 0.4389, + "mean_token_accuracy": 0.8525843024253845, + "num_tokens": 142258147.0, + "step": 3663 + }, + { + "epoch": 0.46609846075562905, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.733455181121826, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8777725696563721, + "num_tokens": 142301662.0, + "step": 3664 + }, + { + "epoch": 0.4662256710342196, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.835389614105225, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8840932250022888, + "num_tokens": 142341868.0, + "step": 3665 + }, + { + "epoch": 0.46635288131281005, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.885875701904297, + "learning_rate": 1e-06, + "loss": 0.4585, + "mean_token_accuracy": 0.8462058305740356, + "num_tokens": 142383203.0, + "step": 3666 + }, + { + "epoch": 0.4664800915914006, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.919382572174072, + "learning_rate": 1e-06, + "loss": 0.4201, + "mean_token_accuracy": 0.8641518354415894, + "num_tokens": 142417829.0, + "step": 3667 + }, + { + "epoch": 0.4666073018699911, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.866537094116211, + "learning_rate": 1e-06, + "loss": 0.4237, + "mean_token_accuracy": 0.8636541366577148, + "num_tokens": 142453180.0, + "step": 3668 + }, + { + "epoch": 0.4667345121485816, + "ewc_loss": 0.0096435546875, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.754925727844238, + "learning_rate": 1e-06, + "loss": 0.4177, + "mean_token_accuracy": 0.8604544401168823, + "num_tokens": 142494271.0, + "step": 3669 + }, + { + "epoch": 0.4668617224271721, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.633167266845703e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.8932695388793945, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.874032735824585, + "num_tokens": 142531250.0, + "step": 3670 + }, + { + "epoch": 0.46698893270576264, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.833217620849609, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8757967948913574, + "num_tokens": 142565746.0, + "step": 3671 + }, + { + "epoch": 0.4671161429843531, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.82330846786499, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.8660306334495544, + "num_tokens": 142604482.0, + "step": 3672 + }, + { + "epoch": 0.46724335326294364, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.92953634262085, + "learning_rate": 1e-06, + "loss": 0.388, + "mean_token_accuracy": 0.8701450228691101, + "num_tokens": 142639510.0, + "step": 3673 + }, + { + "epoch": 0.46737056354153417, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.621246337890625e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.748969078063965, + "learning_rate": 1e-06, + "loss": 0.3947, + "mean_token_accuracy": 0.865852952003479, + "num_tokens": 142678571.0, + "step": 3674 + }, + { + "epoch": 0.46749777382012464, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.633167266845703e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.864599704742432, + "learning_rate": 1e-06, + "loss": 0.403, + "mean_token_accuracy": 0.8612306118011475, + "num_tokens": 142716971.0, + "step": 3675 + }, + { + "epoch": 0.46762498409871517, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.633167266845703e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.868329048156738, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8881849050521851, + "num_tokens": 142754972.0, + "step": 3676 + }, + { + "epoch": 0.4677521943773057, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.633167266845703e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.776193141937256, + "learning_rate": 1e-06, + "loss": 0.4172, + "mean_token_accuracy": 0.8619208931922913, + "num_tokens": 142798270.0, + "step": 3677 + }, + { + "epoch": 0.46787940465589617, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.633167266845703e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.838590145111084, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8754743337631226, + "num_tokens": 142839467.0, + "step": 3678 + }, + { + "epoch": 0.4680066149344867, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.633167266845703e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.814380168914795, + "learning_rate": 1e-06, + "loss": 0.4151, + "mean_token_accuracy": 0.8609256744384766, + "num_tokens": 142879080.0, + "step": 3679 + }, + { + "epoch": 0.4681338252130772, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.633167266845703e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.883268356323242, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.8754586577415466, + "num_tokens": 142914896.0, + "step": 3680 + }, + { + "epoch": 0.46826103549166775, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.633167266845703e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.874848365783691, + "learning_rate": 1e-06, + "loss": 0.4476, + "mean_token_accuracy": 0.851582944393158, + "num_tokens": 142956153.0, + "step": 3681 + }, + { + "epoch": 0.4683882457702582, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.633167266845703e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.837588310241699, + "learning_rate": 1e-06, + "loss": 0.4092, + "mean_token_accuracy": 0.8591137528419495, + "num_tokens": 142997537.0, + "step": 3682 + }, + { + "epoch": 0.46851545604884876, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.633167266845703e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.84037446975708, + "learning_rate": 1e-06, + "loss": 0.4035, + "mean_token_accuracy": 0.8650977611541748, + "num_tokens": 143041119.0, + "step": 3683 + }, + { + "epoch": 0.4686426663274393, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.633167266845703e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 8.036989212036133, + "learning_rate": 1e-06, + "loss": 0.4592, + "mean_token_accuracy": 0.8553178310394287, + "num_tokens": 143074417.0, + "step": 3684 + }, + { + "epoch": 0.46876987660602976, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.633167266845703e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.912391185760498, + "learning_rate": 1e-06, + "loss": 0.3775, + "mean_token_accuracy": 0.8697446584701538, + "num_tokens": 143106657.0, + "step": 3685 + }, + { + "epoch": 0.4688970868846203, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.801033973693848, + "learning_rate": 1e-06, + "loss": 0.4118, + "mean_token_accuracy": 0.8634135127067566, + "num_tokens": 143151919.0, + "step": 3686 + }, + { + "epoch": 0.4690242971632108, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 8.025955200195312, + "learning_rate": 1e-06, + "loss": 0.4205, + "mean_token_accuracy": 0.8581292629241943, + "num_tokens": 143191770.0, + "step": 3687 + }, + { + "epoch": 0.4691515074418013, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.633167266845703e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.838383674621582, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.870335578918457, + "num_tokens": 143227710.0, + "step": 3688 + }, + { + "epoch": 0.4692787177203918, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.8731160163879395, + "learning_rate": 1e-06, + "loss": 0.3966, + "mean_token_accuracy": 0.8630675673484802, + "num_tokens": 143264455.0, + "step": 3689 + }, + { + "epoch": 0.46940592799898234, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.851040363311768, + "learning_rate": 1e-06, + "loss": 0.3978, + "mean_token_accuracy": 0.868843674659729, + "num_tokens": 143307016.0, + "step": 3690 + }, + { + "epoch": 0.4695331382775728, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.839569568634033, + "learning_rate": 1e-06, + "loss": 0.3946, + "mean_token_accuracy": 0.8680799007415771, + "num_tokens": 143348131.0, + "step": 3691 + }, + { + "epoch": 0.46966034855616334, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.905862808227539, + "learning_rate": 1e-06, + "loss": 0.4061, + "mean_token_accuracy": 0.8667870163917542, + "num_tokens": 143384033.0, + "step": 3692 + }, + { + "epoch": 0.46978755883475387, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.757840633392334, + "learning_rate": 1e-06, + "loss": 0.3918, + "mean_token_accuracy": 0.8678860664367676, + "num_tokens": 143426243.0, + "step": 3693 + }, + { + "epoch": 0.46991476911334434, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.783332824707031, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8840065002441406, + "num_tokens": 143465943.0, + "step": 3694 + }, + { + "epoch": 0.47004197939193487, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.881044864654541, + "learning_rate": 1e-06, + "loss": 0.3966, + "mean_token_accuracy": 0.8651041984558105, + "num_tokens": 143505306.0, + "step": 3695 + }, + { + "epoch": 0.4701691896705254, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.920897960662842, + "learning_rate": 1e-06, + "loss": 0.4069, + "mean_token_accuracy": 0.8640402555465698, + "num_tokens": 143538697.0, + "step": 3696 + }, + { + "epoch": 0.4702963999491159, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.722097873687744, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8881542086601257, + "num_tokens": 143582063.0, + "step": 3697 + }, + { + "epoch": 0.4704236102277064, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.909221649169922, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8854889869689941, + "num_tokens": 143616496.0, + "step": 3698 + }, + { + "epoch": 0.47055082050629693, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.866007328033447, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8810611367225647, + "num_tokens": 143652933.0, + "step": 3699 + }, + { + "epoch": 0.4706780307848874, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.897689342498779, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8674678206443787, + "num_tokens": 143688332.0, + "step": 3700 + }, + { + "epoch": 0.47080524106347793, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.972823619842529, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.8720145225524902, + "num_tokens": 143720873.0, + "step": 3701 + }, + { + "epoch": 0.47093245134206846, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.789446830749512, + "learning_rate": 1e-06, + "loss": 0.3886, + "mean_token_accuracy": 0.8693864941596985, + "num_tokens": 143763937.0, + "step": 3702 + }, + { + "epoch": 0.47105966162065893, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.801972389221191, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8802987337112427, + "num_tokens": 143803407.0, + "step": 3703 + }, + { + "epoch": 0.47118687189924946, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.922104358673096, + "learning_rate": 1e-06, + "loss": 0.3834, + "mean_token_accuracy": 0.8696969747543335, + "num_tokens": 143837599.0, + "step": 3704 + }, + { + "epoch": 0.47131408217784, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.892716407775879, + "learning_rate": 1e-06, + "loss": 0.4016, + "mean_token_accuracy": 0.8645453453063965, + "num_tokens": 143871871.0, + "step": 3705 + }, + { + "epoch": 0.47144129245643046, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.956827640533447, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8781832456588745, + "num_tokens": 143903536.0, + "step": 3706 + }, + { + "epoch": 0.471568502735021, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.854487419128418, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8866236805915833, + "num_tokens": 143936383.0, + "step": 3707 + }, + { + "epoch": 0.4716957130136115, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.867105484008789, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8897873163223267, + "num_tokens": 143968861.0, + "step": 3708 + }, + { + "epoch": 0.471822923292202, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.887775897979736, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8815434575080872, + "num_tokens": 144007205.0, + "step": 3709 + }, + { + "epoch": 0.4719501335707925, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.792331695556641, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.8774290680885315, + "num_tokens": 144051224.0, + "step": 3710 + }, + { + "epoch": 0.47207734384938305, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.880638599395752, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8751636743545532, + "num_tokens": 144087304.0, + "step": 3711 + }, + { + "epoch": 0.4722045541279735, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.838748931884766, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8859659433364868, + "num_tokens": 144125190.0, + "step": 3712 + }, + { + "epoch": 0.47233176440656405, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.780391693115234, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8794695734977722, + "num_tokens": 144166469.0, + "step": 3713 + }, + { + "epoch": 0.4724589746851546, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.831964015960693, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8812769651412964, + "num_tokens": 144203844.0, + "step": 3714 + }, + { + "epoch": 0.47258618496374505, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.988022804260254, + "learning_rate": 1e-06, + "loss": 0.4227, + "mean_token_accuracy": 0.8553625345230103, + "num_tokens": 144235538.0, + "step": 3715 + }, + { + "epoch": 0.4727133952423356, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.929291725158691, + "learning_rate": 1e-06, + "loss": 0.4282, + "mean_token_accuracy": 0.8585827946662903, + "num_tokens": 144269349.0, + "step": 3716 + }, + { + "epoch": 0.4728406055209261, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.722901344299316, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8797914981842041, + "num_tokens": 144310020.0, + "step": 3717 + }, + { + "epoch": 0.4729678157995166, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.744480609893799, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8802080154418945, + "num_tokens": 144349587.0, + "step": 3718 + }, + { + "epoch": 0.4730950260781071, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.79289436340332, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8867896199226379, + "num_tokens": 144388454.0, + "step": 3719 + }, + { + "epoch": 0.47322223635669763, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.787805080413818, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8847619295120239, + "num_tokens": 144429477.0, + "step": 3720 + }, + { + "epoch": 0.4733494466352881, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.858898162841797, + "learning_rate": 1e-06, + "loss": 0.3915, + "mean_token_accuracy": 0.8677686452865601, + "num_tokens": 144469433.0, + "step": 3721 + }, + { + "epoch": 0.47347665691387864, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.8338398933410645, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8718793392181396, + "num_tokens": 144508417.0, + "step": 3722 + }, + { + "epoch": 0.47360386719246916, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.999187469482422, + "learning_rate": 1e-06, + "loss": 0.3793, + "mean_token_accuracy": 0.8702313899993896, + "num_tokens": 144540762.0, + "step": 3723 + }, + { + "epoch": 0.47373107747105964, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.976960182189941, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.876659631729126, + "num_tokens": 144571474.0, + "step": 3724 + }, + { + "epoch": 0.47385828774965016, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.792625427246094, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8888903260231018, + "num_tokens": 144608879.0, + "step": 3725 + }, + { + "epoch": 0.4739854980282407, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.963276386260986, + "learning_rate": 1e-06, + "loss": 0.4425, + "mean_token_accuracy": 0.8581606149673462, + "num_tokens": 144644507.0, + "step": 3726 + }, + { + "epoch": 0.47411270830683117, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.743216514587402, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8703956007957458, + "num_tokens": 144689066.0, + "step": 3727 + }, + { + "epoch": 0.4742399185854217, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.824525356292725, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8769926428794861, + "num_tokens": 144725992.0, + "step": 3728 + }, + { + "epoch": 0.4743671288640122, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.928281784057617, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8732332587242126, + "num_tokens": 144765476.0, + "step": 3729 + }, + { + "epoch": 0.4744943391426027, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.7154221534729, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8779234290122986, + "num_tokens": 144813245.0, + "step": 3730 + }, + { + "epoch": 0.4746215494211932, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.977530479431152, + "learning_rate": 1e-06, + "loss": 0.4356, + "mean_token_accuracy": 0.8597427606582642, + "num_tokens": 144850535.0, + "step": 3731 + }, + { + "epoch": 0.47474875969978375, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.9561052322387695, + "learning_rate": 1e-06, + "loss": 0.4079, + "mean_token_accuracy": 0.8644176721572876, + "num_tokens": 144887784.0, + "step": 3732 + }, + { + "epoch": 0.4748759699783743, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.877956867218018, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.8706070780754089, + "num_tokens": 144921138.0, + "step": 3733 + }, + { + "epoch": 0.47500318025696475, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.822907447814941, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8842390775680542, + "num_tokens": 144958006.0, + "step": 3734 + }, + { + "epoch": 0.4751303905355553, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.871980667114258, + "learning_rate": 1e-06, + "loss": 0.4089, + "mean_token_accuracy": 0.8637840747833252, + "num_tokens": 144998957.0, + "step": 3735 + }, + { + "epoch": 0.4752576008141458, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.935018539428711, + "learning_rate": 1e-06, + "loss": 0.4332, + "mean_token_accuracy": 0.8562939763069153, + "num_tokens": 145034454.0, + "step": 3736 + }, + { + "epoch": 0.4753848110927363, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.866324424743652, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8764729499816895, + "num_tokens": 145075706.0, + "step": 3737 + }, + { + "epoch": 0.4755120213713268, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.76666259765625, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8799021244049072, + "num_tokens": 145113402.0, + "step": 3738 + }, + { + "epoch": 0.47563923164991734, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.888873100280762, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8690919876098633, + "num_tokens": 145152060.0, + "step": 3739 + }, + { + "epoch": 0.4757664419285078, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.762037754058838, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.877363920211792, + "num_tokens": 145194611.0, + "step": 3740 + }, + { + "epoch": 0.47589365220709834, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.812558174133301, + "learning_rate": 1e-06, + "loss": 0.4042, + "mean_token_accuracy": 0.8653854131698608, + "num_tokens": 145237551.0, + "step": 3741 + }, + { + "epoch": 0.47602086248568887, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.990296363830566, + "learning_rate": 1e-06, + "loss": 0.4324, + "mean_token_accuracy": 0.8530391454696655, + "num_tokens": 145271111.0, + "step": 3742 + }, + { + "epoch": 0.47614807276427934, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.810403823852539, + "learning_rate": 1e-06, + "loss": 0.396, + "mean_token_accuracy": 0.8680465817451477, + "num_tokens": 145313203.0, + "step": 3743 + }, + { + "epoch": 0.47627528304286987, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.930628299713135, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8736065626144409, + "num_tokens": 145347542.0, + "step": 3744 + }, + { + "epoch": 0.4764024933214604, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.889101505279541, + "learning_rate": 1e-06, + "loss": 0.3788, + "mean_token_accuracy": 0.8707031011581421, + "num_tokens": 145383349.0, + "step": 3745 + }, + { + "epoch": 0.47652970360005087, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.86427640914917, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8850415945053101, + "num_tokens": 145419598.0, + "step": 3746 + }, + { + "epoch": 0.4766569138786414, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.8403000831604, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8829518556594849, + "num_tokens": 145454550.0, + "step": 3747 + }, + { + "epoch": 0.4767841241572319, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.875095367431641, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8836458325386047, + "num_tokens": 145492415.0, + "step": 3748 + }, + { + "epoch": 0.4769113344358224, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.820606708526611, + "learning_rate": 1e-06, + "loss": 0.4043, + "mean_token_accuracy": 0.8662118911743164, + "num_tokens": 145536551.0, + "step": 3749 + }, + { + "epoch": 0.4770385447144129, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.854363441467285, + "learning_rate": 1e-06, + "loss": 0.4411, + "mean_token_accuracy": 0.8522766828536987, + "num_tokens": 145579387.0, + "step": 3750 + }, + { + "epoch": 0.47716575499300345, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.911800861358643, + "learning_rate": 1e-06, + "loss": 0.4346, + "mean_token_accuracy": 0.8507481217384338, + "num_tokens": 145617311.0, + "step": 3751 + }, + { + "epoch": 0.4772929652715939, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.846258640289307, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8815074563026428, + "num_tokens": 145652754.0, + "step": 3752 + }, + { + "epoch": 0.47742017555018446, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.83926248550415, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.883545994758606, + "num_tokens": 145690811.0, + "step": 3753 + }, + { + "epoch": 0.477547385828775, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.895270347595215, + "learning_rate": 1e-06, + "loss": 0.4023, + "mean_token_accuracy": 0.8668138980865479, + "num_tokens": 145728181.0, + "step": 3754 + }, + { + "epoch": 0.47767459610736546, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.829178810119629, + "learning_rate": 1e-06, + "loss": 0.4048, + "mean_token_accuracy": 0.8626424074172974, + "num_tokens": 145770607.0, + "step": 3755 + }, + { + "epoch": 0.477801806385956, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.903163909912109, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.876219630241394, + "num_tokens": 145804787.0, + "step": 3756 + }, + { + "epoch": 0.4779290166645465, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.899669647216797, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8650179505348206, + "num_tokens": 145840741.0, + "step": 3757 + }, + { + "epoch": 0.478056226943137, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.805687427520752, + "learning_rate": 1e-06, + "loss": 0.3786, + "mean_token_accuracy": 0.8750937581062317, + "num_tokens": 145876844.0, + "step": 3758 + }, + { + "epoch": 0.4781834372217275, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.907105445861816, + "learning_rate": 1e-06, + "loss": 0.3868, + "mean_token_accuracy": 0.8691626191139221, + "num_tokens": 145914895.0, + "step": 3759 + }, + { + "epoch": 0.47831064750031804, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.9006667137146, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8718543648719788, + "num_tokens": 145954434.0, + "step": 3760 + }, + { + "epoch": 0.4784378577789085, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.879955291748047, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8769645690917969, + "num_tokens": 145991900.0, + "step": 3761 + }, + { + "epoch": 0.47856506805749904, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.888546943664551, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8800734281539917, + "num_tokens": 146026326.0, + "step": 3762 + }, + { + "epoch": 0.47869227833608957, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.9226555824279785, + "learning_rate": 1e-06, + "loss": 0.3963, + "mean_token_accuracy": 0.8671059012413025, + "num_tokens": 146064201.0, + "step": 3763 + }, + { + "epoch": 0.47881948861468004, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.8899006843566895, + "learning_rate": 1e-06, + "loss": 0.4006, + "mean_token_accuracy": 0.8702435493469238, + "num_tokens": 146099623.0, + "step": 3764 + }, + { + "epoch": 0.4789466988932706, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.919042110443115, + "learning_rate": 1e-06, + "loss": 0.4422, + "mean_token_accuracy": 0.851219654083252, + "num_tokens": 146138665.0, + "step": 3765 + }, + { + "epoch": 0.4790739091718611, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.969348907470703, + "learning_rate": 1e-06, + "loss": 0.4333, + "mean_token_accuracy": 0.8546338081359863, + "num_tokens": 146175546.0, + "step": 3766 + }, + { + "epoch": 0.4792011194504516, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.9086198806762695, + "learning_rate": 1e-06, + "loss": 0.4111, + "mean_token_accuracy": 0.8613579869270325, + "num_tokens": 146214518.0, + "step": 3767 + }, + { + "epoch": 0.4793283297290421, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.790027618408203, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.8791813850402832, + "num_tokens": 146254315.0, + "step": 3768 + }, + { + "epoch": 0.47945554000763263, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.903704643249512, + "learning_rate": 1e-06, + "loss": 0.3979, + "mean_token_accuracy": 0.865035355091095, + "num_tokens": 146289787.0, + "step": 3769 + }, + { + "epoch": 0.4795827502862231, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.715851783752441, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8842063546180725, + "num_tokens": 146336118.0, + "step": 3770 + }, + { + "epoch": 0.47970996056481363, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.833721160888672, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8915942311286926, + "num_tokens": 146370774.0, + "step": 3771 + }, + { + "epoch": 0.47983717084340416, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.766952991485596, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8847465515136719, + "num_tokens": 146409679.0, + "step": 3772 + }, + { + "epoch": 0.47996438112199463, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.928300857543945, + "learning_rate": 1e-06, + "loss": 0.3865, + "mean_token_accuracy": 0.8703435659408569, + "num_tokens": 146447342.0, + "step": 3773 + }, + { + "epoch": 0.48009159140058516, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.8414459228515625, + "learning_rate": 1e-06, + "loss": 0.3752, + "mean_token_accuracy": 0.8749457001686096, + "num_tokens": 146491359.0, + "step": 3774 + }, + { + "epoch": 0.4802188016791757, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.776801109313965, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8816412091255188, + "num_tokens": 146534991.0, + "step": 3775 + }, + { + "epoch": 0.48034601195776616, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.832269668579102, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8817484378814697, + "num_tokens": 146573258.0, + "step": 3776 + }, + { + "epoch": 0.4804732222363567, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.8109917640686035, + "learning_rate": 1e-06, + "loss": 0.3981, + "mean_token_accuracy": 0.8653495907783508, + "num_tokens": 146613610.0, + "step": 3777 + }, + { + "epoch": 0.4806004325149472, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.876372814178467, + "learning_rate": 1e-06, + "loss": 0.3897, + "mean_token_accuracy": 0.8671501874923706, + "num_tokens": 146653970.0, + "step": 3778 + }, + { + "epoch": 0.4807276427935377, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.792504787445068, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8773742914199829, + "num_tokens": 146695966.0, + "step": 3779 + }, + { + "epoch": 0.4808548530721282, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.908420562744141, + "learning_rate": 1e-06, + "loss": 0.3961, + "mean_token_accuracy": 0.8673766255378723, + "num_tokens": 146736933.0, + "step": 3780 + }, + { + "epoch": 0.48098206335071875, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.776760578155518, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8770625591278076, + "num_tokens": 146780750.0, + "step": 3781 + }, + { + "epoch": 0.4811092736293093, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.997234344482422, + "learning_rate": 1e-06, + "loss": 0.423, + "mean_token_accuracy": 0.8512284755706787, + "num_tokens": 146816706.0, + "step": 3782 + }, + { + "epoch": 0.48123648390789975, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.889039993286133, + "learning_rate": 1e-06, + "loss": 0.3849, + "mean_token_accuracy": 0.8728008270263672, + "num_tokens": 146855995.0, + "step": 3783 + }, + { + "epoch": 0.4813636941864903, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.779473781585693, + "learning_rate": 1e-06, + "loss": 0.4107, + "mean_token_accuracy": 0.8629120588302612, + "num_tokens": 146900564.0, + "step": 3784 + }, + { + "epoch": 0.4814909044650808, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.926390647888184, + "learning_rate": 1e-06, + "loss": 0.41, + "mean_token_accuracy": 0.8619312047958374, + "num_tokens": 146937647.0, + "step": 3785 + }, + { + "epoch": 0.4816181147436713, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.88584041595459, + "learning_rate": 1e-06, + "loss": 0.3859, + "mean_token_accuracy": 0.8657432198524475, + "num_tokens": 146974600.0, + "step": 3786 + }, + { + "epoch": 0.4817453250222618, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.793871879577637, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8812234401702881, + "num_tokens": 147017503.0, + "step": 3787 + }, + { + "epoch": 0.48187253530085233, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.83432149887085, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.881259560585022, + "num_tokens": 147057890.0, + "step": 3788 + }, + { + "epoch": 0.4819997455794428, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.766726016998291, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.8681888580322266, + "num_tokens": 147101893.0, + "step": 3789 + }, + { + "epoch": 0.48212695585803333, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.870898246765137, + "learning_rate": 1e-06, + "loss": 0.3878, + "mean_token_accuracy": 0.8671220541000366, + "num_tokens": 147143821.0, + "step": 3790 + }, + { + "epoch": 0.48225416613662386, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.832900047302246, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8771558403968811, + "num_tokens": 147184860.0, + "step": 3791 + }, + { + "epoch": 0.48238137641521434, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.949424743652344, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8769241571426392, + "num_tokens": 147220914.0, + "step": 3792 + }, + { + "epoch": 0.48250858669380486, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.854422569274902, + "learning_rate": 1e-06, + "loss": 0.4188, + "mean_token_accuracy": 0.8615018129348755, + "num_tokens": 147262536.0, + "step": 3793 + }, + { + "epoch": 0.4826357969723954, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.843193054199219, + "learning_rate": 1e-06, + "loss": 0.3818, + "mean_token_accuracy": 0.870418906211853, + "num_tokens": 147303614.0, + "step": 3794 + }, + { + "epoch": 0.48276300725098586, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.884812831878662, + "learning_rate": 1e-06, + "loss": 0.4022, + "mean_token_accuracy": 0.8676934242248535, + "num_tokens": 147344504.0, + "step": 3795 + }, + { + "epoch": 0.4828902175295764, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.856328964233398, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8783358335494995, + "num_tokens": 147383452.0, + "step": 3796 + }, + { + "epoch": 0.4830174278081669, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.8219499588012695, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8749996423721313, + "num_tokens": 147423590.0, + "step": 3797 + }, + { + "epoch": 0.4831446380867574, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.83643102645874, + "learning_rate": 1e-06, + "loss": 0.3793, + "mean_token_accuracy": 0.870730459690094, + "num_tokens": 147464418.0, + "step": 3798 + }, + { + "epoch": 0.4832718483653479, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.879634380340576, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8726453185081482, + "num_tokens": 147499824.0, + "step": 3799 + }, + { + "epoch": 0.48339905864393845, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6450881958007812e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.797612190246582, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8856147527694702, + "num_tokens": 147539461.0, + "step": 3800 + }, + { + "epoch": 0.4835262689225289, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.830348491668701, + "learning_rate": 1e-06, + "loss": 0.3752, + "mean_token_accuracy": 0.8703488111495972, + "num_tokens": 147581473.0, + "step": 3801 + }, + { + "epoch": 0.48365347920111945, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.922055244445801, + "learning_rate": 1e-06, + "loss": 0.3894, + "mean_token_accuracy": 0.8661432266235352, + "num_tokens": 147615003.0, + "step": 3802 + }, + { + "epoch": 0.48378068947971, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.8753252029418945, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8794132471084595, + "num_tokens": 147650446.0, + "step": 3803 + }, + { + "epoch": 0.48390789975830045, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.829500675201416, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8848482370376587, + "num_tokens": 147686408.0, + "step": 3804 + }, + { + "epoch": 0.484035110036891, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.882535457611084, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8708018064498901, + "num_tokens": 147722680.0, + "step": 3805 + }, + { + "epoch": 0.4841623203154815, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.779446125030518, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8914011120796204, + "num_tokens": 147761083.0, + "step": 3806 + }, + { + "epoch": 0.484289530594072, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.771460056304932, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8821274638175964, + "num_tokens": 147804131.0, + "step": 3807 + }, + { + "epoch": 0.4844167408726625, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.91867208480835, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8767064213752747, + "num_tokens": 147841273.0, + "step": 3808 + }, + { + "epoch": 0.48454395115125304, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.787685394287109, + "learning_rate": 1e-06, + "loss": 0.3664, + "mean_token_accuracy": 0.8762539625167847, + "num_tokens": 147884347.0, + "step": 3809 + }, + { + "epoch": 0.4846711614298435, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.959986209869385, + "learning_rate": 1e-06, + "loss": 0.4088, + "mean_token_accuracy": 0.8654829859733582, + "num_tokens": 147921910.0, + "step": 3810 + }, + { + "epoch": 0.48479837170843404, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.918105125427246, + "learning_rate": 1e-06, + "loss": 0.4115, + "mean_token_accuracy": 0.8625649809837341, + "num_tokens": 147958513.0, + "step": 3811 + }, + { + "epoch": 0.48492558198702457, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.814183712005615, + "learning_rate": 1e-06, + "loss": 0.3806, + "mean_token_accuracy": 0.8727571368217468, + "num_tokens": 147996237.0, + "step": 3812 + }, + { + "epoch": 0.48505279226561504, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 7.878223419189453, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8737046122550964, + "num_tokens": 148038669.0, + "step": 3813 + }, + { + "epoch": 0.48518000254420557, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.816563129425049, + "learning_rate": 1e-06, + "loss": 0.3654, + "mean_token_accuracy": 0.8805129528045654, + "num_tokens": 148075887.0, + "step": 3814 + }, + { + "epoch": 0.4853072128227961, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.724583625793457, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8911281824111938, + "num_tokens": 148115598.0, + "step": 3815 + }, + { + "epoch": 0.48543442310138657, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.875253677368164, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8855929374694824, + "num_tokens": 148152559.0, + "step": 3816 + }, + { + "epoch": 0.4855616333799771, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.8843255043029785, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8857555389404297, + "num_tokens": 148188309.0, + "step": 3817 + }, + { + "epoch": 0.4856888436585676, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 8.022943496704102, + "learning_rate": 1e-06, + "loss": 0.4479, + "mean_token_accuracy": 0.8553231358528137, + "num_tokens": 148223799.0, + "step": 3818 + }, + { + "epoch": 0.4858160539371581, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.875435829162598, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8755770921707153, + "num_tokens": 148262465.0, + "step": 3819 + }, + { + "epoch": 0.4859432642157486, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.988945007324219, + "learning_rate": 1e-06, + "loss": 0.4103, + "mean_token_accuracy": 0.861495852470398, + "num_tokens": 148296619.0, + "step": 3820 + }, + { + "epoch": 0.48607047449433916, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.843638896942139, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8770867586135864, + "num_tokens": 148333028.0, + "step": 3821 + }, + { + "epoch": 0.48619768477292963, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.922115325927734, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8812052607536316, + "num_tokens": 148370408.0, + "step": 3822 + }, + { + "epoch": 0.48632489505152016, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.858909606933594, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.878271758556366, + "num_tokens": 148408445.0, + "step": 3823 + }, + { + "epoch": 0.4864521053301107, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.893612861633301, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8733313679695129, + "num_tokens": 148448217.0, + "step": 3824 + }, + { + "epoch": 0.48657931560870116, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.835336208343506, + "learning_rate": 1e-06, + "loss": 0.4019, + "mean_token_accuracy": 0.8689765930175781, + "num_tokens": 148487674.0, + "step": 3825 + }, + { + "epoch": 0.4867065258872917, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.752642631530762, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.8722931146621704, + "num_tokens": 148537237.0, + "step": 3826 + }, + { + "epoch": 0.4868337361658822, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.878185749053955, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8722534775733948, + "num_tokens": 148578264.0, + "step": 3827 + }, + { + "epoch": 0.4869609464444727, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.8352580070495605, + "learning_rate": 1e-06, + "loss": 0.3921, + "mean_token_accuracy": 0.8724566698074341, + "num_tokens": 148619311.0, + "step": 3828 + }, + { + "epoch": 0.4870881567230632, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.834413051605225, + "learning_rate": 1e-06, + "loss": 0.4034, + "mean_token_accuracy": 0.8655286431312561, + "num_tokens": 148664046.0, + "step": 3829 + }, + { + "epoch": 0.48721536700165374, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.86591911315918, + "learning_rate": 1e-06, + "loss": 0.3793, + "mean_token_accuracy": 0.8706532716751099, + "num_tokens": 148702486.0, + "step": 3830 + }, + { + "epoch": 0.48734257728024427, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.870757579803467, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8821842670440674, + "num_tokens": 148744362.0, + "step": 3831 + }, + { + "epoch": 0.48746978755883474, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.81213903427124, + "learning_rate": 1e-06, + "loss": 0.3892, + "mean_token_accuracy": 0.8702479004859924, + "num_tokens": 148788034.0, + "step": 3832 + }, + { + "epoch": 0.48759699783742527, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.905616283416748, + "learning_rate": 1e-06, + "loss": 0.4471, + "mean_token_accuracy": 0.8482264876365662, + "num_tokens": 148827699.0, + "step": 3833 + }, + { + "epoch": 0.4877242081160158, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.908505439758301, + "learning_rate": 1e-06, + "loss": 0.425, + "mean_token_accuracy": 0.8587135076522827, + "num_tokens": 148872401.0, + "step": 3834 + }, + { + "epoch": 0.4878514183946063, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6570091247558594e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.871208190917969, + "learning_rate": 1e-06, + "loss": 0.3987, + "mean_token_accuracy": 0.8662770986557007, + "num_tokens": 148912374.0, + "step": 3835 + }, + { + "epoch": 0.4879786286731968, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.833763599395752, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8793004751205444, + "num_tokens": 148948963.0, + "step": 3836 + }, + { + "epoch": 0.48810583895178733, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.812099933624268, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8800452947616577, + "num_tokens": 148985004.0, + "step": 3837 + }, + { + "epoch": 0.4882330492303778, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.898566722869873, + "learning_rate": 1e-06, + "loss": 0.3994, + "mean_token_accuracy": 0.865458607673645, + "num_tokens": 149021645.0, + "step": 3838 + }, + { + "epoch": 0.48836025950896833, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.881458282470703, + "learning_rate": 1e-06, + "loss": 0.4045, + "mean_token_accuracy": 0.8645467758178711, + "num_tokens": 149059663.0, + "step": 3839 + }, + { + "epoch": 0.48848746978755886, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.78460693359375, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.8733124732971191, + "num_tokens": 149104688.0, + "step": 3840 + }, + { + "epoch": 0.48861468006614933, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.820071220397949, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8797299861907959, + "num_tokens": 149143745.0, + "step": 3841 + }, + { + "epoch": 0.48874189034473986, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.8740692138671875, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.881394624710083, + "num_tokens": 149185027.0, + "step": 3842 + }, + { + "epoch": 0.4888691006233304, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.744882583618164, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8848699331283569, + "num_tokens": 149229124.0, + "step": 3843 + }, + { + "epoch": 0.48899631090192086, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.907103538513184, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8706201314926147, + "num_tokens": 149264712.0, + "step": 3844 + }, + { + "epoch": 0.4891235211805114, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.841466426849365, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8712267875671387, + "num_tokens": 149304876.0, + "step": 3845 + }, + { + "epoch": 0.4892507314591019, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.842203617095947, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.877792477607727, + "num_tokens": 149342715.0, + "step": 3846 + }, + { + "epoch": 0.4893779417376924, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.8644537925720215, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8746901750564575, + "num_tokens": 149383608.0, + "step": 3847 + }, + { + "epoch": 0.4895051520162829, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.765936851501465, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8827335834503174, + "num_tokens": 149423959.0, + "step": 3848 + }, + { + "epoch": 0.48963236229487345, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 8.04043197631836, + "learning_rate": 1e-06, + "loss": 0.3905, + "mean_token_accuracy": 0.867672324180603, + "num_tokens": 149459835.0, + "step": 3849 + }, + { + "epoch": 0.4897595725734639, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.831597328186035, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8814369440078735, + "num_tokens": 149501666.0, + "step": 3850 + }, + { + "epoch": 0.48988678285205445, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.813474655151367, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.8740593194961548, + "num_tokens": 149540689.0, + "step": 3851 + }, + { + "epoch": 0.490013993130645, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.996440410614014, + "learning_rate": 1e-06, + "loss": 0.4276, + "mean_token_accuracy": 0.8600472211837769, + "num_tokens": 149575786.0, + "step": 3852 + }, + { + "epoch": 0.49014120340923545, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.857974052429199, + "learning_rate": 1e-06, + "loss": 0.3797, + "mean_token_accuracy": 0.8699374198913574, + "num_tokens": 149613490.0, + "step": 3853 + }, + { + "epoch": 0.490268413687826, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.76834774017334, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8776004314422607, + "num_tokens": 149655531.0, + "step": 3854 + }, + { + "epoch": 0.4903956239664165, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.788976192474365, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8798565864562988, + "num_tokens": 149696131.0, + "step": 3855 + }, + { + "epoch": 0.490522834245007, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.89152193069458, + "learning_rate": 1e-06, + "loss": 0.3957, + "mean_token_accuracy": 0.8672481775283813, + "num_tokens": 149732197.0, + "step": 3856 + }, + { + "epoch": 0.4906500445235975, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.851617813110352, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8793071508407593, + "num_tokens": 149772181.0, + "step": 3857 + }, + { + "epoch": 0.49077725480218803, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.742526054382324, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8861421346664429, + "num_tokens": 149813298.0, + "step": 3858 + }, + { + "epoch": 0.4909044650807785, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.899716854095459, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8811169862747192, + "num_tokens": 149849260.0, + "step": 3859 + }, + { + "epoch": 0.49103167535936904, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.833908557891846, + "learning_rate": 1e-06, + "loss": 0.3817, + "mean_token_accuracy": 0.8702288269996643, + "num_tokens": 149894407.0, + "step": 3860 + }, + { + "epoch": 0.49115888563795956, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.803954601287842, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8772240877151489, + "num_tokens": 149938448.0, + "step": 3861 + }, + { + "epoch": 0.49128609591655004, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.9099225997924805, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8785667419433594, + "num_tokens": 149974933.0, + "step": 3862 + }, + { + "epoch": 0.49141330619514056, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.77164888381958, + "learning_rate": 1e-06, + "loss": 0.3847, + "mean_token_accuracy": 0.8696342706680298, + "num_tokens": 150017597.0, + "step": 3863 + }, + { + "epoch": 0.4915405164737311, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.919925212860107, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8799859285354614, + "num_tokens": 150054883.0, + "step": 3864 + }, + { + "epoch": 0.49166772675232157, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.89008092880249, + "learning_rate": 1e-06, + "loss": 0.3825, + "mean_token_accuracy": 0.8695945739746094, + "num_tokens": 150091965.0, + "step": 3865 + }, + { + "epoch": 0.4917949370309121, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.902195930480957, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8818660974502563, + "num_tokens": 150126796.0, + "step": 3866 + }, + { + "epoch": 0.4919221473095026, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.692960262298584, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8890438079833984, + "num_tokens": 150172048.0, + "step": 3867 + }, + { + "epoch": 0.4920493575880931, + "ewc_loss": 0.00970458984375, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.046627044677734e-06, + "grad_norm": 8.010385513305664, + "learning_rate": 1e-06, + "loss": 0.4717, + "mean_token_accuracy": 0.8452946543693542, + "num_tokens": 150210199.0, + "step": 3868 + }, + { + "epoch": 0.4921765678666836, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.817163944244385, + "learning_rate": 1e-06, + "loss": 0.4168, + "mean_token_accuracy": 0.8588004112243652, + "num_tokens": 150256837.0, + "step": 3869 + }, + { + "epoch": 0.49230377814527415, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.853621482849121, + "learning_rate": 1e-06, + "loss": 0.4097, + "mean_token_accuracy": 0.8633117079734802, + "num_tokens": 150302812.0, + "step": 3870 + }, + { + "epoch": 0.4924309884238646, + "ewc_loss": 0.009765625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.791232585906982, + "learning_rate": 1e-06, + "loss": 0.3911, + "mean_token_accuracy": 0.8675185441970825, + "num_tokens": 150347044.0, + "step": 3871 + }, + { + "epoch": 0.49255819870245515, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.966553211212158, + "learning_rate": 1e-06, + "loss": 0.4048, + "mean_token_accuracy": 0.8580816984176636, + "num_tokens": 150382237.0, + "step": 3872 + }, + { + "epoch": 0.4926854089810457, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.877044677734375, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8835732936859131, + "num_tokens": 150418523.0, + "step": 3873 + }, + { + "epoch": 0.49281261925963615, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.827887058258057, + "learning_rate": 1e-06, + "loss": 0.4181, + "mean_token_accuracy": 0.860823392868042, + "num_tokens": 150464851.0, + "step": 3874 + }, + { + "epoch": 0.4929398295382267, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.951188087463379, + "learning_rate": 1e-06, + "loss": 0.4227, + "mean_token_accuracy": 0.862717866897583, + "num_tokens": 150502809.0, + "step": 3875 + }, + { + "epoch": 0.4930670398168172, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6689300537109375e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.865843296051025, + "learning_rate": 1e-06, + "loss": 0.4424, + "mean_token_accuracy": 0.8505077958106995, + "num_tokens": 150545344.0, + "step": 3876 + }, + { + "epoch": 0.4931942500954077, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6808509826660156e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.852963924407959, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.8666586875915527, + "num_tokens": 150582752.0, + "step": 3877 + }, + { + "epoch": 0.4933214603739982, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6808509826660156e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.831144332885742, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8787177801132202, + "num_tokens": 150621881.0, + "step": 3878 + }, + { + "epoch": 0.49344867065258874, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6808509826660156e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.813370227813721, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.8771617412567139, + "num_tokens": 150663600.0, + "step": 3879 + }, + { + "epoch": 0.4935758809311792, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6808509826660156e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.948139190673828, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.8741326928138733, + "num_tokens": 150701981.0, + "step": 3880 + }, + { + "epoch": 0.49370309120976974, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6808509826660156e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.835176467895508, + "learning_rate": 1e-06, + "loss": 0.3906, + "mean_token_accuracy": 0.8692177534103394, + "num_tokens": 150742400.0, + "step": 3881 + }, + { + "epoch": 0.49383030148836027, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6808509826660156e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.820179462432861, + "learning_rate": 1e-06, + "loss": 0.3889, + "mean_token_accuracy": 0.8671077489852905, + "num_tokens": 150785865.0, + "step": 3882 + }, + { + "epoch": 0.4939575117669508, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6808509826660156e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.982841968536377, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8840188980102539, + "num_tokens": 150817205.0, + "step": 3883 + }, + { + "epoch": 0.49408472204554127, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6808509826660156e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.788744926452637, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8844022750854492, + "num_tokens": 150858166.0, + "step": 3884 + }, + { + "epoch": 0.4942119323241318, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6808509826660156e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.898491382598877, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8761646747589111, + "num_tokens": 150892867.0, + "step": 3885 + }, + { + "epoch": 0.4943391426027223, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6808509826660156e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.914451599121094, + "learning_rate": 1e-06, + "loss": 0.4006, + "mean_token_accuracy": 0.8684459924697876, + "num_tokens": 150931484.0, + "step": 3886 + }, + { + "epoch": 0.4944663528813128, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6808509826660156e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.851194858551025, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8805640935897827, + "num_tokens": 150966939.0, + "step": 3887 + }, + { + "epoch": 0.4945935631599033, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6808509826660156e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.939445495605469, + "learning_rate": 1e-06, + "loss": 0.3986, + "mean_token_accuracy": 0.8690675497055054, + "num_tokens": 151008775.0, + "step": 3888 + }, + { + "epoch": 0.49472077343849385, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.852702617645264, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8789460062980652, + "num_tokens": 151045505.0, + "step": 3889 + }, + { + "epoch": 0.4948479837170843, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.843477725982666, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8754608631134033, + "num_tokens": 151083113.0, + "step": 3890 + }, + { + "epoch": 0.49497519399567486, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.936801433563232, + "learning_rate": 1e-06, + "loss": 0.3811, + "mean_token_accuracy": 0.8722140789031982, + "num_tokens": 151118552.0, + "step": 3891 + }, + { + "epoch": 0.4951024042742654, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.807812690734863, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8767250776290894, + "num_tokens": 151158468.0, + "step": 3892 + }, + { + "epoch": 0.49522961455285586, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.8903889656066895, + "learning_rate": 1e-06, + "loss": 0.3908, + "mean_token_accuracy": 0.8657588362693787, + "num_tokens": 151196969.0, + "step": 3893 + }, + { + "epoch": 0.4953568248314464, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.832787036895752, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8708320260047913, + "num_tokens": 151231986.0, + "step": 3894 + }, + { + "epoch": 0.4954840351100369, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.80525016784668, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8765993714332581, + "num_tokens": 151273833.0, + "step": 3895 + }, + { + "epoch": 0.4956112453886274, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.908737659454346, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8760550618171692, + "num_tokens": 151313131.0, + "step": 3896 + }, + { + "epoch": 0.4957384556672179, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.9099836349487305, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8826215863227844, + "num_tokens": 151348958.0, + "step": 3897 + }, + { + "epoch": 0.49586566594580844, + "ewc_loss": 0.00982666015625, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.809015274047852, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8858962059020996, + "num_tokens": 151388273.0, + "step": 3898 + }, + { + "epoch": 0.4959928762243989, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.888484477996826, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8822822570800781, + "num_tokens": 151423994.0, + "step": 3899 + }, + { + "epoch": 0.49612008650298944, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.805272579193115, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8933233618736267, + "num_tokens": 151466799.0, + "step": 3900 + }, + { + "epoch": 0.49624729678157997, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.9373955726623535, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8811095952987671, + "num_tokens": 151500873.0, + "step": 3901 + }, + { + "epoch": 0.49637450706017044, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.937420845031738, + "learning_rate": 1e-06, + "loss": 0.4237, + "mean_token_accuracy": 0.858758807182312, + "num_tokens": 151537657.0, + "step": 3902 + }, + { + "epoch": 0.496501717338761, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.983493804931641, + "learning_rate": 1e-06, + "loss": 0.4028, + "mean_token_accuracy": 0.860692024230957, + "num_tokens": 151574187.0, + "step": 3903 + }, + { + "epoch": 0.4966289276173515, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.6927719116210938e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.881319046020508, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8799451589584351, + "num_tokens": 151608906.0, + "step": 3904 + }, + { + "epoch": 0.496756137895942, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.9286909103393555, + "learning_rate": 1e-06, + "loss": 0.3856, + "mean_token_accuracy": 0.8724642992019653, + "num_tokens": 151644957.0, + "step": 3905 + }, + { + "epoch": 0.4968833481745325, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.900528430938721, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8758935928344727, + "num_tokens": 151682007.0, + "step": 3906 + }, + { + "epoch": 0.49701055845312303, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.846250057220459, + "learning_rate": 1e-06, + "loss": 0.4289, + "mean_token_accuracy": 0.8597363233566284, + "num_tokens": 151728221.0, + "step": 3907 + }, + { + "epoch": 0.4971377687317135, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.95492696762085, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8809794187545776, + "num_tokens": 151763363.0, + "step": 3908 + }, + { + "epoch": 0.49726497901030403, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.878506660461426, + "learning_rate": 1e-06, + "loss": 0.4115, + "mean_token_accuracy": 0.862898588180542, + "num_tokens": 151806518.0, + "step": 3909 + }, + { + "epoch": 0.49739218928889456, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.929863929748535, + "learning_rate": 1e-06, + "loss": 0.3818, + "mean_token_accuracy": 0.8733532428741455, + "num_tokens": 151841307.0, + "step": 3910 + }, + { + "epoch": 0.49751939956748503, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.8362884521484375, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.8710294961929321, + "num_tokens": 151883195.0, + "step": 3911 + }, + { + "epoch": 0.49764660984607556, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.879907131195068, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8793580532073975, + "num_tokens": 151921413.0, + "step": 3912 + }, + { + "epoch": 0.4977738201246661, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.893296718597412, + "learning_rate": 1e-06, + "loss": 0.3786, + "mean_token_accuracy": 0.8708617091178894, + "num_tokens": 151960053.0, + "step": 3913 + }, + { + "epoch": 0.49790103040325656, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.934444427490234, + "learning_rate": 1e-06, + "loss": 0.4105, + "mean_token_accuracy": 0.8652291297912598, + "num_tokens": 152000644.0, + "step": 3914 + }, + { + "epoch": 0.4980282406818471, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.842277526855469, + "learning_rate": 1e-06, + "loss": 0.3941, + "mean_token_accuracy": 0.8671415448188782, + "num_tokens": 152042072.0, + "step": 3915 + }, + { + "epoch": 0.4981554509604376, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.911384582519531, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8715105056762695, + "num_tokens": 152080923.0, + "step": 3916 + }, + { + "epoch": 0.4982826612390281, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.846652030944824, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8864579796791077, + "num_tokens": 152122207.0, + "step": 3917 + }, + { + "epoch": 0.4984098715176186, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.882033348083496, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.8636407852172852, + "num_tokens": 152159441.0, + "step": 3918 + }, + { + "epoch": 0.49853708179620915, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.704692840576172e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.895277976989746, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.8692590594291687, + "num_tokens": 152200953.0, + "step": 3919 + }, + { + "epoch": 0.4986642920747996, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.926416397094727, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8764568567276001, + "num_tokens": 152238445.0, + "step": 3920 + }, + { + "epoch": 0.49879150235339015, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.832647323608398, + "learning_rate": 1e-06, + "loss": 0.3858, + "mean_token_accuracy": 0.8711866140365601, + "num_tokens": 152277354.0, + "step": 3921 + }, + { + "epoch": 0.4989187126319807, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.8821210861206055, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8841577172279358, + "num_tokens": 152315205.0, + "step": 3922 + }, + { + "epoch": 0.49904592291057115, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.93839168548584, + "learning_rate": 1e-06, + "loss": 0.3927, + "mean_token_accuracy": 0.8655439615249634, + "num_tokens": 152349912.0, + "step": 3923 + }, + { + "epoch": 0.4991731331891617, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.843560695648193, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8761035799980164, + "num_tokens": 152391028.0, + "step": 3924 + }, + { + "epoch": 0.4993003434677522, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 8.010604858398438, + "learning_rate": 1e-06, + "loss": 0.4619, + "mean_token_accuracy": 0.8437581062316895, + "num_tokens": 152428564.0, + "step": 3925 + }, + { + "epoch": 0.4994275537463427, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.896961212158203, + "learning_rate": 1e-06, + "loss": 0.407, + "mean_token_accuracy": 0.8655518889427185, + "num_tokens": 152472046.0, + "step": 3926 + }, + { + "epoch": 0.4995547640249332, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.827561855316162, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.8722407817840576, + "num_tokens": 152511255.0, + "step": 3927 + }, + { + "epoch": 0.49968197430352373, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.942287921905518, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.870941162109375, + "num_tokens": 152547156.0, + "step": 3928 + }, + { + "epoch": 0.4998091845821142, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.9100799560546875, + "learning_rate": 1e-06, + "loss": 0.4034, + "mean_token_accuracy": 0.8611984848976135, + "num_tokens": 152584625.0, + "step": 3929 + }, + { + "epoch": 0.49993639486070474, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.879663944244385, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.890768826007843, + "num_tokens": 152620709.0, + "step": 3930 + }, + { + "epoch": 0.5000636051392953, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.894796371459961, + "learning_rate": 1e-06, + "loss": 0.4323, + "mean_token_accuracy": 0.8573706746101379, + "num_tokens": 152662790.0, + "step": 3931 + }, + { + "epoch": 0.5001908154178858, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.8199687004089355, + "learning_rate": 1e-06, + "loss": 0.3754, + "mean_token_accuracy": 0.8734157085418701, + "num_tokens": 152704518.0, + "step": 3932 + }, + { + "epoch": 0.5003180256964763, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.8908371925354, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8730749487876892, + "num_tokens": 152745083.0, + "step": 3933 + }, + { + "epoch": 0.5004452359750667, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.794486999511719, + "learning_rate": 1e-06, + "loss": 0.4051, + "mean_token_accuracy": 0.8659729957580566, + "num_tokens": 152789228.0, + "step": 3934 + }, + { + "epoch": 0.5005724462536573, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.895627498626709, + "learning_rate": 1e-06, + "loss": 0.4249, + "mean_token_accuracy": 0.8598713874816895, + "num_tokens": 152836567.0, + "step": 3935 + }, + { + "epoch": 0.5006996565322478, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.899261474609375, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.8743789792060852, + "num_tokens": 152879214.0, + "step": 3936 + }, + { + "epoch": 0.5008268668108383, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.8063836097717285, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8884570598602295, + "num_tokens": 152919946.0, + "step": 3937 + }, + { + "epoch": 0.5009540770894289, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.880483627319336, + "learning_rate": 1e-06, + "loss": 0.381, + "mean_token_accuracy": 0.8707396984100342, + "num_tokens": 152964347.0, + "step": 3938 + }, + { + "epoch": 0.5010812873680194, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.9585394859313965, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8849478363990784, + "num_tokens": 153005473.0, + "step": 3939 + }, + { + "epoch": 0.5012084976466098, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.910190105438232, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8747601509094238, + "num_tokens": 153039349.0, + "step": 3940 + }, + { + "epoch": 0.5013357079252003, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.861363887786865, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.876878023147583, + "num_tokens": 153076569.0, + "step": 3941 + }, + { + "epoch": 0.5014629182037909, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.71661376953125e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.897364616394043, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8797093629837036, + "num_tokens": 153116498.0, + "step": 3942 + }, + { + "epoch": 0.5015901284823814, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.728534698486328e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.916262626647949, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8750389814376831, + "num_tokens": 153153812.0, + "step": 3943 + }, + { + "epoch": 0.5017173387609719, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.728534698486328e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.925369739532471, + "learning_rate": 1e-06, + "loss": 0.3965, + "mean_token_accuracy": 0.8665394186973572, + "num_tokens": 153190563.0, + "step": 3944 + }, + { + "epoch": 0.5018445490395624, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.728534698486328e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.879918098449707, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8756322860717773, + "num_tokens": 153227204.0, + "step": 3945 + }, + { + "epoch": 0.5019717593181529, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.728534698486328e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.906630516052246, + "learning_rate": 1e-06, + "loss": 0.4079, + "mean_token_accuracy": 0.860556960105896, + "num_tokens": 153266160.0, + "step": 3946 + }, + { + "epoch": 0.5020989695967434, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.728534698486328e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.994065284729004, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8824558258056641, + "num_tokens": 153295139.0, + "step": 3947 + }, + { + "epoch": 0.5022261798753339, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.728534698486328e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.8471598625183105, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8768296241760254, + "num_tokens": 153337734.0, + "step": 3948 + }, + { + "epoch": 0.5023533901539244, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.728534698486328e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.9151692390441895, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8792861700057983, + "num_tokens": 153373215.0, + "step": 3949 + }, + { + "epoch": 0.502480600432515, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.728534698486328e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.856927871704102, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8724005222320557, + "num_tokens": 153422199.0, + "step": 3950 + }, + { + "epoch": 0.5026078107111055, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.728534698486328e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.8661394119262695, + "learning_rate": 1e-06, + "loss": 0.3796, + "mean_token_accuracy": 0.8730559349060059, + "num_tokens": 153463299.0, + "step": 3951 + }, + { + "epoch": 0.5027350209896959, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.847273349761963, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8733986616134644, + "num_tokens": 153502397.0, + "step": 3952 + }, + { + "epoch": 0.5028622312682864, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.937478065490723, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.8710625171661377, + "num_tokens": 153539837.0, + "step": 3953 + }, + { + "epoch": 0.502989441546877, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.900931358337402, + "learning_rate": 1e-06, + "loss": 0.3798, + "mean_token_accuracy": 0.8712502717971802, + "num_tokens": 153576263.0, + "step": 3954 + }, + { + "epoch": 0.5031166518254675, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.910581111907959, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8722705841064453, + "num_tokens": 153612627.0, + "step": 3955 + }, + { + "epoch": 0.503243862104058, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.823662757873535, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8833719491958618, + "num_tokens": 153651073.0, + "step": 3956 + }, + { + "epoch": 0.5033710723826486, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.845746994018555, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8730847239494324, + "num_tokens": 153696246.0, + "step": 3957 + }, + { + "epoch": 0.5034982826612391, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.823506832122803, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8779246211051941, + "num_tokens": 153738030.0, + "step": 3958 + }, + { + "epoch": 0.5036254929398295, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 8.028593063354492, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8703545331954956, + "num_tokens": 153770457.0, + "step": 3959 + }, + { + "epoch": 0.50375270321842, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.951427936553955, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8787767887115479, + "num_tokens": 153801465.0, + "step": 3960 + }, + { + "epoch": 0.5038799134970106, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.915567398071289, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8715522289276123, + "num_tokens": 153840098.0, + "step": 3961 + }, + { + "epoch": 0.5040071237756011, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.951221942901611, + "learning_rate": 1e-06, + "loss": 0.4119, + "mean_token_accuracy": 0.8593966364860535, + "num_tokens": 153880575.0, + "step": 3962 + }, + { + "epoch": 0.5041343340541916, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.861497402191162, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.873869001865387, + "num_tokens": 153918585.0, + "step": 3963 + }, + { + "epoch": 0.5042615443327821, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.961981296539307, + "learning_rate": 1e-06, + "loss": 0.3773, + "mean_token_accuracy": 0.8721300363540649, + "num_tokens": 153954557.0, + "step": 3964 + }, + { + "epoch": 0.5043887546113726, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.87879753112793, + "learning_rate": 1e-06, + "loss": 0.4089, + "mean_token_accuracy": 0.8627548217773438, + "num_tokens": 153994751.0, + "step": 3965 + }, + { + "epoch": 0.5045159648899631, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.912840366363525, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8858134150505066, + "num_tokens": 154033301.0, + "step": 3966 + }, + { + "epoch": 0.5046431751685536, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.847228050231934, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.875964879989624, + "num_tokens": 154078855.0, + "step": 3967 + }, + { + "epoch": 0.5047703854471441, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.927968502044678, + "learning_rate": 1e-06, + "loss": 0.3924, + "mean_token_accuracy": 0.866695761680603, + "num_tokens": 154120518.0, + "step": 3968 + }, + { + "epoch": 0.5048975957257347, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.912692546844482, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8908714056015015, + "num_tokens": 154155685.0, + "step": 3969 + }, + { + "epoch": 0.5050248060043252, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.959989547729492, + "learning_rate": 1e-06, + "loss": 0.3971, + "mean_token_accuracy": 0.8674530386924744, + "num_tokens": 154193225.0, + "step": 3970 + }, + { + "epoch": 0.5051520162829156, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.850579261779785, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8771493434906006, + "num_tokens": 154232674.0, + "step": 3971 + }, + { + "epoch": 0.5052792265615061, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.839545726776123, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8734889030456543, + "num_tokens": 154274353.0, + "step": 3972 + }, + { + "epoch": 0.5054064368400967, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.888486862182617, + "learning_rate": 1e-06, + "loss": 0.386, + "mean_token_accuracy": 0.8706793785095215, + "num_tokens": 154314289.0, + "step": 3973 + }, + { + "epoch": 0.5055336471186872, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.954469680786133, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8849694132804871, + "num_tokens": 154346581.0, + "step": 3974 + }, + { + "epoch": 0.5056608573972777, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.936598777770996, + "learning_rate": 1e-06, + "loss": 0.4241, + "mean_token_accuracy": 0.8625056743621826, + "num_tokens": 154387471.0, + "step": 3975 + }, + { + "epoch": 0.5057880676758683, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.925629615783691, + "learning_rate": 1e-06, + "loss": 0.4153, + "mean_token_accuracy": 0.858857274055481, + "num_tokens": 154427284.0, + "step": 3976 + }, + { + "epoch": 0.5059152779544587, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 8.038278579711914, + "learning_rate": 1e-06, + "loss": 0.4319, + "mean_token_accuracy": 0.8550102114677429, + "num_tokens": 154462022.0, + "step": 3977 + }, + { + "epoch": 0.5060424882330492, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.9407429695129395, + "learning_rate": 1e-06, + "loss": 0.3986, + "mean_token_accuracy": 0.8683802485466003, + "num_tokens": 154500003.0, + "step": 3978 + }, + { + "epoch": 0.5061696985116397, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.847341537475586, + "learning_rate": 1e-06, + "loss": 0.3733, + "mean_token_accuracy": 0.8739409446716309, + "num_tokens": 154543174.0, + "step": 3979 + }, + { + "epoch": 0.5062969087902303, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.910292148590088, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8869597911834717, + "num_tokens": 154578390.0, + "step": 3980 + }, + { + "epoch": 0.5064241190688208, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.852205276489258, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8770275115966797, + "num_tokens": 154621768.0, + "step": 3981 + }, + { + "epoch": 0.5065513293474113, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.903055667877197, + "learning_rate": 1e-06, + "loss": 0.3861, + "mean_token_accuracy": 0.8698520660400391, + "num_tokens": 154658409.0, + "step": 3982 + }, + { + "epoch": 0.5066785396260017, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.83234167098999, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8702184557914734, + "num_tokens": 154701744.0, + "step": 3983 + }, + { + "epoch": 0.5068057499045923, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7404556274414062e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.927282333374023, + "learning_rate": 1e-06, + "loss": 0.3831, + "mean_token_accuracy": 0.867257297039032, + "num_tokens": 154739399.0, + "step": 3984 + }, + { + "epoch": 0.5069329601831828, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.845339298248291, + "learning_rate": 1e-06, + "loss": 0.3953, + "mean_token_accuracy": 0.8682912588119507, + "num_tokens": 154782922.0, + "step": 3985 + }, + { + "epoch": 0.5070601704617733, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.939332962036133, + "learning_rate": 1e-06, + "loss": 0.3783, + "mean_token_accuracy": 0.8755950927734375, + "num_tokens": 154819070.0, + "step": 3986 + }, + { + "epoch": 0.5071873807403638, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.936301231384277, + "learning_rate": 1e-06, + "loss": 0.4206, + "mean_token_accuracy": 0.8588190078735352, + "num_tokens": 154857489.0, + "step": 3987 + }, + { + "epoch": 0.5073145910189544, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.897846221923828, + "learning_rate": 1e-06, + "loss": 0.3929, + "mean_token_accuracy": 0.8686015605926514, + "num_tokens": 154896590.0, + "step": 3988 + }, + { + "epoch": 0.5074418012975448, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.836702346801758, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8773889541625977, + "num_tokens": 154939745.0, + "step": 3989 + }, + { + "epoch": 0.5075690115761353, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.868020057678223, + "learning_rate": 1e-06, + "loss": 0.3961, + "mean_token_accuracy": 0.8693302869796753, + "num_tokens": 154979374.0, + "step": 3990 + }, + { + "epoch": 0.5076962218547258, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.881648540496826, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8992390036582947, + "num_tokens": 155012685.0, + "step": 3991 + }, + { + "epoch": 0.5078234321333164, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.878387451171875, + "learning_rate": 1e-06, + "loss": 0.4192, + "mean_token_accuracy": 0.8579052090644836, + "num_tokens": 155057585.0, + "step": 3992 + }, + { + "epoch": 0.5079506424119069, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.903271675109863, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8792255520820618, + "num_tokens": 155093310.0, + "step": 3993 + }, + { + "epoch": 0.5080778526904974, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 8.071409225463867, + "learning_rate": 1e-06, + "loss": 0.4103, + "mean_token_accuracy": 0.8590829372406006, + "num_tokens": 155125893.0, + "step": 3994 + }, + { + "epoch": 0.5082050629690879, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.907412052154541, + "learning_rate": 1e-06, + "loss": 0.4014, + "mean_token_accuracy": 0.8663485646247864, + "num_tokens": 155163029.0, + "step": 3995 + }, + { + "epoch": 0.5083322732476784, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.710108280181885, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.894461452960968, + "num_tokens": 155205348.0, + "step": 3996 + }, + { + "epoch": 0.5084594835262689, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.857125759124756, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.873778223991394, + "num_tokens": 155245589.0, + "step": 3997 + }, + { + "epoch": 0.5085866938048594, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.8370490074157715, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8768002390861511, + "num_tokens": 155286350.0, + "step": 3998 + }, + { + "epoch": 0.50871390408345, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.776516914367676, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8848958015441895, + "num_tokens": 155331723.0, + "step": 3999 + }, + { + "epoch": 0.5088411143620405, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 8.028801918029785, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8713167309761047, + "num_tokens": 155365161.0, + "step": 4000 + }, + { + "epoch": 0.5089683246406309, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.8774213790893555, + "learning_rate": 1e-06, + "loss": 0.3838, + "mean_token_accuracy": 0.8694139122962952, + "num_tokens": 155403516.0, + "step": 4001 + }, + { + "epoch": 0.5090955349192214, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.904745578765869, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.8732450604438782, + "num_tokens": 155442698.0, + "step": 4002 + }, + { + "epoch": 0.509222745197812, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.908270359039307, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8779469728469849, + "num_tokens": 155480165.0, + "step": 4003 + }, + { + "epoch": 0.5093499554764025, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.86643648147583, + "learning_rate": 1e-06, + "loss": 0.4028, + "mean_token_accuracy": 0.8599655628204346, + "num_tokens": 155522731.0, + "step": 4004 + }, + { + "epoch": 0.509477165754993, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.817955493927002, + "learning_rate": 1e-06, + "loss": 0.4029, + "mean_token_accuracy": 0.8652896881103516, + "num_tokens": 155569341.0, + "step": 4005 + }, + { + "epoch": 0.5096043760335836, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.7523765563964844e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.878958702087402, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.871271014213562, + "num_tokens": 155610977.0, + "step": 4006 + }, + { + "epoch": 0.5097315863121741, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7642974853515625e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.771445274353027, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8688130974769592, + "num_tokens": 155655390.0, + "step": 4007 + }, + { + "epoch": 0.5098587965907645, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.7642974853515625e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.916637420654297, + "learning_rate": 1e-06, + "loss": 0.3664, + "mean_token_accuracy": 0.87396240234375, + "num_tokens": 155692105.0, + "step": 4008 + }, + { + "epoch": 0.509986006869355, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.7642974853515625e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.832491874694824, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.8656115531921387, + "num_tokens": 155734579.0, + "step": 4009 + }, + { + "epoch": 0.5101132171479456, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.7642974853515625e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.889094352722168, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8853769302368164, + "num_tokens": 155773096.0, + "step": 4010 + }, + { + "epoch": 0.5102404274265361, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.7642974853515625e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.870276927947998, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8792448043823242, + "num_tokens": 155809156.0, + "step": 4011 + }, + { + "epoch": 0.5103676377051266, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.7642974853515625e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.82016658782959, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.8705867528915405, + "num_tokens": 155851294.0, + "step": 4012 + }, + { + "epoch": 0.5104948479837171, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.7642974853515625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.875829696655273, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8868447542190552, + "num_tokens": 155893156.0, + "step": 4013 + }, + { + "epoch": 0.5106220582623076, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.7642974853515625e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.851194858551025, + "learning_rate": 1e-06, + "loss": 0.3923, + "mean_token_accuracy": 0.8690229058265686, + "num_tokens": 155937335.0, + "step": 4014 + }, + { + "epoch": 0.5107492685408981, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.7642974853515625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.762225151062012, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8925908803939819, + "num_tokens": 155978891.0, + "step": 4015 + }, + { + "epoch": 0.5108764788194886, + "ewc_loss": 0.00994873046875, + "ewc_loss_diag": 1.7642974853515625e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.991817951202393, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8732711672782898, + "num_tokens": 156011228.0, + "step": 4016 + }, + { + "epoch": 0.5110036890980791, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.899865627288818, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8734217882156372, + "num_tokens": 156051648.0, + "step": 4017 + }, + { + "epoch": 0.5111308993766697, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.910473823547363, + "learning_rate": 1e-06, + "loss": 0.4079, + "mean_token_accuracy": 0.8658499717712402, + "num_tokens": 156087609.0, + "step": 4018 + }, + { + "epoch": 0.5112581096552602, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.954920768737793, + "learning_rate": 1e-06, + "loss": 0.4073, + "mean_token_accuracy": 0.8621700406074524, + "num_tokens": 156127344.0, + "step": 4019 + }, + { + "epoch": 0.5113853199338506, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.867379665374756, + "learning_rate": 1e-06, + "loss": 0.3826, + "mean_token_accuracy": 0.8699616193771362, + "num_tokens": 156166745.0, + "step": 4020 + }, + { + "epoch": 0.5115125302124411, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.93113899230957, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.87930828332901, + "num_tokens": 156202543.0, + "step": 4021 + }, + { + "epoch": 0.5116397404910317, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.90582799911499, + "learning_rate": 1e-06, + "loss": 0.4197, + "mean_token_accuracy": 0.8610920310020447, + "num_tokens": 156243167.0, + "step": 4022 + }, + { + "epoch": 0.5117669507696222, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 8.032796859741211, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.891840398311615, + "num_tokens": 156285984.0, + "step": 4023 + }, + { + "epoch": 0.5118941610482127, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.8845038414001465, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8818293809890747, + "num_tokens": 156323871.0, + "step": 4024 + }, + { + "epoch": 0.5120213713268033, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.92616605758667, + "learning_rate": 1e-06, + "loss": 0.3942, + "mean_token_accuracy": 0.8699100017547607, + "num_tokens": 156364496.0, + "step": 4025 + }, + { + "epoch": 0.5121485816053937, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.8957600593566895, + "learning_rate": 1e-06, + "loss": 0.3938, + "mean_token_accuracy": 0.8687349557876587, + "num_tokens": 156404605.0, + "step": 4026 + }, + { + "epoch": 0.5122757918839842, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.996415615081787, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8813982009887695, + "num_tokens": 156437659.0, + "step": 4027 + }, + { + "epoch": 0.5124030021625747, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.988773822784424, + "learning_rate": 1e-06, + "loss": 0.4249, + "mean_token_accuracy": 0.8617660999298096, + "num_tokens": 156472735.0, + "step": 4028 + }, + { + "epoch": 0.5125302124411653, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.755338668823242, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8825185298919678, + "num_tokens": 156518924.0, + "step": 4029 + }, + { + "epoch": 0.5126574227197558, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.7762184143066406e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.905786514282227, + "learning_rate": 1e-06, + "loss": 0.4039, + "mean_token_accuracy": 0.8666691184043884, + "num_tokens": 156556239.0, + "step": 4030 + }, + { + "epoch": 0.5127846329983463, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.860477924346924, + "learning_rate": 1e-06, + "loss": 0.3475, + "mean_token_accuracy": 0.8816415071487427, + "num_tokens": 156597571.0, + "step": 4031 + }, + { + "epoch": 0.5129118432769367, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.900930881500244, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.879148542881012, + "num_tokens": 156634264.0, + "step": 4032 + }, + { + "epoch": 0.5130390535555273, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.985347747802734, + "learning_rate": 1e-06, + "loss": 0.4228, + "mean_token_accuracy": 0.8585972189903259, + "num_tokens": 156675901.0, + "step": 4033 + }, + { + "epoch": 0.5131662638341178, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.788145065307617, + "learning_rate": 1e-06, + "loss": 0.3842, + "mean_token_accuracy": 0.8702521920204163, + "num_tokens": 156721400.0, + "step": 4034 + }, + { + "epoch": 0.5132934741127083, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.869145393371582, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8841330409049988, + "num_tokens": 156759769.0, + "step": 4035 + }, + { + "epoch": 0.5134206843912988, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.92143440246582, + "learning_rate": 1e-06, + "loss": 0.4072, + "mean_token_accuracy": 0.8628142476081848, + "num_tokens": 156799841.0, + "step": 4036 + }, + { + "epoch": 0.5135478946698894, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.85033655166626, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8767632842063904, + "num_tokens": 156841368.0, + "step": 4037 + }, + { + "epoch": 0.5136751049484798, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 8.073702812194824, + "learning_rate": 1e-06, + "loss": 0.3846, + "mean_token_accuracy": 0.8710259795188904, + "num_tokens": 156872918.0, + "step": 4038 + }, + { + "epoch": 0.5138023152270703, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.851773738861084, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8868141174316406, + "num_tokens": 156909830.0, + "step": 4039 + }, + { + "epoch": 0.5139295255056608, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.830951690673828, + "learning_rate": 1e-06, + "loss": 0.3861, + "mean_token_accuracy": 0.8709262609481812, + "num_tokens": 156956990.0, + "step": 4040 + }, + { + "epoch": 0.5140567357842514, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.906090259552002, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8783067464828491, + "num_tokens": 156995647.0, + "step": 4041 + }, + { + "epoch": 0.5141839460628419, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.840491771697998, + "learning_rate": 1e-06, + "loss": 0.39, + "mean_token_accuracy": 0.8695947527885437, + "num_tokens": 157040469.0, + "step": 4042 + }, + { + "epoch": 0.5143111563414324, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.97008752822876, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8747113943099976, + "num_tokens": 157075814.0, + "step": 4043 + }, + { + "epoch": 0.5144383666200228, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.981709003448486, + "learning_rate": 1e-06, + "loss": 0.4176, + "mean_token_accuracy": 0.861865758895874, + "num_tokens": 157114473.0, + "step": 4044 + }, + { + "epoch": 0.5145655768986134, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.965925216674805, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8845254182815552, + "num_tokens": 157153874.0, + "step": 4045 + }, + { + "epoch": 0.5146927871772039, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 8.144521713256836, + "learning_rate": 1e-06, + "loss": 0.3959, + "mean_token_accuracy": 0.8657835125923157, + "num_tokens": 157182539.0, + "step": 4046 + }, + { + "epoch": 0.5148199974557944, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.898153781890869, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8822305798530579, + "num_tokens": 157222011.0, + "step": 4047 + }, + { + "epoch": 0.514947207734385, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.912272930145264, + "learning_rate": 1e-06, + "loss": 0.4209, + "mean_token_accuracy": 0.8637629747390747, + "num_tokens": 157262676.0, + "step": 4048 + }, + { + "epoch": 0.5150744180129755, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.856105327606201, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8863111734390259, + "num_tokens": 157305078.0, + "step": 4049 + }, + { + "epoch": 0.5152016282915659, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.7881393432617188e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.791266441345215, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8751864433288574, + "num_tokens": 157349102.0, + "step": 4050 + }, + { + "epoch": 0.5153288385701564, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 8.106821060180664, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8847488760948181, + "num_tokens": 157378948.0, + "step": 4051 + }, + { + "epoch": 0.515456048848747, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.974524021148682, + "learning_rate": 1e-06, + "loss": 0.4736, + "mean_token_accuracy": 0.8406984210014343, + "num_tokens": 157422368.0, + "step": 4052 + }, + { + "epoch": 0.5155832591273375, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.76185417175293, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8829324841499329, + "num_tokens": 157464617.0, + "step": 4053 + }, + { + "epoch": 0.515710469405928, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.864654064178467, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8728572130203247, + "num_tokens": 157505406.0, + "step": 4054 + }, + { + "epoch": 0.5158376796845185, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.859581470489502, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8718066215515137, + "num_tokens": 157551146.0, + "step": 4055 + }, + { + "epoch": 0.5159648899631091, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.9427008628845215, + "learning_rate": 1e-06, + "loss": 0.4443, + "mean_token_accuracy": 0.8508625626564026, + "num_tokens": 157591322.0, + "step": 4056 + }, + { + "epoch": 0.5160921002416995, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.9118804931640625, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.874752402305603, + "num_tokens": 157637886.0, + "step": 4057 + }, + { + "epoch": 0.51621931052029, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.9025750160217285, + "learning_rate": 1e-06, + "loss": 0.409, + "mean_token_accuracy": 0.8653824329376221, + "num_tokens": 157680532.0, + "step": 4058 + }, + { + "epoch": 0.5163465207988805, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.934493064880371, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8821735382080078, + "num_tokens": 157716913.0, + "step": 4059 + }, + { + "epoch": 0.5164737310774711, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.844357013702393, + "learning_rate": 1e-06, + "loss": 0.4098, + "mean_token_accuracy": 0.8626235723495483, + "num_tokens": 157766109.0, + "step": 4060 + }, + { + "epoch": 0.5166009413560616, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.925241947174072, + "learning_rate": 1e-06, + "loss": 0.3898, + "mean_token_accuracy": 0.8676513433456421, + "num_tokens": 157805829.0, + "step": 4061 + }, + { + "epoch": 0.5167281516346521, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.87159538269043, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.8713475465774536, + "num_tokens": 157846937.0, + "step": 4062 + }, + { + "epoch": 0.5168553619132426, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.928558826446533, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8932997584342957, + "num_tokens": 157883133.0, + "step": 4063 + }, + { + "epoch": 0.5169825721918331, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.974048614501953, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8824293613433838, + "num_tokens": 157924772.0, + "step": 4064 + }, + { + "epoch": 0.5171097824704236, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.803936958312988, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.8746896386146545, + "num_tokens": 157972110.0, + "step": 4065 + }, + { + "epoch": 0.5172369927490141, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.932721138000488, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8827910423278809, + "num_tokens": 158009534.0, + "step": 4066 + }, + { + "epoch": 0.5173642030276047, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 8.008095741271973, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8828902840614319, + "num_tokens": 158040637.0, + "step": 4067 + }, + { + "epoch": 0.5174914133061952, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.976451873779297, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8665756583213806, + "num_tokens": 158075983.0, + "step": 4068 + }, + { + "epoch": 0.5176186235847856, + "ewc_loss": 0.0098876953125, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.106231689453125e-06, + "grad_norm": 7.879358291625977, + "learning_rate": 1e-06, + "loss": 0.3676, + "mean_token_accuracy": 0.8760083317756653, + "num_tokens": 158116100.0, + "step": 4069 + }, + { + "epoch": 0.5177458338633761, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.165836334228516e-06, + "grad_norm": 7.907617568969727, + "learning_rate": 1e-06, + "loss": 0.4389, + "mean_token_accuracy": 0.8543530106544495, + "num_tokens": 158162419.0, + "step": 4070 + }, + { + "epoch": 0.5178730441419667, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.867409706115723, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.8642542362213135, + "num_tokens": 158204083.0, + "step": 4071 + }, + { + "epoch": 0.5180002544205572, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.90071439743042, + "learning_rate": 1e-06, + "loss": 0.389, + "mean_token_accuracy": 0.8678843975067139, + "num_tokens": 158245645.0, + "step": 4072 + }, + { + "epoch": 0.5181274646991477, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.833442687988281, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8918313980102539, + "num_tokens": 158284148.0, + "step": 4073 + }, + { + "epoch": 0.5182546749777382, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.8511786460876465, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8806345462799072, + "num_tokens": 158325012.0, + "step": 4074 + }, + { + "epoch": 0.5183818852563287, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.9518022537231445, + "learning_rate": 1e-06, + "loss": 0.3714, + "mean_token_accuracy": 0.8752799034118652, + "num_tokens": 158359674.0, + "step": 4075 + }, + { + "epoch": 0.5185090955349192, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.893442630767822, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.8725161552429199, + "num_tokens": 158402049.0, + "step": 4076 + }, + { + "epoch": 0.5186363058135097, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.850276947021484, + "learning_rate": 1e-06, + "loss": 0.4118, + "mean_token_accuracy": 0.8631254434585571, + "num_tokens": 158448540.0, + "step": 4077 + }, + { + "epoch": 0.5187635160921003, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.908388614654541, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8770443201065063, + "num_tokens": 158486117.0, + "step": 4078 + }, + { + "epoch": 0.5188907263706908, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.878482341766357, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8721966743469238, + "num_tokens": 158527618.0, + "step": 4079 + }, + { + "epoch": 0.5190179366492813, + "ewc_loss": 0.010009765625, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.834333419799805, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8736076354980469, + "num_tokens": 158575259.0, + "step": 4080 + }, + { + "epoch": 0.5191451469278717, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.800060272216797e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.957754611968994, + "learning_rate": 1e-06, + "loss": 0.3967, + "mean_token_accuracy": 0.8643032312393188, + "num_tokens": 158612179.0, + "step": 4081 + }, + { + "epoch": 0.5192723572064623, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.90104866027832, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.8717813491821289, + "num_tokens": 158652532.0, + "step": 4082 + }, + { + "epoch": 0.5193995674850528, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.90566873550415, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.879812479019165, + "num_tokens": 158690688.0, + "step": 4083 + }, + { + "epoch": 0.5195267777636433, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.966052532196045, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.8740420341491699, + "num_tokens": 158725430.0, + "step": 4084 + }, + { + "epoch": 0.5196539880422338, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.9006171226501465, + "learning_rate": 1e-06, + "loss": 0.3875, + "mean_token_accuracy": 0.8702090382575989, + "num_tokens": 158766302.0, + "step": 4085 + }, + { + "epoch": 0.5197811983208244, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.917028903961182, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.8712657690048218, + "num_tokens": 158807156.0, + "step": 4086 + }, + { + "epoch": 0.5199084085994148, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.809863090515137, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8786565065383911, + "num_tokens": 158848448.0, + "step": 4087 + }, + { + "epoch": 0.5200356188780053, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.924058437347412, + "learning_rate": 1e-06, + "loss": 0.4006, + "mean_token_accuracy": 0.8665092587471008, + "num_tokens": 158885356.0, + "step": 4088 + }, + { + "epoch": 0.5201628291565958, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.939083099365234, + "learning_rate": 1e-06, + "loss": 0.3904, + "mean_token_accuracy": 0.8677502870559692, + "num_tokens": 158923187.0, + "step": 4089 + }, + { + "epoch": 0.5202900394351864, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.787352085113525, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.877663791179657, + "num_tokens": 158968177.0, + "step": 4090 + }, + { + "epoch": 0.5204172497137769, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.905725955963135, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8819196224212646, + "num_tokens": 159007696.0, + "step": 4091 + }, + { + "epoch": 0.5205444599923674, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.921285152435303, + "learning_rate": 1e-06, + "loss": 0.4118, + "mean_token_accuracy": 0.8642702698707581, + "num_tokens": 159047486.0, + "step": 4092 + }, + { + "epoch": 0.5206716702709578, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.993302822113037, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8775961399078369, + "num_tokens": 159080164.0, + "step": 4093 + }, + { + "epoch": 0.5207988805495484, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 8.02187442779541, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8784900903701782, + "num_tokens": 159111758.0, + "step": 4094 + }, + { + "epoch": 0.5209260908281389, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.853449821472168, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8751129508018494, + "num_tokens": 159151146.0, + "step": 4095 + }, + { + "epoch": 0.5210533011067294, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.849667549133301, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8711138367652893, + "num_tokens": 159190667.0, + "step": 4096 + }, + { + "epoch": 0.52118051138532, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.911076545715332, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8825188279151917, + "num_tokens": 159222921.0, + "step": 4097 + }, + { + "epoch": 0.5213077216639105, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 8.009489059448242, + "learning_rate": 1e-06, + "loss": 0.4241, + "mean_token_accuracy": 0.8599121570587158, + "num_tokens": 159264122.0, + "step": 4098 + }, + { + "epoch": 0.5214349319425009, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.994657039642334, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8759372234344482, + "num_tokens": 159299540.0, + "step": 4099 + }, + { + "epoch": 0.5215621422210914, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.879947662353516, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8741379976272583, + "num_tokens": 159339330.0, + "step": 4100 + }, + { + "epoch": 0.521689352499682, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.803623199462891, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.873504102230072, + "num_tokens": 159384383.0, + "step": 4101 + }, + { + "epoch": 0.5218165627782725, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.984990119934082, + "learning_rate": 1e-06, + "loss": 0.414, + "mean_token_accuracy": 0.8600583672523499, + "num_tokens": 159422335.0, + "step": 4102 + }, + { + "epoch": 0.521943773056863, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.823902130126953e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.82404899597168, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8877787590026855, + "num_tokens": 159465079.0, + "step": 4103 + }, + { + "epoch": 0.5220709833354535, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.937546253204346, + "learning_rate": 1e-06, + "loss": 0.3916, + "mean_token_accuracy": 0.8715090751647949, + "num_tokens": 159506959.0, + "step": 4104 + }, + { + "epoch": 0.522198193614044, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.975707054138184, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8733454942703247, + "num_tokens": 159544569.0, + "step": 4105 + }, + { + "epoch": 0.5223254038926345, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.884629726409912, + "learning_rate": 1e-06, + "loss": 0.3844, + "mean_token_accuracy": 0.8714506030082703, + "num_tokens": 159583182.0, + "step": 4106 + }, + { + "epoch": 0.522452614171225, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.943022727966309, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8785098195075989, + "num_tokens": 159622138.0, + "step": 4107 + }, + { + "epoch": 0.5225798244498155, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.952621936798096, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.878922164440155, + "num_tokens": 159664237.0, + "step": 4108 + }, + { + "epoch": 0.5227070347284061, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.87098503112793, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8792245984077454, + "num_tokens": 159705139.0, + "step": 4109 + }, + { + "epoch": 0.5228342450069966, + "ewc_loss": 0.01007080078125, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.225440979003906e-06, + "grad_norm": 7.85249137878418, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8783559799194336, + "num_tokens": 159746672.0, + "step": 4110 + }, + { + "epoch": 0.5229614552855871, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 8.001874923706055, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8837506771087646, + "num_tokens": 159779185.0, + "step": 4111 + }, + { + "epoch": 0.5230886655641775, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.93358850479126, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8733546137809753, + "num_tokens": 159814956.0, + "step": 4112 + }, + { + "epoch": 0.5232158758427681, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.943631649017334, + "learning_rate": 1e-06, + "loss": 0.4437, + "mean_token_accuracy": 0.8525168299674988, + "num_tokens": 159861118.0, + "step": 4113 + }, + { + "epoch": 0.5233430861213586, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.935271739959717, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.8715526461601257, + "num_tokens": 159898959.0, + "step": 4114 + }, + { + "epoch": 0.5234702963999491, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.898313045501709, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8757095336914062, + "num_tokens": 159937078.0, + "step": 4115 + }, + { + "epoch": 0.5235975066785397, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.917563438415527, + "learning_rate": 1e-06, + "loss": 0.3797, + "mean_token_accuracy": 0.8730735778808594, + "num_tokens": 159977770.0, + "step": 4116 + }, + { + "epoch": 0.5237247169571302, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.805735111236572, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8697487115859985, + "num_tokens": 160027918.0, + "step": 4117 + }, + { + "epoch": 0.5238519272357206, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.964290142059326, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8794896006584167, + "num_tokens": 160062642.0, + "step": 4118 + }, + { + "epoch": 0.5239791375143111, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.913790702819824, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.881523847579956, + "num_tokens": 160102605.0, + "step": 4119 + }, + { + "epoch": 0.5241063477929017, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.941470623016357, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8718408346176147, + "num_tokens": 160139683.0, + "step": 4120 + }, + { + "epoch": 0.5242335580714922, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.907285213470459, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8783184289932251, + "num_tokens": 160177857.0, + "step": 4121 + }, + { + "epoch": 0.5243607683500827, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.933042526245117, + "learning_rate": 1e-06, + "loss": 0.3768, + "mean_token_accuracy": 0.8716159462928772, + "num_tokens": 160216954.0, + "step": 4122 + }, + { + "epoch": 0.5244879786286732, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8358230590820312e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.9901533126831055, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8721314668655396, + "num_tokens": 160252098.0, + "step": 4123 + }, + { + "epoch": 0.5246151889072637, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.860716342926025, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8818862438201904, + "num_tokens": 160291351.0, + "step": 4124 + }, + { + "epoch": 0.5247423991858542, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 8.020819664001465, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.873843789100647, + "num_tokens": 160323949.0, + "step": 4125 + }, + { + "epoch": 0.5248696094644447, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.93247127532959, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8715451955795288, + "num_tokens": 160363309.0, + "step": 4126 + }, + { + "epoch": 0.5249968197430352, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.834902763366699, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8805623054504395, + "num_tokens": 160406280.0, + "step": 4127 + }, + { + "epoch": 0.5251240300216258, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.961194038391113, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8855363130569458, + "num_tokens": 160440950.0, + "step": 4128 + }, + { + "epoch": 0.5252512403002163, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.879352569580078, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8815584182739258, + "num_tokens": 160483389.0, + "step": 4129 + }, + { + "epoch": 0.5253784505788067, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.995350360870361, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8693848848342896, + "num_tokens": 160516350.0, + "step": 4130 + }, + { + "epoch": 0.5255056608573972, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.884698390960693, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8829736113548279, + "num_tokens": 160553099.0, + "step": 4131 + }, + { + "epoch": 0.5256328711359878, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.924645900726318, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.8749288320541382, + "num_tokens": 160590960.0, + "step": 4132 + }, + { + "epoch": 0.5257600814145783, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.879568576812744, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8821374773979187, + "num_tokens": 160630441.0, + "step": 4133 + }, + { + "epoch": 0.5258872916931688, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.907947540283203, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8793071508407593, + "num_tokens": 160665839.0, + "step": 4134 + }, + { + "epoch": 0.5260145019717594, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.914802074432373, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8721222877502441, + "num_tokens": 160707265.0, + "step": 4135 + }, + { + "epoch": 0.5261417122503498, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.954006671905518, + "learning_rate": 1e-06, + "loss": 0.3878, + "mean_token_accuracy": 0.8684048652648926, + "num_tokens": 160748044.0, + "step": 4136 + }, + { + "epoch": 0.5262689225289403, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.985615253448486, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8658909201622009, + "num_tokens": 160785726.0, + "step": 4137 + }, + { + "epoch": 0.5263961328075308, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.848602771759033, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8825827836990356, + "num_tokens": 160828706.0, + "step": 4138 + }, + { + "epoch": 0.5265233430861214, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.921546936035156, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.8704239130020142, + "num_tokens": 160867599.0, + "step": 4139 + }, + { + "epoch": 0.5266505533647119, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.9801106452941895, + "learning_rate": 1e-06, + "loss": 0.4048, + "mean_token_accuracy": 0.8639625310897827, + "num_tokens": 160906299.0, + "step": 4140 + }, + { + "epoch": 0.5267777636433024, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.949588298797607, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.879877507686615, + "num_tokens": 160946742.0, + "step": 4141 + }, + { + "epoch": 0.5269049739218928, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8477439880371094e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.8851189613342285, + "learning_rate": 1e-06, + "loss": 0.4091, + "mean_token_accuracy": 0.8637915253639221, + "num_tokens": 160989780.0, + "step": 4142 + }, + { + "epoch": 0.5270321842004834, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.926103591918945, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8772339820861816, + "num_tokens": 161026178.0, + "step": 4143 + }, + { + "epoch": 0.5271593944790739, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.95975923538208, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8861439824104309, + "num_tokens": 161059322.0, + "step": 4144 + }, + { + "epoch": 0.5272866047576644, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.970520496368408, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8830435276031494, + "num_tokens": 161090634.0, + "step": 4145 + }, + { + "epoch": 0.527413815036255, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.773010730743408, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.886122465133667, + "num_tokens": 161136088.0, + "step": 4146 + }, + { + "epoch": 0.5275410253148455, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.860642433166504, + "learning_rate": 1e-06, + "loss": 0.3744, + "mean_token_accuracy": 0.8736723065376282, + "num_tokens": 161183652.0, + "step": 4147 + }, + { + "epoch": 0.5276682355934359, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.852793216705322, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8845475316047668, + "num_tokens": 161225664.0, + "step": 4148 + }, + { + "epoch": 0.5277954458720264, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.894989013671875, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.885981559753418, + "num_tokens": 161264662.0, + "step": 4149 + }, + { + "epoch": 0.527922656150617, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.904977798461914, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8772894144058228, + "num_tokens": 161305586.0, + "step": 4150 + }, + { + "epoch": 0.5280498664292075, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.913252353668213, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8782514333724976, + "num_tokens": 161342612.0, + "step": 4151 + }, + { + "epoch": 0.528177076707798, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.83336877822876, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8820360898971558, + "num_tokens": 161393362.0, + "step": 4152 + }, + { + "epoch": 0.5283042869863885, + "ewc_loss": 0.0101318359375, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.909450531005859, + "learning_rate": 1e-06, + "loss": 0.4083, + "mean_token_accuracy": 0.8641026616096497, + "num_tokens": 161431805.0, + "step": 4153 + }, + { + "epoch": 0.528431497264979, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.9811110496521, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8730255365371704, + "num_tokens": 161472875.0, + "step": 4154 + }, + { + "epoch": 0.5285587075435695, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.935280799865723, + "learning_rate": 1e-06, + "loss": 0.3956, + "mean_token_accuracy": 0.8615399599075317, + "num_tokens": 161508932.0, + "step": 4155 + }, + { + "epoch": 0.52868591782216, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 8.02115535736084, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8748742341995239, + "num_tokens": 161547793.0, + "step": 4156 + }, + { + "epoch": 0.5288131281007505, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.969189643859863, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8714428544044495, + "num_tokens": 161589050.0, + "step": 4157 + }, + { + "epoch": 0.5289403383793411, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 8.002669334411621, + "learning_rate": 1e-06, + "loss": 0.4367, + "mean_token_accuracy": 0.8559975624084473, + "num_tokens": 161627950.0, + "step": 4158 + }, + { + "epoch": 0.5290675486579316, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8596649169921875e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.939815521240234, + "learning_rate": 1e-06, + "loss": 0.3954, + "mean_token_accuracy": 0.8657715916633606, + "num_tokens": 161667878.0, + "step": 4159 + }, + { + "epoch": 0.5291947589365221, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 8.032686233520508, + "learning_rate": 1e-06, + "loss": 0.3959, + "mean_token_accuracy": 0.869148850440979, + "num_tokens": 161703776.0, + "step": 4160 + }, + { + "epoch": 0.5293219692151125, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.852097034454346, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8813810348510742, + "num_tokens": 161747437.0, + "step": 4161 + }, + { + "epoch": 0.5294491794937031, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.9516377449035645, + "learning_rate": 1e-06, + "loss": 0.3724, + "mean_token_accuracy": 0.8723440170288086, + "num_tokens": 161787422.0, + "step": 4162 + }, + { + "epoch": 0.5295763897722936, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.886467456817627, + "learning_rate": 1e-06, + "loss": 0.3913, + "mean_token_accuracy": 0.8663601875305176, + "num_tokens": 161826970.0, + "step": 4163 + }, + { + "epoch": 0.5297036000508841, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.940404415130615, + "learning_rate": 1e-06, + "loss": 0.3865, + "mean_token_accuracy": 0.8682829737663269, + "num_tokens": 161864705.0, + "step": 4164 + }, + { + "epoch": 0.5298308103294747, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.970011234283447, + "learning_rate": 1e-06, + "loss": 0.3769, + "mean_token_accuracy": 0.8742114305496216, + "num_tokens": 161902853.0, + "step": 4165 + }, + { + "epoch": 0.5299580206080652, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 8.029195785522461, + "learning_rate": 1e-06, + "loss": 0.4456, + "mean_token_accuracy": 0.8531508445739746, + "num_tokens": 161940666.0, + "step": 4166 + }, + { + "epoch": 0.5300852308866556, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.858320713043213, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8937472105026245, + "num_tokens": 161982192.0, + "step": 4167 + }, + { + "epoch": 0.5302124411652461, + "ewc_loss": 0.01019287109375, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.285045623779297e-06, + "grad_norm": 7.909458160400391, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8848427534103394, + "num_tokens": 162019095.0, + "step": 4168 + }, + { + "epoch": 0.5303396514438367, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.880560398101807, + "learning_rate": 1e-06, + "loss": 0.3757, + "mean_token_accuracy": 0.8709498047828674, + "num_tokens": 162060811.0, + "step": 4169 + }, + { + "epoch": 0.5304668617224272, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.931166648864746, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8797522187232971, + "num_tokens": 162096244.0, + "step": 4170 + }, + { + "epoch": 0.5305940720010177, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.957379341125488, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8785003423690796, + "num_tokens": 162133211.0, + "step": 4171 + }, + { + "epoch": 0.5307212822796082, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.861148834228516, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8797956109046936, + "num_tokens": 162175769.0, + "step": 4172 + }, + { + "epoch": 0.5308484925581987, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 8.057296752929688, + "learning_rate": 1e-06, + "loss": 0.4009, + "mean_token_accuracy": 0.8668380379676819, + "num_tokens": 162211904.0, + "step": 4173 + }, + { + "epoch": 0.5309757028367892, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.931084156036377, + "learning_rate": 1e-06, + "loss": 0.3858, + "mean_token_accuracy": 0.8702986240386963, + "num_tokens": 162247064.0, + "step": 4174 + }, + { + "epoch": 0.5311029131153797, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.967182159423828, + "learning_rate": 1e-06, + "loss": 0.4062, + "mean_token_accuracy": 0.8630856871604919, + "num_tokens": 162283332.0, + "step": 4175 + }, + { + "epoch": 0.5312301233939702, + "ewc_loss": 0.01031494140625, + "ewc_loss_diag": 1.8715858459472656e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.808286666870117, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8803322911262512, + "num_tokens": 162326684.0, + "step": 4176 + }, + { + "epoch": 0.5313573336725608, + "ewc_loss": 0.01025390625, + "ewc_loss_diag": 1.8835067749023438e-05, + "ewc_loss_parallel": 8.344650268554688e-06, + "grad_norm": 7.898188591003418, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8800212144851685, + "num_tokens": 162366260.0, + "step": 4177 + }, + { + "epoch": 0.5314845439511513, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.8835067749023438e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 8.049180030822754, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8768954277038574, + "num_tokens": 162397517.0, + "step": 4178 + }, + { + "epoch": 0.5316117542297417, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.8835067749023438e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.986636161804199, + "learning_rate": 1e-06, + "loss": 0.3967, + "mean_token_accuracy": 0.8647170066833496, + "num_tokens": 162432621.0, + "step": 4179 + }, + { + "epoch": 0.5317389645083322, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.8835067749023438e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.919891357421875, + "learning_rate": 1e-06, + "loss": 0.3779, + "mean_token_accuracy": 0.8727041482925415, + "num_tokens": 162475851.0, + "step": 4180 + }, + { + "epoch": 0.5318661747869228, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.8835067749023438e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.895729064941406, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8823772072792053, + "num_tokens": 162516035.0, + "step": 4181 + }, + { + "epoch": 0.5319933850655133, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.8835067749023438e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.879939556121826, + "learning_rate": 1e-06, + "loss": 0.4092, + "mean_token_accuracy": 0.8639696836471558, + "num_tokens": 162561870.0, + "step": 4182 + }, + { + "epoch": 0.5321205953441038, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.8835067749023438e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.815826892852783, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8765765428543091, + "num_tokens": 162603042.0, + "step": 4183 + }, + { + "epoch": 0.5322478056226944, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.8835067749023438e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 8.062716484069824, + "learning_rate": 1e-06, + "loss": 0.4026, + "mean_token_accuracy": 0.867008626461029, + "num_tokens": 162639439.0, + "step": 4184 + }, + { + "epoch": 0.5323750159012848, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.8835067749023438e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.962066650390625, + "learning_rate": 1e-06, + "loss": 0.3859, + "mean_token_accuracy": 0.8697868585586548, + "num_tokens": 162676392.0, + "step": 4185 + }, + { + "epoch": 0.5325022261798753, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.8835067749023438e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.893640041351318, + "learning_rate": 1e-06, + "loss": 0.4205, + "mean_token_accuracy": 0.8642512559890747, + "num_tokens": 162715197.0, + "step": 4186 + }, + { + "epoch": 0.5326294364584658, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.0475435256958, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8709038496017456, + "num_tokens": 162750022.0, + "step": 4187 + }, + { + "epoch": 0.5327566467370564, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.923147201538086, + "learning_rate": 1e-06, + "loss": 0.3924, + "mean_token_accuracy": 0.8682548999786377, + "num_tokens": 162787428.0, + "step": 4188 + }, + { + "epoch": 0.5328838570156469, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.931750774383545, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8688172101974487, + "num_tokens": 162825015.0, + "step": 4189 + }, + { + "epoch": 0.5330110672942374, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.926370143890381, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8831480741500854, + "num_tokens": 162861506.0, + "step": 4190 + }, + { + "epoch": 0.5331382775728278, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.851493835449219, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8842056393623352, + "num_tokens": 162902766.0, + "step": 4191 + }, + { + "epoch": 0.5332654878514184, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.956731796264648, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.8630947470664978, + "num_tokens": 162939782.0, + "step": 4192 + }, + { + "epoch": 0.5333926981300089, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9001264572143555, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8836618661880493, + "num_tokens": 162977358.0, + "step": 4193 + }, + { + "epoch": 0.5335199084085994, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.015869140625, + "learning_rate": 1e-06, + "loss": 0.4279, + "mean_token_accuracy": 0.8595995903015137, + "num_tokens": 163016065.0, + "step": 4194 + }, + { + "epoch": 0.53364711868719, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.063895225524902, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8734350204467773, + "num_tokens": 163047609.0, + "step": 4195 + }, + { + "epoch": 0.5337743289657805, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.8438401222229, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.8775046467781067, + "num_tokens": 163094330.0, + "step": 4196 + }, + { + "epoch": 0.5339015392443709, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.8984575271606445, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8759088516235352, + "num_tokens": 163132592.0, + "step": 4197 + }, + { + "epoch": 0.5340287495229614, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.03842544555664, + "learning_rate": 1e-06, + "loss": 0.393, + "mean_token_accuracy": 0.8674386143684387, + "num_tokens": 163167181.0, + "step": 4198 + }, + { + "epoch": 0.534155959801552, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.8665618896484375, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8891997933387756, + "num_tokens": 163205988.0, + "step": 4199 + }, + { + "epoch": 0.5342831700801425, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.985703468322754, + "learning_rate": 1e-06, + "loss": 0.4261, + "mean_token_accuracy": 0.8541408777236938, + "num_tokens": 163242438.0, + "step": 4200 + }, + { + "epoch": 0.534410380358733, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.929717063903809, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8820972442626953, + "num_tokens": 163284015.0, + "step": 4201 + }, + { + "epoch": 0.5345375906373235, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9318013191223145, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8695278763771057, + "num_tokens": 163320172.0, + "step": 4202 + }, + { + "epoch": 0.534664800915914, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.905550956726074, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8838995695114136, + "num_tokens": 163359235.0, + "step": 4203 + }, + { + "epoch": 0.5347920111945045, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.9522223472595215, + "learning_rate": 1e-06, + "loss": 0.3944, + "mean_token_accuracy": 0.8679582476615906, + "num_tokens": 163399912.0, + "step": 4204 + }, + { + "epoch": 0.534919221473095, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.087140083312988, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8804358243942261, + "num_tokens": 163429238.0, + "step": 4205 + }, + { + "epoch": 0.5350464317516855, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.875100612640381, + "learning_rate": 1e-06, + "loss": 0.3972, + "mean_token_accuracy": 0.8696776628494263, + "num_tokens": 163469610.0, + "step": 4206 + }, + { + "epoch": 0.5351736420302761, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.031764030456543, + "learning_rate": 1e-06, + "loss": 0.3939, + "mean_token_accuracy": 0.8669844269752502, + "num_tokens": 163507742.0, + "step": 4207 + }, + { + "epoch": 0.5353008523088666, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.920011043548584, + "learning_rate": 1e-06, + "loss": 0.4128, + "mean_token_accuracy": 0.8614968657493591, + "num_tokens": 163546670.0, + "step": 4208 + }, + { + "epoch": 0.5354280625874571, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.847346305847168, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8852143287658691, + "num_tokens": 163587927.0, + "step": 4209 + }, + { + "epoch": 0.5355552728660475, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.872866630554199, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8799265623092651, + "num_tokens": 163634170.0, + "step": 4210 + }, + { + "epoch": 0.5356824831446381, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.849485874176025, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8888139128684998, + "num_tokens": 163678941.0, + "step": 4211 + }, + { + "epoch": 0.5358096934232286, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.974260330200195, + "learning_rate": 1e-06, + "loss": 0.3952, + "mean_token_accuracy": 0.8651326298713684, + "num_tokens": 163718842.0, + "step": 4212 + }, + { + "epoch": 0.5359369037018191, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.965034008026123, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8816648125648499, + "num_tokens": 163760403.0, + "step": 4213 + }, + { + "epoch": 0.5360641139804097, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.977416515350342, + "learning_rate": 1e-06, + "loss": 0.3993, + "mean_token_accuracy": 0.8679368495941162, + "num_tokens": 163800206.0, + "step": 4214 + }, + { + "epoch": 0.5361913242590002, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.966975212097168, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.8715436458587646, + "num_tokens": 163841044.0, + "step": 4215 + }, + { + "epoch": 0.5363185345375906, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.898119926452637, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.8766691088676453, + "num_tokens": 163879736.0, + "step": 4216 + }, + { + "epoch": 0.5364457448161811, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.910701274871826, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8979891538619995, + "num_tokens": 163917281.0, + "step": 4217 + }, + { + "epoch": 0.5365729550947717, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.980777263641357, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.883018970489502, + "num_tokens": 163952708.0, + "step": 4218 + }, + { + "epoch": 0.5367001653733622, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.951566696166992, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8778328895568848, + "num_tokens": 163993182.0, + "step": 4219 + }, + { + "epoch": 0.5368273756519527, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 8.14399528503418, + "learning_rate": 1e-06, + "loss": 0.4148, + "mean_token_accuracy": 0.8601179122924805, + "num_tokens": 164024158.0, + "step": 4220 + }, + { + "epoch": 0.5369545859305432, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.920602798461914, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.8707014322280884, + "num_tokens": 164059717.0, + "step": 4221 + }, + { + "epoch": 0.5370817962091337, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.836419105529785, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8915106058120728, + "num_tokens": 164096902.0, + "step": 4222 + }, + { + "epoch": 0.5372090064877242, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.911096096038818, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8804394006729126, + "num_tokens": 164138748.0, + "step": 4223 + }, + { + "epoch": 0.5373362167663147, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.990592956542969, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.8729561567306519, + "num_tokens": 164173827.0, + "step": 4224 + }, + { + "epoch": 0.5374634270449052, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.862611770629883, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8805210590362549, + "num_tokens": 164214369.0, + "step": 4225 + }, + { + "epoch": 0.5375906373234958, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.929169178009033, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8853121995925903, + "num_tokens": 164249124.0, + "step": 4226 + }, + { + "epoch": 0.5377178476020863, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.955987930297852, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8758244514465332, + "num_tokens": 164287927.0, + "step": 4227 + }, + { + "epoch": 0.5378450578806767, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.996103763580322, + "learning_rate": 1e-06, + "loss": 0.3923, + "mean_token_accuracy": 0.8682853579521179, + "num_tokens": 164326374.0, + "step": 4228 + }, + { + "epoch": 0.5379722681592672, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.895427703857422e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.998956680297852, + "learning_rate": 1e-06, + "loss": 0.4216, + "mean_token_accuracy": 0.858540952205658, + "num_tokens": 164363176.0, + "step": 4229 + }, + { + "epoch": 0.5380994784378578, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.919269561767578e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.896298885345459, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8827159404754639, + "num_tokens": 164398653.0, + "step": 4230 + }, + { + "epoch": 0.5382266887164483, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.919269561767578e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.999448299407959, + "learning_rate": 1e-06, + "loss": 0.3768, + "mean_token_accuracy": 0.8726706504821777, + "num_tokens": 164434920.0, + "step": 4231 + }, + { + "epoch": 0.5383538989950388, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.919269561767578e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.968955039978027, + "learning_rate": 1e-06, + "loss": 0.4263, + "mean_token_accuracy": 0.8583230972290039, + "num_tokens": 164473306.0, + "step": 4232 + }, + { + "epoch": 0.5384811092736294, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.919269561767578e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.92924165725708, + "learning_rate": 1e-06, + "loss": 0.3859, + "mean_token_accuracy": 0.8727939128875732, + "num_tokens": 164511353.0, + "step": 4233 + }, + { + "epoch": 0.5386083195522198, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.919269561767578e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.87485933303833, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8776836395263672, + "num_tokens": 164552527.0, + "step": 4234 + }, + { + "epoch": 0.5387355298308103, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.919269561767578e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.864066123962402, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.880948007106781, + "num_tokens": 164591854.0, + "step": 4235 + }, + { + "epoch": 0.5388627401094008, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.919269561767578e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.995108127593994, + "learning_rate": 1e-06, + "loss": 0.3784, + "mean_token_accuracy": 0.8730649352073669, + "num_tokens": 164627628.0, + "step": 4236 + }, + { + "epoch": 0.5389899503879914, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.919269561767578e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.931429386138916, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8777826428413391, + "num_tokens": 164662520.0, + "step": 4237 + }, + { + "epoch": 0.5391171606665819, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.919269561767578e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.987522125244141, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8733153343200684, + "num_tokens": 164699128.0, + "step": 4238 + }, + { + "epoch": 0.5392443709451724, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.919269561767578e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.895349979400635, + "learning_rate": 1e-06, + "loss": 0.4084, + "mean_token_accuracy": 0.8610039949417114, + "num_tokens": 164740623.0, + "step": 4239 + }, + { + "epoch": 0.5393715812237628, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.919269561767578e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.072407722473145, + "learning_rate": 1e-06, + "loss": 0.4243, + "mean_token_accuracy": 0.8561325669288635, + "num_tokens": 164777246.0, + "step": 4240 + }, + { + "epoch": 0.5394987915023534, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.919269561767578e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.974409580230713, + "learning_rate": 1e-06, + "loss": 0.4082, + "mean_token_accuracy": 0.8627821207046509, + "num_tokens": 164812848.0, + "step": 4241 + }, + { + "epoch": 0.5396260017809439, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.919269561767578e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.8952507972717285, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8812447786331177, + "num_tokens": 164850018.0, + "step": 4242 + }, + { + "epoch": 0.5397532120595344, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.843047142028809, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8846311569213867, + "num_tokens": 164888980.0, + "step": 4243 + }, + { + "epoch": 0.5398804223381249, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.065185546875, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.877319872379303, + "num_tokens": 164923771.0, + "step": 4244 + }, + { + "epoch": 0.5400076326167155, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.844457626342773, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8757013082504272, + "num_tokens": 164964901.0, + "step": 4245 + }, + { + "epoch": 0.5401348428953059, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9791131019592285, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8779311776161194, + "num_tokens": 165002540.0, + "step": 4246 + }, + { + "epoch": 0.5402620531738964, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.964443683624268, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.884162187576294, + "num_tokens": 165041104.0, + "step": 4247 + }, + { + "epoch": 0.540389263452487, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.860504150390625, + "learning_rate": 1e-06, + "loss": 0.3817, + "mean_token_accuracy": 0.8706638813018799, + "num_tokens": 165083199.0, + "step": 4248 + }, + { + "epoch": 0.5405164737310775, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.935793876647949, + "learning_rate": 1e-06, + "loss": 0.3843, + "mean_token_accuracy": 0.8678773045539856, + "num_tokens": 165125974.0, + "step": 4249 + }, + { + "epoch": 0.540643684009668, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.904290676116943, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8890878558158875, + "num_tokens": 165165708.0, + "step": 4250 + }, + { + "epoch": 0.5407708942882585, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.994935035705566, + "learning_rate": 1e-06, + "loss": 0.3885, + "mean_token_accuracy": 0.8674357533454895, + "num_tokens": 165203841.0, + "step": 4251 + }, + { + "epoch": 0.540898104566849, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.944214820861816, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8774020671844482, + "num_tokens": 165239646.0, + "step": 4252 + }, + { + "epoch": 0.5410253148454395, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.963168144226074, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8734951019287109, + "num_tokens": 165279024.0, + "step": 4253 + }, + { + "epoch": 0.54115252512403, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.023005485534668, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8945052623748779, + "num_tokens": 165309803.0, + "step": 4254 + }, + { + "epoch": 0.5412797354026205, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.921065330505371, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.8767229318618774, + "num_tokens": 165353000.0, + "step": 4255 + }, + { + "epoch": 0.5414069456812111, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9073486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.987717628479004, + "learning_rate": 1e-06, + "loss": 0.4108, + "mean_token_accuracy": 0.8641825318336487, + "num_tokens": 165389726.0, + "step": 4256 + }, + { + "epoch": 0.5415341559598016, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.974250793457031, + "learning_rate": 1e-06, + "loss": 0.387, + "mean_token_accuracy": 0.8756440281867981, + "num_tokens": 165429882.0, + "step": 4257 + }, + { + "epoch": 0.5416613662383921, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.938892841339111, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8719144463539124, + "num_tokens": 165469407.0, + "step": 4258 + }, + { + "epoch": 0.5417885765169825, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.955061912536621, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8777835965156555, + "num_tokens": 165503834.0, + "step": 4259 + }, + { + "epoch": 0.5419157867955731, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.854804515838623, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8916939496994019, + "num_tokens": 165544566.0, + "step": 4260 + }, + { + "epoch": 0.5420429970741636, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.974327564239502, + "learning_rate": 1e-06, + "loss": 0.4188, + "mean_token_accuracy": 0.8587002158164978, + "num_tokens": 165585492.0, + "step": 4261 + }, + { + "epoch": 0.5421702073527541, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.921751499176025, + "learning_rate": 1e-06, + "loss": 0.3884, + "mean_token_accuracy": 0.869458794593811, + "num_tokens": 165624038.0, + "step": 4262 + }, + { + "epoch": 0.5422974176313446, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.890736103057861, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8685165047645569, + "num_tokens": 165663084.0, + "step": 4263 + }, + { + "epoch": 0.5424246279099352, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.91903829574585, + "learning_rate": 1e-06, + "loss": 0.3887, + "mean_token_accuracy": 0.8721113204956055, + "num_tokens": 165709979.0, + "step": 4264 + }, + { + "epoch": 0.5425518381885256, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.008816719055176, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8788866996765137, + "num_tokens": 165746255.0, + "step": 4265 + }, + { + "epoch": 0.5426790484671161, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.940022945404053, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8783890008926392, + "num_tokens": 165782972.0, + "step": 4266 + }, + { + "epoch": 0.5428062587457066, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.066069602966309, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8754870295524597, + "num_tokens": 165813552.0, + "step": 4267 + }, + { + "epoch": 0.5429334690242972, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.987135410308838, + "learning_rate": 1e-06, + "loss": 0.4183, + "mean_token_accuracy": 0.8629003763198853, + "num_tokens": 165853554.0, + "step": 4268 + }, + { + "epoch": 0.5430606793028877, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.916209697723389, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8829931020736694, + "num_tokens": 165889377.0, + "step": 4269 + }, + { + "epoch": 0.5431878895814782, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9311904907226562e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.06336784362793, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.8654310703277588, + "num_tokens": 165926218.0, + "step": 4270 + }, + { + "epoch": 0.5433150998600687, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 1.9431114196777344e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.89406681060791, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8876432180404663, + "num_tokens": 165966927.0, + "step": 4271 + }, + { + "epoch": 0.5434423101386592, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9431114196777344e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.9242167472839355, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8891938924789429, + "num_tokens": 166004390.0, + "step": 4272 + }, + { + "epoch": 0.5435695204172497, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9431114196777344e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.990420818328857, + "learning_rate": 1e-06, + "loss": 0.3984, + "mean_token_accuracy": 0.8685598373413086, + "num_tokens": 166045247.0, + "step": 4273 + }, + { + "epoch": 0.5436967306958402, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9431114196777344e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.040008544921875, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8715806007385254, + "num_tokens": 166079284.0, + "step": 4274 + }, + { + "epoch": 0.5438239409744308, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 1.9431114196777344e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.995304584503174, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8806518316268921, + "num_tokens": 166113262.0, + "step": 4275 + }, + { + "epoch": 0.5439511512530213, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9431114196777344e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.895505905151367, + "learning_rate": 1e-06, + "loss": 0.3892, + "mean_token_accuracy": 0.8693764209747314, + "num_tokens": 166158810.0, + "step": 4276 + }, + { + "epoch": 0.5440783615316117, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9431114196777344e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.849585056304932, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8824166059494019, + "num_tokens": 166204178.0, + "step": 4277 + }, + { + "epoch": 0.5442055718102022, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9431114196777344e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.021533012390137, + "learning_rate": 1e-06, + "loss": 0.4137, + "mean_token_accuracy": 0.8647403120994568, + "num_tokens": 166246590.0, + "step": 4278 + }, + { + "epoch": 0.5443327820887928, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 1.9431114196777344e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.983253002166748, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8765467405319214, + "num_tokens": 166279587.0, + "step": 4279 + }, + { + "epoch": 0.5444599923673833, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9431114196777344e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.982254505157471, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.877827525138855, + "num_tokens": 166313595.0, + "step": 4280 + }, + { + "epoch": 0.5445872026459738, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.031832695007324, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8780730366706848, + "num_tokens": 166347805.0, + "step": 4281 + }, + { + "epoch": 0.5447144129245644, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.886281490325928, + "learning_rate": 1e-06, + "loss": 0.3814, + "mean_token_accuracy": 0.8718280792236328, + "num_tokens": 166387352.0, + "step": 4282 + }, + { + "epoch": 0.5448416232031548, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.048851013183594, + "learning_rate": 1e-06, + "loss": 0.4198, + "mean_token_accuracy": 0.860537588596344, + "num_tokens": 166423921.0, + "step": 4283 + }, + { + "epoch": 0.5449688334817453, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.986292362213135, + "learning_rate": 1e-06, + "loss": 0.4593, + "mean_token_accuracy": 0.8462345600128174, + "num_tokens": 166467394.0, + "step": 4284 + }, + { + "epoch": 0.5450960437603358, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.99890661239624, + "learning_rate": 1e-06, + "loss": 0.4038, + "mean_token_accuracy": 0.8624350428581238, + "num_tokens": 166505947.0, + "step": 4285 + }, + { + "epoch": 0.5452232540389264, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.946135997772217, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8957988023757935, + "num_tokens": 166538811.0, + "step": 4286 + }, + { + "epoch": 0.5453504643175169, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.965414524078369, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8805679082870483, + "num_tokens": 166574728.0, + "step": 4287 + }, + { + "epoch": 0.5454776745961074, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.013141632080078, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8875083923339844, + "num_tokens": 166608765.0, + "step": 4288 + }, + { + "epoch": 0.5456048848746978, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.914030075073242, + "learning_rate": 1e-06, + "loss": 0.3733, + "mean_token_accuracy": 0.8740831017494202, + "num_tokens": 166650168.0, + "step": 4289 + }, + { + "epoch": 0.5457320951532884, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.040454864501953, + "learning_rate": 1e-06, + "loss": 0.384, + "mean_token_accuracy": 0.8707373738288879, + "num_tokens": 166684251.0, + "step": 4290 + }, + { + "epoch": 0.5458593054318789, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.982519149780273, + "learning_rate": 1e-06, + "loss": 0.3861, + "mean_token_accuracy": 0.8698636293411255, + "num_tokens": 166722546.0, + "step": 4291 + }, + { + "epoch": 0.5459865157104694, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.968716621398926, + "learning_rate": 1e-06, + "loss": 0.387, + "mean_token_accuracy": 0.8715530633926392, + "num_tokens": 166758358.0, + "step": 4292 + }, + { + "epoch": 0.5461137259890599, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.8998332023620605, + "learning_rate": 1e-06, + "loss": 0.3947, + "mean_token_accuracy": 0.8635631799697876, + "num_tokens": 166799565.0, + "step": 4293 + }, + { + "epoch": 0.5462409362676505, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.869503974914551, + "learning_rate": 1e-06, + "loss": 0.4318, + "mean_token_accuracy": 0.8556782007217407, + "num_tokens": 166844373.0, + "step": 4294 + }, + { + "epoch": 0.5463681465462409, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.010557174682617, + "learning_rate": 1e-06, + "loss": 0.4898, + "mean_token_accuracy": 0.8380976915359497, + "num_tokens": 166892881.0, + "step": 4295 + }, + { + "epoch": 0.5464953568248314, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.860584735870361, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8942012190818787, + "num_tokens": 166933887.0, + "step": 4296 + }, + { + "epoch": 0.5466225671034219, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.038514137268066, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8737851977348328, + "num_tokens": 166969805.0, + "step": 4297 + }, + { + "epoch": 0.5467497773820125, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.165948867797852, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8823947310447693, + "num_tokens": 167002408.0, + "step": 4298 + }, + { + "epoch": 0.546876987660603, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.88859748840332, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.8708466291427612, + "num_tokens": 167043390.0, + "step": 4299 + }, + { + "epoch": 0.5470041979391935, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.864448070526123, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.881469190120697, + "num_tokens": 167086696.0, + "step": 4300 + }, + { + "epoch": 0.5471314082177839, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.972481727600098, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8703908324241638, + "num_tokens": 167123410.0, + "step": 4301 + }, + { + "epoch": 0.5472586184963745, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.926242351531982, + "learning_rate": 1e-06, + "loss": 0.381, + "mean_token_accuracy": 0.8684300184249878, + "num_tokens": 167164241.0, + "step": 4302 + }, + { + "epoch": 0.547385828774965, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.955543518066406, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8752830028533936, + "num_tokens": 167204974.0, + "step": 4303 + }, + { + "epoch": 0.5475130390535555, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.934111595153809, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.8651145100593567, + "num_tokens": 167243638.0, + "step": 4304 + }, + { + "epoch": 0.5476402493321461, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.97454309463501, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8793079853057861, + "num_tokens": 167279012.0, + "step": 4305 + }, + { + "epoch": 0.5477674596107366, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.872797012329102, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8782692551612854, + "num_tokens": 167322563.0, + "step": 4306 + }, + { + "epoch": 0.5478946698893271, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.04957103729248, + "learning_rate": 1e-06, + "loss": 0.3919, + "mean_token_accuracy": 0.8656142950057983, + "num_tokens": 167359010.0, + "step": 4307 + }, + { + "epoch": 0.5480218801679175, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.860881805419922, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8750662207603455, + "num_tokens": 167402286.0, + "step": 4308 + }, + { + "epoch": 0.5481490904465081, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.9336466789245605, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8768739700317383, + "num_tokens": 167440511.0, + "step": 4309 + }, + { + "epoch": 0.5482763007250986, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.010954856872559, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.871307373046875, + "num_tokens": 167477261.0, + "step": 4310 + }, + { + "epoch": 0.5484035110036891, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.982210159301758, + "learning_rate": 1e-06, + "loss": 0.3643, + "mean_token_accuracy": 0.8744126558303833, + "num_tokens": 167512660.0, + "step": 4311 + }, + { + "epoch": 0.5485307212822796, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.90946102142334, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8820053339004517, + "num_tokens": 167551197.0, + "step": 4312 + }, + { + "epoch": 0.5486579315608702, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.9210309982299805, + "learning_rate": 1e-06, + "loss": 0.3825, + "mean_token_accuracy": 0.8727251291275024, + "num_tokens": 167592696.0, + "step": 4313 + }, + { + "epoch": 0.5487851418394606, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.969976902008057, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8754146695137024, + "num_tokens": 167632457.0, + "step": 4314 + }, + { + "epoch": 0.5489123521180511, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.940695762634277, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.88715660572052, + "num_tokens": 167670877.0, + "step": 4315 + }, + { + "epoch": 0.5490395623966416, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.980348587036133, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8759093284606934, + "num_tokens": 167709388.0, + "step": 4316 + }, + { + "epoch": 0.5491667726752322, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.932560920715332, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.886377215385437, + "num_tokens": 167745459.0, + "step": 4317 + }, + { + "epoch": 0.5492939829538227, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.869085788726807, + "learning_rate": 1e-06, + "loss": 0.3814, + "mean_token_accuracy": 0.8710587620735168, + "num_tokens": 167794590.0, + "step": 4318 + }, + { + "epoch": 0.5494211932324132, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.961010456085205, + "learning_rate": 1e-06, + "loss": 0.3965, + "mean_token_accuracy": 0.8728443384170532, + "num_tokens": 167836670.0, + "step": 4319 + }, + { + "epoch": 0.5495484035110036, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9550323486328125e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.055691719055176, + "learning_rate": 1e-06, + "loss": 0.3826, + "mean_token_accuracy": 0.869419276714325, + "num_tokens": 167871944.0, + "step": 4320 + }, + { + "epoch": 0.5496756137895942, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.914347171783447, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.885776698589325, + "num_tokens": 167910174.0, + "step": 4321 + }, + { + "epoch": 0.5498028240681847, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.907100200653076, + "learning_rate": 1e-06, + "loss": 0.3962, + "mean_token_accuracy": 0.865408182144165, + "num_tokens": 167957583.0, + "step": 4322 + }, + { + "epoch": 0.5499300343467752, + "ewc_loss": 0.0103759765625, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.404254913330078e-06, + "grad_norm": 7.915789604187012, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.8735411167144775, + "num_tokens": 168002446.0, + "step": 4323 + }, + { + "epoch": 0.5500572446253658, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.941570281982422, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8892607688903809, + "num_tokens": 168042132.0, + "step": 4324 + }, + { + "epoch": 0.5501844549039563, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.934423923492432, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8888500332832336, + "num_tokens": 168077148.0, + "step": 4325 + }, + { + "epoch": 0.5503116651825467, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.952001094818115, + "learning_rate": 1e-06, + "loss": 0.3783, + "mean_token_accuracy": 0.871658444404602, + "num_tokens": 168116676.0, + "step": 4326 + }, + { + "epoch": 0.5504388754611372, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.972330570220947, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.8782916069030762, + "num_tokens": 168150042.0, + "step": 4327 + }, + { + "epoch": 0.5505660857397278, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.989586353302002, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8729104399681091, + "num_tokens": 168186001.0, + "step": 4328 + }, + { + "epoch": 0.5506932960183183, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.911549091339111, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8734354972839355, + "num_tokens": 168221879.0, + "step": 4329 + }, + { + "epoch": 0.5508205062969088, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.968817234039307, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8887592554092407, + "num_tokens": 168259177.0, + "step": 4330 + }, + { + "epoch": 0.5509477165754993, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.973987579345703, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8722015619277954, + "num_tokens": 168296797.0, + "step": 4331 + }, + { + "epoch": 0.5510749268540898, + "ewc_loss": 0.01043701171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 8.015071868896484, + "learning_rate": 1e-06, + "loss": 0.4244, + "mean_token_accuracy": 0.859052300453186, + "num_tokens": 168334891.0, + "step": 4332 + }, + { + "epoch": 0.5512021371326803, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.9110822677612305, + "learning_rate": 1e-06, + "loss": 0.4379, + "mean_token_accuracy": 0.853295624256134, + "num_tokens": 168378715.0, + "step": 4333 + }, + { + "epoch": 0.5513293474112708, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.888576984405518, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.886454164981842, + "num_tokens": 168418558.0, + "step": 4334 + }, + { + "epoch": 0.5514565576898613, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.03230094909668, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8698741793632507, + "num_tokens": 168452548.0, + "step": 4335 + }, + { + "epoch": 0.5515837679684519, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.929860591888428, + "learning_rate": 1e-06, + "loss": 0.3843, + "mean_token_accuracy": 0.8705622553825378, + "num_tokens": 168492541.0, + "step": 4336 + }, + { + "epoch": 0.5517109782470424, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.9873738288879395, + "learning_rate": 1e-06, + "loss": 0.386, + "mean_token_accuracy": 0.8689403533935547, + "num_tokens": 168532570.0, + "step": 4337 + }, + { + "epoch": 0.5518381885256328, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.959326267242432, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8774772882461548, + "num_tokens": 168572341.0, + "step": 4338 + }, + { + "epoch": 0.5519653988042234, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.97763204574585, + "learning_rate": 1e-06, + "loss": 0.3949, + "mean_token_accuracy": 0.8679684400558472, + "num_tokens": 168611966.0, + "step": 4339 + }, + { + "epoch": 0.5520926090828139, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.905543327331543, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8886130452156067, + "num_tokens": 168649011.0, + "step": 4340 + }, + { + "epoch": 0.5522198193614044, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.95013952255249, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8803828954696655, + "num_tokens": 168684654.0, + "step": 4341 + }, + { + "epoch": 0.5523470296399949, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.8409199714660645, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8759072422981262, + "num_tokens": 168731386.0, + "step": 4342 + }, + { + "epoch": 0.5524742399185855, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 8.035810470581055, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8772144913673401, + "num_tokens": 168761664.0, + "step": 4343 + }, + { + "epoch": 0.5526014501971759, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.073492050170898, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8770303130149841, + "num_tokens": 168796033.0, + "step": 4344 + }, + { + "epoch": 0.5527286604757664, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.844690799713135, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8825730085372925, + "num_tokens": 168844190.0, + "step": 4345 + }, + { + "epoch": 0.5528558707543569, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.994402885437012, + "learning_rate": 1e-06, + "loss": 0.4207, + "mean_token_accuracy": 0.8593482971191406, + "num_tokens": 168888902.0, + "step": 4346 + }, + { + "epoch": 0.5529830810329475, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.023225784301758, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8693447113037109, + "num_tokens": 168922064.0, + "step": 4347 + }, + { + "epoch": 0.553110291311538, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9669532775878906e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.9339919090271, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8761662244796753, + "num_tokens": 168962309.0, + "step": 4348 + }, + { + "epoch": 0.5532375015901285, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9788742065429688e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.880202293395996, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8702186346054077, + "num_tokens": 169004669.0, + "step": 4349 + }, + { + "epoch": 0.5533647118687189, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9788742065429688e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.006638526916504, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8818663954734802, + "num_tokens": 169046117.0, + "step": 4350 + }, + { + "epoch": 0.5534919221473095, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9788742065429688e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.9753737449646, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8707249164581299, + "num_tokens": 169085983.0, + "step": 4351 + }, + { + "epoch": 0.5536191324259, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.9842939376831055, + "learning_rate": 1e-06, + "loss": 0.386, + "mean_token_accuracy": 0.8694398999214172, + "num_tokens": 169126493.0, + "step": 4352 + }, + { + "epoch": 0.5537463427044905, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.930121898651123, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8754370212554932, + "num_tokens": 169164489.0, + "step": 4353 + }, + { + "epoch": 0.553873552983081, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.970325946807861, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8765811324119568, + "num_tokens": 169205999.0, + "step": 4354 + }, + { + "epoch": 0.5540007632616716, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.0182466506958, + "learning_rate": 1e-06, + "loss": 0.4026, + "mean_token_accuracy": 0.86513352394104, + "num_tokens": 169244272.0, + "step": 4355 + }, + { + "epoch": 0.554127973540262, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9788742065429688e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.996366500854492, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8704166412353516, + "num_tokens": 169281793.0, + "step": 4356 + }, + { + "epoch": 0.5542551838188525, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 7.908365726470947, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8804574608802795, + "num_tokens": 169323809.0, + "step": 4357 + }, + { + "epoch": 0.554382394097443, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 8.083230972290039, + "learning_rate": 1e-06, + "loss": 0.4492, + "mean_token_accuracy": 0.8527894020080566, + "num_tokens": 169360670.0, + "step": 4358 + }, + { + "epoch": 0.5545096043760336, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.91619873046875, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8770831823348999, + "num_tokens": 169400403.0, + "step": 4359 + }, + { + "epoch": 0.5546368146546241, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9788742065429688e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.944422721862793, + "learning_rate": 1e-06, + "loss": 0.3999, + "mean_token_accuracy": 0.8685445785522461, + "num_tokens": 169443510.0, + "step": 4360 + }, + { + "epoch": 0.5547640249332146, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.9788742065429688e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.938446044921875, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8771622180938721, + "num_tokens": 169480799.0, + "step": 4361 + }, + { + "epoch": 0.5548912352118052, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9788742065429688e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.039804458618164, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8685474991798401, + "num_tokens": 169515397.0, + "step": 4362 + }, + { + "epoch": 0.5550184454903956, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9788742065429688e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.976594924926758, + "learning_rate": 1e-06, + "loss": 0.3947, + "mean_token_accuracy": 0.8687923550605774, + "num_tokens": 169552237.0, + "step": 4363 + }, + { + "epoch": 0.5551456557689861, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.9788742065429688e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.976153373718262, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8846685886383057, + "num_tokens": 169589312.0, + "step": 4364 + }, + { + "epoch": 0.5552728660475766, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9788742065429688e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.031465530395508, + "learning_rate": 1e-06, + "loss": 0.4228, + "mean_token_accuracy": 0.8588485717773438, + "num_tokens": 169627542.0, + "step": 4365 + }, + { + "epoch": 0.5554000763261672, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.9788742065429688e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.893426418304443, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8743031620979309, + "num_tokens": 169669738.0, + "step": 4366 + }, + { + "epoch": 0.5555272866047577, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.985861778259277, + "learning_rate": 1e-06, + "loss": 0.3773, + "mean_token_accuracy": 0.869988203048706, + "num_tokens": 169706902.0, + "step": 4367 + }, + { + "epoch": 0.5556544968833482, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.064163208007812, + "learning_rate": 1e-06, + "loss": 0.399, + "mean_token_accuracy": 0.8650435209274292, + "num_tokens": 169738965.0, + "step": 4368 + }, + { + "epoch": 0.5557817071619386, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.03357982635498, + "learning_rate": 1e-06, + "loss": 0.3816, + "mean_token_accuracy": 0.8735426068305969, + "num_tokens": 169772988.0, + "step": 4369 + }, + { + "epoch": 0.5559089174405292, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.973530292510986, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8697510957717896, + "num_tokens": 169810039.0, + "step": 4370 + }, + { + "epoch": 0.5560361277191197, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.905124187469482, + "learning_rate": 1e-06, + "loss": 0.3676, + "mean_token_accuracy": 0.8744386434555054, + "num_tokens": 169851937.0, + "step": 4371 + }, + { + "epoch": 0.5561633379977102, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9915876388549805, + "learning_rate": 1e-06, + "loss": 0.4056, + "mean_token_accuracy": 0.862183690071106, + "num_tokens": 169889071.0, + "step": 4372 + }, + { + "epoch": 0.5562905482763008, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.073259353637695, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8761032223701477, + "num_tokens": 169926757.0, + "step": 4373 + }, + { + "epoch": 0.5564177585548913, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9131317138671875, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8686469793319702, + "num_tokens": 169966800.0, + "step": 4374 + }, + { + "epoch": 0.5565449688334817, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.052835464477539, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8847067356109619, + "num_tokens": 170002016.0, + "step": 4375 + }, + { + "epoch": 0.5566721791120722, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.951536178588867, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8800637125968933, + "num_tokens": 170040695.0, + "step": 4376 + }, + { + "epoch": 0.5567993893906628, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9656500816345215, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8728312253952026, + "num_tokens": 170082238.0, + "step": 4377 + }, + { + "epoch": 0.5569265996692533, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.893805027008057, + "learning_rate": 1e-06, + "loss": 0.3733, + "mean_token_accuracy": 0.8723252415657043, + "num_tokens": 170126148.0, + "step": 4378 + }, + { + "epoch": 0.5570538099478438, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.062141418457031, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.8729079365730286, + "num_tokens": 170161802.0, + "step": 4379 + }, + { + "epoch": 0.5571810202264343, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.052495956420898, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8770411014556885, + "num_tokens": 170197548.0, + "step": 4380 + }, + { + "epoch": 0.5573082305050248, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.020861625671387, + "learning_rate": 1e-06, + "loss": 0.4045, + "mean_token_accuracy": 0.8624963760375977, + "num_tokens": 170230954.0, + "step": 4381 + }, + { + "epoch": 0.5574354407836153, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.002588272094727, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.871367335319519, + "num_tokens": 170268513.0, + "step": 4382 + }, + { + "epoch": 0.5575626510622058, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.982995986938477, + "learning_rate": 1e-06, + "loss": 0.3962, + "mean_token_accuracy": 0.8656021356582642, + "num_tokens": 170310914.0, + "step": 4383 + }, + { + "epoch": 0.5576898613407963, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.980795383453369, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8880719542503357, + "num_tokens": 170348166.0, + "step": 4384 + }, + { + "epoch": 0.5578170716193869, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.981068134307861, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.87436842918396, + "num_tokens": 170384201.0, + "step": 4385 + }, + { + "epoch": 0.5579442818979774, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 1.990795135498047e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.956532001495361, + "learning_rate": 1e-06, + "loss": 0.3754, + "mean_token_accuracy": 0.8775489330291748, + "num_tokens": 170422227.0, + "step": 4386 + }, + { + "epoch": 0.5580714921765678, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.002716064453125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.852250576019287, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8772571682929993, + "num_tokens": 170461553.0, + "step": 4387 + }, + { + "epoch": 0.5581987024551583, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.002716064453125e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.05964469909668, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8714576363563538, + "num_tokens": 170500935.0, + "step": 4388 + }, + { + "epoch": 0.5583259127337489, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.002716064453125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.941643238067627, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8784379959106445, + "num_tokens": 170544187.0, + "step": 4389 + }, + { + "epoch": 0.5584531230123394, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.931699275970459, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8818620443344116, + "num_tokens": 170584342.0, + "step": 4390 + }, + { + "epoch": 0.5585803332909299, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.936262607574463, + "learning_rate": 1e-06, + "loss": 0.3911, + "mean_token_accuracy": 0.8672537803649902, + "num_tokens": 170631147.0, + "step": 4391 + }, + { + "epoch": 0.5587075435695205, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.900239944458008, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8759647011756897, + "num_tokens": 170671820.0, + "step": 4392 + }, + { + "epoch": 0.5588347538481109, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.032946586608887, + "learning_rate": 1e-06, + "loss": 0.4118, + "mean_token_accuracy": 0.8610634803771973, + "num_tokens": 170710231.0, + "step": 4393 + }, + { + "epoch": 0.5589619641267014, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.979547023773193, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8804833889007568, + "num_tokens": 170748677.0, + "step": 4394 + }, + { + "epoch": 0.5590891744052919, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.938720226287842, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8943474888801575, + "num_tokens": 170786932.0, + "step": 4395 + }, + { + "epoch": 0.5592163846838825, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.004622459411621, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8815445899963379, + "num_tokens": 170821109.0, + "step": 4396 + }, + { + "epoch": 0.559343594962473, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.002716064453125e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.885322093963623, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8744345307350159, + "num_tokens": 170865622.0, + "step": 4397 + }, + { + "epoch": 0.5594708052410635, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 8.015460968017578, + "learning_rate": 1e-06, + "loss": 0.3831, + "mean_token_accuracy": 0.8679946064949036, + "num_tokens": 170901624.0, + "step": 4398 + }, + { + "epoch": 0.5595980155196539, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.973709583282471, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8752861022949219, + "num_tokens": 170942938.0, + "step": 4399 + }, + { + "epoch": 0.5597252257982445, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.931069850921631, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8887059092521667, + "num_tokens": 170979210.0, + "step": 4400 + }, + { + "epoch": 0.559852436076835, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 8.029369354248047, + "learning_rate": 1e-06, + "loss": 0.3922, + "mean_token_accuracy": 0.8640098571777344, + "num_tokens": 171016228.0, + "step": 4401 + }, + { + "epoch": 0.5599796463554255, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.018394470214844, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8793509006500244, + "num_tokens": 171046006.0, + "step": 4402 + }, + { + "epoch": 0.560106856634016, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.872257232666016, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8789619207382202, + "num_tokens": 171092733.0, + "step": 4403 + }, + { + "epoch": 0.5602340669126066, + "ewc_loss": 0.010498046875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.463859558105469e-06, + "grad_norm": 7.972711086273193, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8788048028945923, + "num_tokens": 171130745.0, + "step": 4404 + }, + { + "epoch": 0.560361277191197, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.015068054199219, + "learning_rate": 1e-06, + "loss": 0.4171, + "mean_token_accuracy": 0.8609439134597778, + "num_tokens": 171170821.0, + "step": 4405 + }, + { + "epoch": 0.5604884874697875, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.980343818664551, + "learning_rate": 1e-06, + "loss": 0.3815, + "mean_token_accuracy": 0.8705849647521973, + "num_tokens": 171212015.0, + "step": 4406 + }, + { + "epoch": 0.560615697748378, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.028285026550293, + "learning_rate": 1e-06, + "loss": 0.3928, + "mean_token_accuracy": 0.868135392665863, + "num_tokens": 171249956.0, + "step": 4407 + }, + { + "epoch": 0.5607429080269686, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.909149169921875, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8789792060852051, + "num_tokens": 171293551.0, + "step": 4408 + }, + { + "epoch": 0.5608701183055591, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.933833599090576, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8814406991004944, + "num_tokens": 171330155.0, + "step": 4409 + }, + { + "epoch": 0.5609973285841496, + "ewc_loss": 0.01055908203125, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 8.021329879760742, + "learning_rate": 1e-06, + "loss": 0.4143, + "mean_token_accuracy": 0.8621190786361694, + "num_tokens": 171368637.0, + "step": 4410 + }, + { + "epoch": 0.5611245388627402, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.028711318969727, + "learning_rate": 1e-06, + "loss": 0.4191, + "mean_token_accuracy": 0.8593666553497314, + "num_tokens": 171407133.0, + "step": 4411 + }, + { + "epoch": 0.5612517491413306, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.907261371612549, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.8737175464630127, + "num_tokens": 171449865.0, + "step": 4412 + }, + { + "epoch": 0.5613789594199211, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.9610772132873535, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8724497556686401, + "num_tokens": 171492330.0, + "step": 4413 + }, + { + "epoch": 0.5615061696985116, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9974236488342285, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8776019215583801, + "num_tokens": 171528426.0, + "step": 4414 + }, + { + "epoch": 0.5616333799771022, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.986217498779297, + "learning_rate": 1e-06, + "loss": 0.4262, + "mean_token_accuracy": 0.8592308163642883, + "num_tokens": 171572767.0, + "step": 4415 + }, + { + "epoch": 0.5617605902556927, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.930437088012695, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8786103129386902, + "num_tokens": 171613339.0, + "step": 4416 + }, + { + "epoch": 0.5618878005342832, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.049765586853027, + "learning_rate": 1e-06, + "loss": 0.4141, + "mean_token_accuracy": 0.8602213859558105, + "num_tokens": 171648170.0, + "step": 4417 + }, + { + "epoch": 0.5620150108128736, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.89994478225708, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.880605936050415, + "num_tokens": 171687933.0, + "step": 4418 + }, + { + "epoch": 0.5621422210914642, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.937163829803467, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8765649795532227, + "num_tokens": 171728461.0, + "step": 4419 + }, + { + "epoch": 0.5622694313700547, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.036649703979492, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.884704053401947, + "num_tokens": 171763459.0, + "step": 4420 + }, + { + "epoch": 0.5623966416486452, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.953649997711182, + "learning_rate": 1e-06, + "loss": 0.3935, + "mean_token_accuracy": 0.8699920177459717, + "num_tokens": 171804606.0, + "step": 4421 + }, + { + "epoch": 0.5625238519272358, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.882442951202393, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.8752764463424683, + "num_tokens": 171848969.0, + "step": 4422 + }, + { + "epoch": 0.5626510622058263, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.014636993408203e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.003622055053711, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.8805904388427734, + "num_tokens": 171884565.0, + "step": 4423 + }, + { + "epoch": 0.5627782724844167, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.939363956451416, + "learning_rate": 1e-06, + "loss": 0.39, + "mean_token_accuracy": 0.8683943152427673, + "num_tokens": 171930210.0, + "step": 4424 + }, + { + "epoch": 0.5629054827630072, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.054628372192383, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.878987193107605, + "num_tokens": 171966248.0, + "step": 4425 + }, + { + "epoch": 0.5630326930415978, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.986850738525391, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.8743127584457397, + "num_tokens": 172002804.0, + "step": 4426 + }, + { + "epoch": 0.5631599033201883, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.975341796875, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8849319219589233, + "num_tokens": 172041059.0, + "step": 4427 + }, + { + "epoch": 0.5632871135987788, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.913143634796143, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.876362681388855, + "num_tokens": 172085168.0, + "step": 4428 + }, + { + "epoch": 0.5634143238773693, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.939803123474121, + "learning_rate": 1e-06, + "loss": 0.3753, + "mean_token_accuracy": 0.8696414828300476, + "num_tokens": 172125529.0, + "step": 4429 + }, + { + "epoch": 0.5635415341559598, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.109489440917969, + "learning_rate": 1e-06, + "loss": 0.3864, + "mean_token_accuracy": 0.8707476854324341, + "num_tokens": 172161637.0, + "step": 4430 + }, + { + "epoch": 0.5636687444345503, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.012726783752441, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8751553893089294, + "num_tokens": 172197206.0, + "step": 4431 + }, + { + "epoch": 0.5637959547131408, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.042664527893066, + "learning_rate": 1e-06, + "loss": 0.3902, + "mean_token_accuracy": 0.8667764067649841, + "num_tokens": 172233369.0, + "step": 4432 + }, + { + "epoch": 0.5639231649917313, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.955655574798584, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.88469398021698, + "num_tokens": 172266989.0, + "step": 4433 + }, + { + "epoch": 0.5640503752703219, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.848803997039795, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8781470060348511, + "num_tokens": 172310129.0, + "step": 4434 + }, + { + "epoch": 0.5641775855489124, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.00279712677002, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8741408586502075, + "num_tokens": 172347835.0, + "step": 4435 + }, + { + "epoch": 0.5643047958275028, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.016366004943848, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8767037391662598, + "num_tokens": 172381388.0, + "step": 4436 + }, + { + "epoch": 0.5644320061060933, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.006235122680664, + "learning_rate": 1e-06, + "loss": 0.3775, + "mean_token_accuracy": 0.8730393052101135, + "num_tokens": 172416445.0, + "step": 4437 + }, + { + "epoch": 0.5645592163846839, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.871444225311279, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8898013830184937, + "num_tokens": 172456191.0, + "step": 4438 + }, + { + "epoch": 0.5646864266632744, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.032835960388184, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8777272701263428, + "num_tokens": 172490270.0, + "step": 4439 + }, + { + "epoch": 0.5648136369418649, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.931750774383545, + "learning_rate": 1e-06, + "loss": 0.3776, + "mean_token_accuracy": 0.872631311416626, + "num_tokens": 172531749.0, + "step": 4440 + }, + { + "epoch": 0.5649408472204555, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.007918357849121, + "learning_rate": 1e-06, + "loss": 0.3986, + "mean_token_accuracy": 0.8667974472045898, + "num_tokens": 172573598.0, + "step": 4441 + }, + { + "epoch": 0.5650680574990459, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9105000495910645, + "learning_rate": 1e-06, + "loss": 0.361, + "mean_token_accuracy": 0.8776646852493286, + "num_tokens": 172614485.0, + "step": 4442 + }, + { + "epoch": 0.5651952677776364, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.974862575531006, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8732543587684631, + "num_tokens": 172650249.0, + "step": 4443 + }, + { + "epoch": 0.5653224780562269, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.07666015625, + "learning_rate": 1e-06, + "loss": 0.4087, + "mean_token_accuracy": 0.8660004138946533, + "num_tokens": 172685562.0, + "step": 4444 + }, + { + "epoch": 0.5654496883348175, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.917486190795898, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8837239742279053, + "num_tokens": 172728025.0, + "step": 4445 + }, + { + "epoch": 0.565576898613408, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.945337295532227, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.869560956954956, + "num_tokens": 172765800.0, + "step": 4446 + }, + { + "epoch": 0.5657041088919985, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.938144683837891, + "learning_rate": 1e-06, + "loss": 0.394, + "mean_token_accuracy": 0.8721556067466736, + "num_tokens": 172810014.0, + "step": 4447 + }, + { + "epoch": 0.5658313191705889, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.126581192016602, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8715512752532959, + "num_tokens": 172840874.0, + "step": 4448 + }, + { + "epoch": 0.5659585294491795, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9867377281188965, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8803350329399109, + "num_tokens": 172882146.0, + "step": 4449 + }, + { + "epoch": 0.56608573972777, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.990645885467529, + "learning_rate": 1e-06, + "loss": 0.4271, + "mean_token_accuracy": 0.8561776876449585, + "num_tokens": 172921002.0, + "step": 4450 + }, + { + "epoch": 0.5662129500063605, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.068939208984375, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8776990175247192, + "num_tokens": 172949491.0, + "step": 4451 + }, + { + "epoch": 0.566340160284951, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.93231725692749, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8885570764541626, + "num_tokens": 172987044.0, + "step": 4452 + }, + { + "epoch": 0.5664673705635416, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.017624855041504, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.872039258480072, + "num_tokens": 173024654.0, + "step": 4453 + }, + { + "epoch": 0.566594580842132, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.022200584411621, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8835285902023315, + "num_tokens": 173059448.0, + "step": 4454 + }, + { + "epoch": 0.5667217911207225, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.904620170593262, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8928358554840088, + "num_tokens": 173101870.0, + "step": 4455 + }, + { + "epoch": 0.566849001399313, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.959543228149414, + "learning_rate": 1e-06, + "loss": 0.3927, + "mean_token_accuracy": 0.8650080561637878, + "num_tokens": 173142512.0, + "step": 4456 + }, + { + "epoch": 0.5669762116779036, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.985082626342773, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8772227764129639, + "num_tokens": 173182483.0, + "step": 4457 + }, + { + "epoch": 0.5671034219564941, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.072728157043457, + "learning_rate": 1e-06, + "loss": 0.3774, + "mean_token_accuracy": 0.8721345663070679, + "num_tokens": 173218252.0, + "step": 4458 + }, + { + "epoch": 0.5672306322350846, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.935662746429443, + "learning_rate": 1e-06, + "loss": 0.3917, + "mean_token_accuracy": 0.8657180666923523, + "num_tokens": 173261072.0, + "step": 4459 + }, + { + "epoch": 0.5673578425136752, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.946972370147705, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8972086310386658, + "num_tokens": 173297674.0, + "step": 4460 + }, + { + "epoch": 0.5674850527922656, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.945685863494873, + "learning_rate": 1e-06, + "loss": 0.3875, + "mean_token_accuracy": 0.8662392497062683, + "num_tokens": 173342899.0, + "step": 4461 + }, + { + "epoch": 0.5676122630708561, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.024717330932617, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8781173229217529, + "num_tokens": 173379194.0, + "step": 4462 + }, + { + "epoch": 0.5677394733494466, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.932718276977539, + "learning_rate": 1e-06, + "loss": 0.3856, + "mean_token_accuracy": 0.8737905621528625, + "num_tokens": 173421415.0, + "step": 4463 + }, + { + "epoch": 0.5678666836280372, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.98169469833374, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8802381753921509, + "num_tokens": 173460392.0, + "step": 4464 + }, + { + "epoch": 0.5679938939066277, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.078901290893555, + "learning_rate": 1e-06, + "loss": 0.3784, + "mean_token_accuracy": 0.8729609847068787, + "num_tokens": 173493088.0, + "step": 4465 + }, + { + "epoch": 0.5681211041852182, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.934296131134033, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.892891526222229, + "num_tokens": 173530602.0, + "step": 4466 + }, + { + "epoch": 0.5682483144638086, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.039278030395508, + "learning_rate": 1e-06, + "loss": 0.4066, + "mean_token_accuracy": 0.8601609468460083, + "num_tokens": 173567774.0, + "step": 4467 + }, + { + "epoch": 0.5683755247423992, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.020386695861816, + "learning_rate": 1e-06, + "loss": 0.3867, + "mean_token_accuracy": 0.8704458475112915, + "num_tokens": 173607471.0, + "step": 4468 + }, + { + "epoch": 0.5685027350209897, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0384788513183594e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.90574836730957, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8936715126037598, + "num_tokens": 173646979.0, + "step": 4469 + }, + { + "epoch": 0.5686299452995802, + "ewc_loss": 0.0106201171875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 7.931572437286377, + "learning_rate": 1e-06, + "loss": 0.3754, + "mean_token_accuracy": 0.8731412887573242, + "num_tokens": 173692012.0, + "step": 4470 + }, + { + "epoch": 0.5687571555781707, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.056788444519043, + "learning_rate": 1e-06, + "loss": 0.3997, + "mean_token_accuracy": 0.859657883644104, + "num_tokens": 173731769.0, + "step": 4471 + }, + { + "epoch": 0.5688843658567613, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.954000473022461, + "learning_rate": 1e-06, + "loss": 0.3894, + "mean_token_accuracy": 0.868407130241394, + "num_tokens": 173774115.0, + "step": 4472 + }, + { + "epoch": 0.5690115761353517, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.929381370544434, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8760330677032471, + "num_tokens": 173813058.0, + "step": 4473 + }, + { + "epoch": 0.5691387864139422, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.039786338806152, + "learning_rate": 1e-06, + "loss": 0.4036, + "mean_token_accuracy": 0.8634544610977173, + "num_tokens": 173855783.0, + "step": 4474 + }, + { + "epoch": 0.5692659966925327, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9237189292907715, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8752286434173584, + "num_tokens": 173895393.0, + "step": 4475 + }, + { + "epoch": 0.5693932069711233, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.027952194213867, + "learning_rate": 1e-06, + "loss": 0.4054, + "mean_token_accuracy": 0.8648691177368164, + "num_tokens": 173933520.0, + "step": 4476 + }, + { + "epoch": 0.5695204172497138, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.013930320739746, + "learning_rate": 1e-06, + "loss": 0.4005, + "mean_token_accuracy": 0.865831732749939, + "num_tokens": 173970403.0, + "step": 4477 + }, + { + "epoch": 0.5696476275283043, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.935489654541016, + "learning_rate": 1e-06, + "loss": 0.3994, + "mean_token_accuracy": 0.8655257225036621, + "num_tokens": 174012236.0, + "step": 4478 + }, + { + "epoch": 0.5697748378068948, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.9896416664123535, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8707269430160522, + "num_tokens": 174048578.0, + "step": 4479 + }, + { + "epoch": 0.5699020480854853, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.946043014526367, + "learning_rate": 1e-06, + "loss": 0.4245, + "mean_token_accuracy": 0.8567523956298828, + "num_tokens": 174089682.0, + "step": 4480 + }, + { + "epoch": 0.5700292583640758, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.06611156463623, + "learning_rate": 1e-06, + "loss": 0.4226, + "mean_token_accuracy": 0.8552924394607544, + "num_tokens": 174127870.0, + "step": 4481 + }, + { + "epoch": 0.5701564686426663, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.013450622558594, + "learning_rate": 1e-06, + "loss": 0.4026, + "mean_token_accuracy": 0.8668249845504761, + "num_tokens": 174165286.0, + "step": 4482 + }, + { + "epoch": 0.5702836789212569, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.991492748260498, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8698740005493164, + "num_tokens": 174203180.0, + "step": 4483 + }, + { + "epoch": 0.5704108891998474, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.998880863189697, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8795861005783081, + "num_tokens": 174237262.0, + "step": 4484 + }, + { + "epoch": 0.5705380994784378, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.934103488922119, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.880273699760437, + "num_tokens": 174276627.0, + "step": 4485 + }, + { + "epoch": 0.5706653097570283, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.901753902435303, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.878251314163208, + "num_tokens": 174319386.0, + "step": 4486 + }, + { + "epoch": 0.5707925200356189, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.995965957641602, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8712090253829956, + "num_tokens": 174358848.0, + "step": 4487 + }, + { + "epoch": 0.5709197303142094, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.087807655334473, + "learning_rate": 1e-06, + "loss": 0.4073, + "mean_token_accuracy": 0.8677990436553955, + "num_tokens": 174395042.0, + "step": 4488 + }, + { + "epoch": 0.5710469405927999, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.079496383666992, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8850851058959961, + "num_tokens": 174426876.0, + "step": 4489 + }, + { + "epoch": 0.5711741508713905, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.969853401184082, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.883132815361023, + "num_tokens": 174463376.0, + "step": 4490 + }, + { + "epoch": 0.5713013611499809, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.95561408996582, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8829431533813477, + "num_tokens": 174503448.0, + "step": 4491 + }, + { + "epoch": 0.5714285714285714, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.948230743408203, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.868175208568573, + "num_tokens": 174546612.0, + "step": 4492 + }, + { + "epoch": 0.5715557817071619, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.96356725692749, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.879989504814148, + "num_tokens": 174585530.0, + "step": 4493 + }, + { + "epoch": 0.5716829919857525, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.999988555908203, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8830261826515198, + "num_tokens": 174620346.0, + "step": 4494 + }, + { + "epoch": 0.571810202264343, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0503997802734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.026456832885742, + "learning_rate": 1e-06, + "loss": 0.4166, + "mean_token_accuracy": 0.8600210547447205, + "num_tokens": 174662407.0, + "step": 4495 + }, + { + "epoch": 0.5719374125429335, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.934878349304199, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8774702548980713, + "num_tokens": 174704605.0, + "step": 4496 + }, + { + "epoch": 0.5720646228215239, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.931704044342041, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.882382333278656, + "num_tokens": 174743555.0, + "step": 4497 + }, + { + "epoch": 0.5721918331001145, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.05963134765625, + "learning_rate": 1e-06, + "loss": 0.4017, + "mean_token_accuracy": 0.866215705871582, + "num_tokens": 174780511.0, + "step": 4498 + }, + { + "epoch": 0.572319043378705, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.165018081665039, + "learning_rate": 1e-06, + "loss": 0.3848, + "mean_token_accuracy": 0.868963360786438, + "num_tokens": 174814359.0, + "step": 4499 + }, + { + "epoch": 0.5724462536572955, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.051511764526367, + "learning_rate": 1e-06, + "loss": 0.3858, + "mean_token_accuracy": 0.8712564706802368, + "num_tokens": 174851763.0, + "step": 4500 + }, + { + "epoch": 0.572573463935886, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.895502090454102, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.8767532110214233, + "num_tokens": 174894324.0, + "step": 4501 + }, + { + "epoch": 0.5727006742144766, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.973534107208252, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.8735498189926147, + "num_tokens": 174930973.0, + "step": 4502 + }, + { + "epoch": 0.572827884493067, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.945955753326416, + "learning_rate": 1e-06, + "loss": 0.389, + "mean_token_accuracy": 0.8723586797714233, + "num_tokens": 174973977.0, + "step": 4503 + }, + { + "epoch": 0.5729550947716575, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.038968086242676, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8944318294525146, + "num_tokens": 175007942.0, + "step": 4504 + }, + { + "epoch": 0.573082305050248, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.951721668243408, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8779321908950806, + "num_tokens": 175047202.0, + "step": 4505 + }, + { + "epoch": 0.5732095153288386, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.05411434173584, + "learning_rate": 1e-06, + "loss": 0.4195, + "mean_token_accuracy": 0.860683798789978, + "num_tokens": 175087033.0, + "step": 4506 + }, + { + "epoch": 0.5733367256074291, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.932683944702148, + "learning_rate": 1e-06, + "loss": 0.3843, + "mean_token_accuracy": 0.8731249570846558, + "num_tokens": 175129975.0, + "step": 4507 + }, + { + "epoch": 0.5734639358860196, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.939186096191406, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8731117844581604, + "num_tokens": 175171225.0, + "step": 4508 + }, + { + "epoch": 0.5735911461646102, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.944090366363525, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8872904777526855, + "num_tokens": 175209577.0, + "step": 4509 + }, + { + "epoch": 0.5737183564432006, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.036077499389648, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8698244094848633, + "num_tokens": 175246126.0, + "step": 4510 + }, + { + "epoch": 0.5738455667217911, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.02037239074707, + "learning_rate": 1e-06, + "loss": 0.3846, + "mean_token_accuracy": 0.8694601655006409, + "num_tokens": 175282735.0, + "step": 4511 + }, + { + "epoch": 0.5739727770003816, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.08283805847168, + "learning_rate": 1e-06, + "loss": 0.3912, + "mean_token_accuracy": 0.8674443960189819, + "num_tokens": 175315382.0, + "step": 4512 + }, + { + "epoch": 0.5740999872789722, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.008210182189941, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8849440217018127, + "num_tokens": 175349599.0, + "step": 4513 + }, + { + "epoch": 0.5742271975575627, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.017241477966309, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.872341513633728, + "num_tokens": 175387723.0, + "step": 4514 + }, + { + "epoch": 0.5743544078361532, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.056249618530273, + "learning_rate": 1e-06, + "loss": 0.4422, + "mean_token_accuracy": 0.8534277677536011, + "num_tokens": 175430967.0, + "step": 4515 + }, + { + "epoch": 0.5744816181147436, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.944748401641846, + "learning_rate": 1e-06, + "loss": 0.3818, + "mean_token_accuracy": 0.874343991279602, + "num_tokens": 175470454.0, + "step": 4516 + }, + { + "epoch": 0.5746088283933342, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.945521831512451, + "learning_rate": 1e-06, + "loss": 0.3861, + "mean_token_accuracy": 0.8719624280929565, + "num_tokens": 175513608.0, + "step": 4517 + }, + { + "epoch": 0.5747360386719247, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.01953125, + "learning_rate": 1e-06, + "loss": 0.3963, + "mean_token_accuracy": 0.8687710762023926, + "num_tokens": 175553611.0, + "step": 4518 + }, + { + "epoch": 0.5748632489505152, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.891661167144775, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8863674402236938, + "num_tokens": 175598882.0, + "step": 4519 + }, + { + "epoch": 0.5749904592291057, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.078250885009766, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.874391496181488, + "num_tokens": 175637534.0, + "step": 4520 + }, + { + "epoch": 0.5751176695076963, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.947300910949707, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8792535066604614, + "num_tokens": 175681134.0, + "step": 4521 + }, + { + "epoch": 0.5752448797862867, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.04066276550293, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8836146593093872, + "num_tokens": 175717223.0, + "step": 4522 + }, + { + "epoch": 0.5753720900648772, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0623207092285156e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.083683967590332, + "learning_rate": 1e-06, + "loss": 0.4433, + "mean_token_accuracy": 0.854212760925293, + "num_tokens": 175753877.0, + "step": 4523 + }, + { + "epoch": 0.5754993003434677, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.087825775146484, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8746665120124817, + "num_tokens": 175791099.0, + "step": 4524 + }, + { + "epoch": 0.5756265106220583, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.069060325622559, + "learning_rate": 1e-06, + "loss": 0.4092, + "mean_token_accuracy": 0.8620142936706543, + "num_tokens": 175826776.0, + "step": 4525 + }, + { + "epoch": 0.5757537209006488, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.990672588348389, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8812222480773926, + "num_tokens": 175865066.0, + "step": 4526 + }, + { + "epoch": 0.5758809311792393, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.0289306640625, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8784663081169128, + "num_tokens": 175899801.0, + "step": 4527 + }, + { + "epoch": 0.5760081414578297, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.067805290222168, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8746688365936279, + "num_tokens": 175935707.0, + "step": 4528 + }, + { + "epoch": 0.5761353517364203, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.98980712890625, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.875461995601654, + "num_tokens": 175973747.0, + "step": 4529 + }, + { + "epoch": 0.5762625620150108, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.108267784118652, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8773728609085083, + "num_tokens": 176004868.0, + "step": 4530 + }, + { + "epoch": 0.5763897722936013, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.150524139404297, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8774979710578918, + "num_tokens": 176037669.0, + "step": 4531 + }, + { + "epoch": 0.5765169825721919, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9459547996521, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8830617070198059, + "num_tokens": 176075503.0, + "step": 4532 + }, + { + "epoch": 0.5766441928507824, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.056429862976074, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.868493914604187, + "num_tokens": 176115278.0, + "step": 4533 + }, + { + "epoch": 0.5767714031293728, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.029414176940918, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8813223838806152, + "num_tokens": 176153347.0, + "step": 4534 + }, + { + "epoch": 0.5768986134079633, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.995749473571777, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8821409940719604, + "num_tokens": 176187730.0, + "step": 4535 + }, + { + "epoch": 0.5770258236865539, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.971166133880615, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.8717674016952515, + "num_tokens": 176230484.0, + "step": 4536 + }, + { + "epoch": 0.5771530339651444, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.939131259918213, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8762050867080688, + "num_tokens": 176276704.0, + "step": 4537 + }, + { + "epoch": 0.5772802442437349, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.071609497070312, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8839348554611206, + "num_tokens": 176315288.0, + "step": 4538 + }, + { + "epoch": 0.5774074545223254, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.971636772155762, + "learning_rate": 1e-06, + "loss": 0.4153, + "mean_token_accuracy": 0.8608699440956116, + "num_tokens": 176354484.0, + "step": 4539 + }, + { + "epoch": 0.5775346648009159, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.126307487487793, + "learning_rate": 1e-06, + "loss": 0.4023, + "mean_token_accuracy": 0.8615206480026245, + "num_tokens": 176388631.0, + "step": 4540 + }, + { + "epoch": 0.5776618750795064, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.966157913208008, + "learning_rate": 1e-06, + "loss": 0.4076, + "mean_token_accuracy": 0.8650360107421875, + "num_tokens": 176428112.0, + "step": 4541 + }, + { + "epoch": 0.5777890853580969, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.031108856201172, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.874382495880127, + "num_tokens": 176468846.0, + "step": 4542 + }, + { + "epoch": 0.5779162956366874, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.944150924682617, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8728939294815063, + "num_tokens": 176505948.0, + "step": 4543 + }, + { + "epoch": 0.578043505915278, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.953726768493652, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8781037330627441, + "num_tokens": 176549101.0, + "step": 4544 + }, + { + "epoch": 0.5781707161938685, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.0742416381835938e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.048808097839355, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8777828216552734, + "num_tokens": 176583060.0, + "step": 4545 + }, + { + "epoch": 0.5782979264724589, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.911032676696777, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8795176148414612, + "num_tokens": 176623422.0, + "step": 4546 + }, + { + "epoch": 0.5784251367510495, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.00692367553711, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.8725886940956116, + "num_tokens": 176661835.0, + "step": 4547 + }, + { + "epoch": 0.57855234702964, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.994541645050049, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8838905096054077, + "num_tokens": 176703546.0, + "step": 4548 + }, + { + "epoch": 0.5786795573082305, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.066746711730957, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8777222633361816, + "num_tokens": 176737977.0, + "step": 4549 + }, + { + "epoch": 0.578806767586821, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.99341344833374, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8797342777252197, + "num_tokens": 176775822.0, + "step": 4550 + }, + { + "epoch": 0.5789339778654116, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.049943923950195, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8784160614013672, + "num_tokens": 176811046.0, + "step": 4551 + }, + { + "epoch": 0.579061188144002, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.974851608276367, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.8770493268966675, + "num_tokens": 176848822.0, + "step": 4552 + }, + { + "epoch": 0.5791883984225925, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.967172622680664, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8800544738769531, + "num_tokens": 176885181.0, + "step": 4553 + }, + { + "epoch": 0.579315608701183, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.997666358947754, + "learning_rate": 1e-06, + "loss": 0.3905, + "mean_token_accuracy": 0.8670432567596436, + "num_tokens": 176925322.0, + "step": 4554 + }, + { + "epoch": 0.5794428189797736, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.976038455963135, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8792222738265991, + "num_tokens": 176965882.0, + "step": 4555 + }, + { + "epoch": 0.5795700292583641, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.000794410705566, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8860799670219421, + "num_tokens": 177004888.0, + "step": 4556 + }, + { + "epoch": 0.5796972395369546, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.970142364501953, + "learning_rate": 1e-06, + "loss": 0.3926, + "mean_token_accuracy": 0.8655174374580383, + "num_tokens": 177045244.0, + "step": 4557 + }, + { + "epoch": 0.5798244498155452, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.246797561645508, + "learning_rate": 1e-06, + "loss": 0.4422, + "mean_token_accuracy": 0.8542421460151672, + "num_tokens": 177075557.0, + "step": 4558 + }, + { + "epoch": 0.5799516600941356, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.96626091003418, + "learning_rate": 1e-06, + "loss": 0.3899, + "mean_token_accuracy": 0.8691201210021973, + "num_tokens": 177114908.0, + "step": 4559 + }, + { + "epoch": 0.5800788703727261, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.184266090393066, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.8682857751846313, + "num_tokens": 177147615.0, + "step": 4560 + }, + { + "epoch": 0.5802060806513166, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 7.993433952331543, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.8715705871582031, + "num_tokens": 177193500.0, + "step": 4561 + }, + { + "epoch": 0.5803332909299072, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.953047752380371, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8775099515914917, + "num_tokens": 177231355.0, + "step": 4562 + }, + { + "epoch": 0.5804605012084977, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.086162567138672e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.170096397399902, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8788242340087891, + "num_tokens": 177261901.0, + "step": 4563 + }, + { + "epoch": 0.5805877114870882, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.996665954589844, + "learning_rate": 1e-06, + "loss": 0.3957, + "mean_token_accuracy": 0.8675955533981323, + "num_tokens": 177302396.0, + "step": 4564 + }, + { + "epoch": 0.5807149217656786, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.002908706665039, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.8738935589790344, + "num_tokens": 177342793.0, + "step": 4565 + }, + { + "epoch": 0.5808421320442692, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.051785469055176, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8795754909515381, + "num_tokens": 177377783.0, + "step": 4566 + }, + { + "epoch": 0.5809693423228597, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.929582595825195, + "learning_rate": 1e-06, + "loss": 0.3859, + "mean_token_accuracy": 0.8743889927864075, + "num_tokens": 177421353.0, + "step": 4567 + }, + { + "epoch": 0.5810965526014502, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.03756046295166, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8770005106925964, + "num_tokens": 177459514.0, + "step": 4568 + }, + { + "epoch": 0.5812237628800407, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.008520126342773, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8758975863456726, + "num_tokens": 177492720.0, + "step": 4569 + }, + { + "epoch": 0.5813509731586313, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.97974157333374, + "learning_rate": 1e-06, + "loss": 0.3952, + "mean_token_accuracy": 0.8670013546943665, + "num_tokens": 177534292.0, + "step": 4570 + }, + { + "epoch": 0.5814781834372217, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.031126022338867, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.875009298324585, + "num_tokens": 177571477.0, + "step": 4571 + }, + { + "epoch": 0.5816053937158122, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.9917707443237305, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8786470293998718, + "num_tokens": 177606092.0, + "step": 4572 + }, + { + "epoch": 0.5817326039944027, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.985296249389648, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8804666996002197, + "num_tokens": 177644430.0, + "step": 4573 + }, + { + "epoch": 0.5818598142729933, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.14258861541748, + "learning_rate": 1e-06, + "loss": 0.4078, + "mean_token_accuracy": 0.861274242401123, + "num_tokens": 177678215.0, + "step": 4574 + }, + { + "epoch": 0.5819870245515838, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.064306259155273, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8695036768913269, + "num_tokens": 177714287.0, + "step": 4575 + }, + { + "epoch": 0.5821142348301743, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.98058557510376, + "learning_rate": 1e-06, + "loss": 0.4002, + "mean_token_accuracy": 0.8671327829360962, + "num_tokens": 177758549.0, + "step": 4576 + }, + { + "epoch": 0.5822414451087647, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.020946502685547, + "learning_rate": 1e-06, + "loss": 0.3874, + "mean_token_accuracy": 0.8671638369560242, + "num_tokens": 177796556.0, + "step": 4577 + }, + { + "epoch": 0.5823686553873553, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.179428100585938, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.8681585192680359, + "num_tokens": 177827534.0, + "step": 4578 + }, + { + "epoch": 0.5824958656659458, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 7.995913028717041, + "learning_rate": 1e-06, + "loss": 0.4075, + "mean_token_accuracy": 0.8651441931724548, + "num_tokens": 177868069.0, + "step": 4579 + }, + { + "epoch": 0.5826230759445363, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.929964542388916, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8770707249641418, + "num_tokens": 177910904.0, + "step": 4580 + }, + { + "epoch": 0.5827502862231269, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.115666389465332, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8706979751586914, + "num_tokens": 177946125.0, + "step": 4581 + }, + { + "epoch": 0.5828774965017174, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.991345405578613, + "learning_rate": 1e-06, + "loss": 0.3576, + "mean_token_accuracy": 0.8773753046989441, + "num_tokens": 177982440.0, + "step": 4582 + }, + { + "epoch": 0.5830047067803078, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.142924308776855, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.8660521507263184, + "num_tokens": 178018516.0, + "step": 4583 + }, + { + "epoch": 0.5831319170588983, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 7.920010566711426, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8745006322860718, + "num_tokens": 178063116.0, + "step": 4584 + }, + { + "epoch": 0.5832591273374889, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.011975288391113, + "learning_rate": 1e-06, + "loss": 0.3874, + "mean_token_accuracy": 0.8700341582298279, + "num_tokens": 178105779.0, + "step": 4585 + }, + { + "epoch": 0.5833863376160794, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.04146957397461, + "learning_rate": 1e-06, + "loss": 0.4064, + "mean_token_accuracy": 0.8620098829269409, + "num_tokens": 178151049.0, + "step": 4586 + }, + { + "epoch": 0.5835135478946699, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.044540405273438, + "learning_rate": 1e-06, + "loss": 0.3903, + "mean_token_accuracy": 0.8702892661094666, + "num_tokens": 178191677.0, + "step": 4587 + }, + { + "epoch": 0.5836407581732604, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.016254425048828, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.873679518699646, + "num_tokens": 178232345.0, + "step": 4588 + }, + { + "epoch": 0.5837679684518509, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.037344932556152, + "learning_rate": 1e-06, + "loss": 0.42, + "mean_token_accuracy": 0.8593425750732422, + "num_tokens": 178275960.0, + "step": 4589 + }, + { + "epoch": 0.5838951787304414, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.017157554626465, + "learning_rate": 1e-06, + "loss": 0.4088, + "mean_token_accuracy": 0.8675718307495117, + "num_tokens": 178320575.0, + "step": 4590 + }, + { + "epoch": 0.5840223890090319, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.150842666625977, + "learning_rate": 1e-06, + "loss": 0.3815, + "mean_token_accuracy": 0.873510479927063, + "num_tokens": 178351417.0, + "step": 4591 + }, + { + "epoch": 0.5841495992876224, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.08879566192627, + "learning_rate": 1e-06, + "loss": 0.385, + "mean_token_accuracy": 0.8651784658432007, + "num_tokens": 178388516.0, + "step": 4592 + }, + { + "epoch": 0.584276809566213, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.919909954071045, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.8677881956100464, + "num_tokens": 178434717.0, + "step": 4593 + }, + { + "epoch": 0.5844040198448035, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.019889831542969, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.887050211429596, + "num_tokens": 178471175.0, + "step": 4594 + }, + { + "epoch": 0.5845312301233939, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.9909162521362305, + "learning_rate": 1e-06, + "loss": 0.4078, + "mean_token_accuracy": 0.8663588166236877, + "num_tokens": 178512343.0, + "step": 4595 + }, + { + "epoch": 0.5846584404019844, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.056053161621094, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8767911195755005, + "num_tokens": 178548711.0, + "step": 4596 + }, + { + "epoch": 0.584785650680575, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.121652603149414, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.881249189376831, + "num_tokens": 178581674.0, + "step": 4597 + }, + { + "epoch": 0.5849128609591655, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.004095077514648, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8773828148841858, + "num_tokens": 178626964.0, + "step": 4598 + }, + { + "epoch": 0.585040071237756, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.110004425048828e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.003637313842773, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8777414560317993, + "num_tokens": 178669895.0, + "step": 4599 + }, + { + "epoch": 0.5851672815163466, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1219253540039062e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.019684791564941, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8746719360351562, + "num_tokens": 178712424.0, + "step": 4600 + }, + { + "epoch": 0.585294491794937, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1219253540039062e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.084688186645508, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8881226778030396, + "num_tokens": 178743547.0, + "step": 4601 + }, + { + "epoch": 0.5854217020735275, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1219253540039062e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.038954734802246, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8756569027900696, + "num_tokens": 178783702.0, + "step": 4602 + }, + { + "epoch": 0.585548912352118, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.1219253540039062e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.056140899658203, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8825695514678955, + "num_tokens": 178819640.0, + "step": 4603 + }, + { + "epoch": 0.5856761226307086, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1219253540039062e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.02483081817627, + "learning_rate": 1e-06, + "loss": 0.3964, + "mean_token_accuracy": 0.8665416836738586, + "num_tokens": 178861088.0, + "step": 4604 + }, + { + "epoch": 0.5858033329092991, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.133930206298828, + "learning_rate": 1e-06, + "loss": 0.4156, + "mean_token_accuracy": 0.8584935665130615, + "num_tokens": 178897216.0, + "step": 4605 + }, + { + "epoch": 0.5859305431878896, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.098904609680176, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8723627328872681, + "num_tokens": 178931018.0, + "step": 4606 + }, + { + "epoch": 0.5860577534664801, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.117029190063477, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.8662718534469604, + "num_tokens": 178968646.0, + "step": 4607 + }, + { + "epoch": 0.5861849637450706, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.00183391571045, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8889355063438416, + "num_tokens": 179004276.0, + "step": 4608 + }, + { + "epoch": 0.5863121740236611, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.996130466461182, + "learning_rate": 1e-06, + "loss": 0.3879, + "mean_token_accuracy": 0.8697640895843506, + "num_tokens": 179047991.0, + "step": 4609 + }, + { + "epoch": 0.5864393843022516, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.109989166259766, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.88419508934021, + "num_tokens": 179076867.0, + "step": 4610 + }, + { + "epoch": 0.5865665945808421, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.071548461914062, + "learning_rate": 1e-06, + "loss": 0.393, + "mean_token_accuracy": 0.8641518354415894, + "num_tokens": 179114661.0, + "step": 4611 + }, + { + "epoch": 0.5866938048594327, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.902368068695068, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8783480525016785, + "num_tokens": 179161431.0, + "step": 4612 + }, + { + "epoch": 0.5868210151380232, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.041999816894531, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8884122967720032, + "num_tokens": 179200422.0, + "step": 4613 + }, + { + "epoch": 0.5869482254166136, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.977492809295654, + "learning_rate": 1e-06, + "loss": 0.3788, + "mean_token_accuracy": 0.8742420673370361, + "num_tokens": 179240136.0, + "step": 4614 + }, + { + "epoch": 0.5870754356952042, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.010876655578613, + "learning_rate": 1e-06, + "loss": 0.4136, + "mean_token_accuracy": 0.8618399500846863, + "num_tokens": 179287568.0, + "step": 4615 + }, + { + "epoch": 0.5872026459737947, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.019941329956055, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8790050745010376, + "num_tokens": 179325381.0, + "step": 4616 + }, + { + "epoch": 0.5873298562523852, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.009711265563965, + "learning_rate": 1e-06, + "loss": 0.3894, + "mean_token_accuracy": 0.8660615086555481, + "num_tokens": 179369912.0, + "step": 4617 + }, + { + "epoch": 0.5874570665309757, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.030254364013672, + "learning_rate": 1e-06, + "loss": 0.3798, + "mean_token_accuracy": 0.8724485635757446, + "num_tokens": 179409365.0, + "step": 4618 + }, + { + "epoch": 0.5875842768095663, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.116538047790527, + "learning_rate": 1e-06, + "loss": 0.3832, + "mean_token_accuracy": 0.8746923208236694, + "num_tokens": 179445707.0, + "step": 4619 + }, + { + "epoch": 0.5877114870881567, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.941240310668945, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8893812894821167, + "num_tokens": 179484924.0, + "step": 4620 + }, + { + "epoch": 0.5878386973667472, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 7.978880882263184, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8826528787612915, + "num_tokens": 179526118.0, + "step": 4621 + }, + { + "epoch": 0.5879659076453377, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.964979648590088, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8891624808311462, + "num_tokens": 179568727.0, + "step": 4622 + }, + { + "epoch": 0.5880931179239283, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.017024993896484, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8778423070907593, + "num_tokens": 179608174.0, + "step": 4623 + }, + { + "epoch": 0.5882203282025188, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.203848838806152, + "learning_rate": 1e-06, + "loss": 0.407, + "mean_token_accuracy": 0.8621813058853149, + "num_tokens": 179643944.0, + "step": 4624 + }, + { + "epoch": 0.5883475384811093, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.921976566314697, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8903272747993469, + "num_tokens": 179683836.0, + "step": 4625 + }, + { + "epoch": 0.5884747487596997, + "ewc_loss": 0.01068115234375, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 8.163935661315918, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.8733144402503967, + "num_tokens": 179717935.0, + "step": 4626 + }, + { + "epoch": 0.5886019590382903, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.017413139343262, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8779385089874268, + "num_tokens": 179757380.0, + "step": 4627 + }, + { + "epoch": 0.5887291693168808, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.037816047668457, + "learning_rate": 1e-06, + "loss": 0.3904, + "mean_token_accuracy": 0.8649505376815796, + "num_tokens": 179794913.0, + "step": 4628 + }, + { + "epoch": 0.5888563795954713, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.010721206665039, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8708252310752869, + "num_tokens": 179836018.0, + "step": 4629 + }, + { + "epoch": 0.5889835898740619, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.14961051940918, + "learning_rate": 1e-06, + "loss": 0.3942, + "mean_token_accuracy": 0.8698196411132812, + "num_tokens": 179871754.0, + "step": 4630 + }, + { + "epoch": 0.5891108001526524, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.049530982971191, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8746931552886963, + "num_tokens": 179910078.0, + "step": 4631 + }, + { + "epoch": 0.5892380104312428, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.58306884765625e-06, + "grad_norm": 8.090362548828125, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.8670729398727417, + "num_tokens": 179945830.0, + "step": 4632 + }, + { + "epoch": 0.5893652207098333, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.061321258544922, + "learning_rate": 1e-06, + "loss": 0.3906, + "mean_token_accuracy": 0.8668829202651978, + "num_tokens": 179983559.0, + "step": 4633 + }, + { + "epoch": 0.5894924309884239, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.98273229598999, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8915371894836426, + "num_tokens": 180019773.0, + "step": 4634 + }, + { + "epoch": 0.5896196412670144, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.029472351074219, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8759652376174927, + "num_tokens": 180060216.0, + "step": 4635 + }, + { + "epoch": 0.5897468515456049, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.06784439086914, + "learning_rate": 1e-06, + "loss": 0.4001, + "mean_token_accuracy": 0.8640357255935669, + "num_tokens": 180101081.0, + "step": 4636 + }, + { + "epoch": 0.5898740618241954, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.140166282653809, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.880303144454956, + "num_tokens": 180131655.0, + "step": 4637 + }, + { + "epoch": 0.5900012721027859, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.123128890991211, + "learning_rate": 1e-06, + "loss": 0.4091, + "mean_token_accuracy": 0.863387942314148, + "num_tokens": 180166898.0, + "step": 4638 + }, + { + "epoch": 0.5901284823813764, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.01671314239502, + "learning_rate": 1e-06, + "loss": 0.4063, + "mean_token_accuracy": 0.8658455014228821, + "num_tokens": 180207003.0, + "step": 4639 + }, + { + "epoch": 0.5902556926599669, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1338462829589844e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.971221923828125, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8765178918838501, + "num_tokens": 180249786.0, + "step": 4640 + }, + { + "epoch": 0.5903829029385574, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.004897117614746, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8860535025596619, + "num_tokens": 180286407.0, + "step": 4641 + }, + { + "epoch": 0.590510113217148, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.11159610748291, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8818780779838562, + "num_tokens": 180324180.0, + "step": 4642 + }, + { + "epoch": 0.5906373234957385, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.16443157196045, + "learning_rate": 1e-06, + "loss": 0.4, + "mean_token_accuracy": 0.8681483268737793, + "num_tokens": 180370195.0, + "step": 4643 + }, + { + "epoch": 0.5907645337743289, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 7.964672088623047, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8771204352378845, + "num_tokens": 180411331.0, + "step": 4644 + }, + { + "epoch": 0.5908917440529194, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.977095127105713, + "learning_rate": 1e-06, + "loss": 0.4083, + "mean_token_accuracy": 0.8628071546554565, + "num_tokens": 180453923.0, + "step": 4645 + }, + { + "epoch": 0.59101895433151, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.117707252502441, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8765354156494141, + "num_tokens": 180483024.0, + "step": 4646 + }, + { + "epoch": 0.5911461646101005, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.965532302856445, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8726586699485779, + "num_tokens": 180523740.0, + "step": 4647 + }, + { + "epoch": 0.591273374888691, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.01262378692627, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.872564435005188, + "num_tokens": 180563920.0, + "step": 4648 + }, + { + "epoch": 0.5914005851672816, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.902613639831543, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8866558074951172, + "num_tokens": 180606160.0, + "step": 4649 + }, + { + "epoch": 0.591527795445872, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.03201675415039, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.868489682674408, + "num_tokens": 180649370.0, + "step": 4650 + }, + { + "epoch": 0.5916550057244625, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.087112426757812, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8835442662239075, + "num_tokens": 180686523.0, + "step": 4651 + }, + { + "epoch": 0.591782216003053, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.066998481750488, + "learning_rate": 1e-06, + "loss": 0.3888, + "mean_token_accuracy": 0.8702199459075928, + "num_tokens": 180719671.0, + "step": 4652 + }, + { + "epoch": 0.5919094262816436, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.976193904876709, + "learning_rate": 1e-06, + "loss": 0.3761, + "mean_token_accuracy": 0.8665187358856201, + "num_tokens": 180760453.0, + "step": 4653 + }, + { + "epoch": 0.5920366365602341, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.026124000549316, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8735970258712769, + "num_tokens": 180796628.0, + "step": 4654 + }, + { + "epoch": 0.5921638468388246, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9013285636901855, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8825491070747375, + "num_tokens": 180841838.0, + "step": 4655 + }, + { + "epoch": 0.592291057117415, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.037429809570312, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.8767606019973755, + "num_tokens": 180880142.0, + "step": 4656 + }, + { + "epoch": 0.5924182673960056, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.047574043273926, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8822423815727234, + "num_tokens": 180913658.0, + "step": 4657 + }, + { + "epoch": 0.5925454776745961, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.00682258605957, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8775081634521484, + "num_tokens": 180947367.0, + "step": 4658 + }, + { + "epoch": 0.5926726879531866, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.090447425842285, + "learning_rate": 1e-06, + "loss": 0.3901, + "mean_token_accuracy": 0.867236852645874, + "num_tokens": 180983388.0, + "step": 4659 + }, + { + "epoch": 0.5927998982317771, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.010340690612793, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8870829343795776, + "num_tokens": 181019071.0, + "step": 4660 + }, + { + "epoch": 0.5929271085103677, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.003754615783691, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8784072995185852, + "num_tokens": 181061749.0, + "step": 4661 + }, + { + "epoch": 0.5930543187889582, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.035795211791992, + "learning_rate": 1e-06, + "loss": 0.4127, + "mean_token_accuracy": 0.859918475151062, + "num_tokens": 181102799.0, + "step": 4662 + }, + { + "epoch": 0.5931815290675486, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.023719787597656, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8750880360603333, + "num_tokens": 181140680.0, + "step": 4663 + }, + { + "epoch": 0.5933087393461391, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.150884628295898, + "learning_rate": 1e-06, + "loss": 0.3925, + "mean_token_accuracy": 0.8670225739479065, + "num_tokens": 181173966.0, + "step": 4664 + }, + { + "epoch": 0.5934359496247297, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.027833938598633, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8731600046157837, + "num_tokens": 181216867.0, + "step": 4665 + }, + { + "epoch": 0.5935631599033202, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.101513862609863, + "learning_rate": 1e-06, + "loss": 0.3938, + "mean_token_accuracy": 0.8703928589820862, + "num_tokens": 181250432.0, + "step": 4666 + }, + { + "epoch": 0.5936903701819107, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.016304969787598, + "learning_rate": 1e-06, + "loss": 0.3689, + "mean_token_accuracy": 0.8741501569747925, + "num_tokens": 181290956.0, + "step": 4667 + }, + { + "epoch": 0.5938175804605013, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.976744651794434, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8877733945846558, + "num_tokens": 181329170.0, + "step": 4668 + }, + { + "epoch": 0.5939447907390917, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1576881408691406e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.042119979858398, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.8835912346839905, + "num_tokens": 181368825.0, + "step": 4669 + }, + { + "epoch": 0.5940720010176822, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1576881408691406e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.997507095336914, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8738446235656738, + "num_tokens": 181408455.0, + "step": 4670 + }, + { + "epoch": 0.5941992112962727, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1576881408691406e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.132424354553223, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8847173452377319, + "num_tokens": 181440402.0, + "step": 4671 + }, + { + "epoch": 0.5943264215748633, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1576881408691406e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.029932022094727, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8757395148277283, + "num_tokens": 181471931.0, + "step": 4672 + }, + { + "epoch": 0.5944536318534538, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.23222541809082, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.868041455745697, + "num_tokens": 181503460.0, + "step": 4673 + }, + { + "epoch": 0.5945808421320443, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.030195236206055, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8788973689079285, + "num_tokens": 181540294.0, + "step": 4674 + }, + { + "epoch": 0.5947080524106347, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.054807662963867, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8778585195541382, + "num_tokens": 181575249.0, + "step": 4675 + }, + { + "epoch": 0.5948352626892253, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.088166236877441, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8747035264968872, + "num_tokens": 181612523.0, + "step": 4676 + }, + { + "epoch": 0.5949624729678158, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.9862751960754395, + "learning_rate": 1e-06, + "loss": 0.4001, + "mean_token_accuracy": 0.8660503625869751, + "num_tokens": 181657393.0, + "step": 4677 + }, + { + "epoch": 0.5950896832464063, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.007974624633789, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8779484033584595, + "num_tokens": 181697704.0, + "step": 4678 + }, + { + "epoch": 0.5952168935249968, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.113590240478516, + "learning_rate": 1e-06, + "loss": 0.4828, + "mean_token_accuracy": 0.8395373821258545, + "num_tokens": 181740959.0, + "step": 4679 + }, + { + "epoch": 0.5953441038035874, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.003995895385742, + "learning_rate": 1e-06, + "loss": 0.3815, + "mean_token_accuracy": 0.8694329261779785, + "num_tokens": 181782478.0, + "step": 4680 + }, + { + "epoch": 0.5954713140821778, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.004225730895996, + "learning_rate": 1e-06, + "loss": 0.3667, + "mean_token_accuracy": 0.8737694621086121, + "num_tokens": 181824580.0, + "step": 4681 + }, + { + "epoch": 0.5955985243607683, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.115976333618164, + "learning_rate": 1e-06, + "loss": 0.3667, + "mean_token_accuracy": 0.8774168491363525, + "num_tokens": 181859387.0, + "step": 4682 + }, + { + "epoch": 0.5957257346393589, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.992184638977051, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8758225440979004, + "num_tokens": 181900763.0, + "step": 4683 + }, + { + "epoch": 0.5958529449179494, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.073807716369629, + "learning_rate": 1e-06, + "loss": 0.4087, + "mean_token_accuracy": 0.8611664175987244, + "num_tokens": 181942888.0, + "step": 4684 + }, + { + "epoch": 0.5959801551965399, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.010574340820312, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8754922151565552, + "num_tokens": 181983532.0, + "step": 4685 + }, + { + "epoch": 0.5961073654751304, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.042037010192871, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8899876475334167, + "num_tokens": 182018127.0, + "step": 4686 + }, + { + "epoch": 0.5962345757537209, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.049229621887207, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8849446773529053, + "num_tokens": 182053550.0, + "step": 4687 + }, + { + "epoch": 0.5963617860323114, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.975400447845459, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.877760648727417, + "num_tokens": 182098584.0, + "step": 4688 + }, + { + "epoch": 0.5964889963109019, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.989289283752441, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8866527080535889, + "num_tokens": 182141411.0, + "step": 4689 + }, + { + "epoch": 0.5966162065894924, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.122305870056152, + "learning_rate": 1e-06, + "loss": 0.4075, + "mean_token_accuracy": 0.8641906380653381, + "num_tokens": 182180516.0, + "step": 4690 + }, + { + "epoch": 0.596743416868083, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.966342449188232, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8894398212432861, + "num_tokens": 182224222.0, + "step": 4691 + }, + { + "epoch": 0.5968706271466735, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.158087730407715, + "learning_rate": 1e-06, + "loss": 0.4326, + "mean_token_accuracy": 0.852682888507843, + "num_tokens": 182260052.0, + "step": 4692 + }, + { + "epoch": 0.5969978374252639, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.111433029174805, + "learning_rate": 1e-06, + "loss": 0.3902, + "mean_token_accuracy": 0.8680561780929565, + "num_tokens": 182295604.0, + "step": 4693 + }, + { + "epoch": 0.5971250477038544, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.079273223876953, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8758127689361572, + "num_tokens": 182327929.0, + "step": 4694 + }, + { + "epoch": 0.597252257982445, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.014286994934082, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8833936452865601, + "num_tokens": 182361724.0, + "step": 4695 + }, + { + "epoch": 0.5973794682610355, + "ewc_loss": 0.01080322265625, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.11939525604248, + "learning_rate": 1e-06, + "loss": 0.4113, + "mean_token_accuracy": 0.8601174354553223, + "num_tokens": 182398094.0, + "step": 4696 + }, + { + "epoch": 0.597506678539626, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.920832633972168, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8936119079589844, + "num_tokens": 182437664.0, + "step": 4697 + }, + { + "epoch": 0.5976338888182166, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.256515502929688, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8722308874130249, + "num_tokens": 182471626.0, + "step": 4698 + }, + { + "epoch": 0.597761099096807, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.920313358306885, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8796408772468567, + "num_tokens": 182517068.0, + "step": 4699 + }, + { + "epoch": 0.5978883093753975, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1457672119140625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.026164054870605, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8728342056274414, + "num_tokens": 182554838.0, + "step": 4700 + }, + { + "epoch": 0.598015519653988, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.060474395751953, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8784892559051514, + "num_tokens": 182594863.0, + "step": 4701 + }, + { + "epoch": 0.5981427299325786, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.023982048034668, + "learning_rate": 1e-06, + "loss": 0.4159, + "mean_token_accuracy": 0.8619163632392883, + "num_tokens": 182640016.0, + "step": 4702 + }, + { + "epoch": 0.5982699402111691, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.957211017608643, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8860892057418823, + "num_tokens": 182683841.0, + "step": 4703 + }, + { + "epoch": 0.5983971504897596, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.005298614501953, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8861558437347412, + "num_tokens": 182723553.0, + "step": 4704 + }, + { + "epoch": 0.59852436076835, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.974515914916992, + "learning_rate": 1e-06, + "loss": 0.3865, + "mean_token_accuracy": 0.8689634799957275, + "num_tokens": 182769014.0, + "step": 4705 + }, + { + "epoch": 0.5986515710469406, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.13090705871582, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8737213611602783, + "num_tokens": 182801671.0, + "step": 4706 + }, + { + "epoch": 0.5987787813255311, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9963059425354, + "learning_rate": 1e-06, + "loss": 0.3716, + "mean_token_accuracy": 0.8747704029083252, + "num_tokens": 182846497.0, + "step": 4707 + }, + { + "epoch": 0.5989059916041216, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.068549156188965, + "learning_rate": 1e-06, + "loss": 0.4278, + "mean_token_accuracy": 0.8612383008003235, + "num_tokens": 182887547.0, + "step": 4708 + }, + { + "epoch": 0.5990332018827121, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9799323081970215, + "learning_rate": 1e-06, + "loss": 0.3775, + "mean_token_accuracy": 0.8746176958084106, + "num_tokens": 182930034.0, + "step": 4709 + }, + { + "epoch": 0.5991604121613027, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.033249855041504, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8777015805244446, + "num_tokens": 182965715.0, + "step": 4710 + }, + { + "epoch": 0.5992876224398932, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.169677734375, + "learning_rate": 1e-06, + "loss": 0.4292, + "mean_token_accuracy": 0.8562221527099609, + "num_tokens": 183000069.0, + "step": 4711 + }, + { + "epoch": 0.5994148327184836, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.03246784210205, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8788843154907227, + "num_tokens": 183037908.0, + "step": 4712 + }, + { + "epoch": 0.5995420429970741, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.033660888671875, + "learning_rate": 1e-06, + "loss": 0.3897, + "mean_token_accuracy": 0.8659730553627014, + "num_tokens": 183077410.0, + "step": 4713 + }, + { + "epoch": 0.5996692532756647, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.99578332901001, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8928890228271484, + "num_tokens": 183112972.0, + "step": 4714 + }, + { + "epoch": 0.5997964635542552, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.086725234985352, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8758773803710938, + "num_tokens": 183153761.0, + "step": 4715 + }, + { + "epoch": 0.5999236738328457, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.973598480224609, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8757985234260559, + "num_tokens": 183195520.0, + "step": 4716 + }, + { + "epoch": 0.6000508841114363, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.977834701538086, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8836610317230225, + "num_tokens": 183234930.0, + "step": 4717 + }, + { + "epoch": 0.6001780943900267, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.9948930740356445, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8792804479598999, + "num_tokens": 183275178.0, + "step": 4718 + }, + { + "epoch": 0.6003053046686172, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.968291282653809, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8839391469955444, + "num_tokens": 183318347.0, + "step": 4719 + }, + { + "epoch": 0.6004325149472077, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.087323188781738, + "learning_rate": 1e-06, + "loss": 0.3863, + "mean_token_accuracy": 0.8710224628448486, + "num_tokens": 183355947.0, + "step": 4720 + }, + { + "epoch": 0.6005597252257983, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.977402210235596, + "learning_rate": 1e-06, + "loss": 0.3861, + "mean_token_accuracy": 0.8714067935943604, + "num_tokens": 183398554.0, + "step": 4721 + }, + { + "epoch": 0.6006869355043888, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.9846367835998535, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8799892663955688, + "num_tokens": 183439863.0, + "step": 4722 + }, + { + "epoch": 0.6008141457829793, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.436656951904297, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.879888653755188, + "num_tokens": 183474155.0, + "step": 4723 + }, + { + "epoch": 0.6009413560615697, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.9574689865112305, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8702802658081055, + "num_tokens": 183516866.0, + "step": 4724 + }, + { + "epoch": 0.6010685663401603, + "ewc_loss": 0.0107421875, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.52346420288086e-06, + "grad_norm": 8.053475379943848, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8794635534286499, + "num_tokens": 183548774.0, + "step": 4725 + }, + { + "epoch": 0.6011957766187508, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.052419662475586, + "learning_rate": 1e-06, + "loss": 0.3811, + "mean_token_accuracy": 0.8722758293151855, + "num_tokens": 183591915.0, + "step": 4726 + }, + { + "epoch": 0.6013229868973413, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.125198364257812, + "learning_rate": 1e-06, + "loss": 0.3908, + "mean_token_accuracy": 0.8703101277351379, + "num_tokens": 183624751.0, + "step": 4727 + }, + { + "epoch": 0.6014501971759318, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.084553718566895, + "learning_rate": 1e-06, + "loss": 0.4192, + "mean_token_accuracy": 0.8589455485343933, + "num_tokens": 183665665.0, + "step": 4728 + }, + { + "epoch": 0.6015774074545224, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.992372989654541, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8757847547531128, + "num_tokens": 183704437.0, + "step": 4729 + }, + { + "epoch": 0.6017046177331128, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.038735389709473, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.877530574798584, + "num_tokens": 183744035.0, + "step": 4730 + }, + { + "epoch": 0.6018318280117033, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.0420560836792, + "learning_rate": 1e-06, + "loss": 0.4001, + "mean_token_accuracy": 0.8662188053131104, + "num_tokens": 183784566.0, + "step": 4731 + }, + { + "epoch": 0.6019590382902938, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.000506401062012, + "learning_rate": 1e-06, + "loss": 0.4077, + "mean_token_accuracy": 0.8633099794387817, + "num_tokens": 183825662.0, + "step": 4732 + }, + { + "epoch": 0.6020862485688844, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.1696090698242188e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.026912689208984, + "learning_rate": 1e-06, + "loss": 0.3949, + "mean_token_accuracy": 0.8649792075157166, + "num_tokens": 183866337.0, + "step": 4733 + }, + { + "epoch": 0.6022134588474749, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.181529998779297e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.958081245422363, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8770430088043213, + "num_tokens": 183908103.0, + "step": 4734 + }, + { + "epoch": 0.6023406691260654, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.181529998779297e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.013411521911621, + "learning_rate": 1e-06, + "loss": 0.3898, + "mean_token_accuracy": 0.8691173791885376, + "num_tokens": 183947798.0, + "step": 4735 + }, + { + "epoch": 0.6024678794046558, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.181529998779297e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.992105484008789, + "learning_rate": 1e-06, + "loss": 0.4037, + "mean_token_accuracy": 0.8657498955726624, + "num_tokens": 183989952.0, + "step": 4736 + }, + { + "epoch": 0.6025950896832464, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.181529998779297e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.930922031402588, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8860700130462646, + "num_tokens": 184033775.0, + "step": 4737 + }, + { + "epoch": 0.6027222999618369, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.181529998779297e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.074313163757324, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.883001446723938, + "num_tokens": 184072562.0, + "step": 4738 + }, + { + "epoch": 0.6028495102404274, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.181529998779297e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 7.999440670013428, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8752405047416687, + "num_tokens": 184113420.0, + "step": 4739 + }, + { + "epoch": 0.602976720519018, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.181529998779297e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.046072006225586, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.8681257963180542, + "num_tokens": 184152541.0, + "step": 4740 + }, + { + "epoch": 0.6031039307976085, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.058356285095215, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8805896043777466, + "num_tokens": 184194074.0, + "step": 4741 + }, + { + "epoch": 0.6032311410761989, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.07513427734375, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8775672912597656, + "num_tokens": 184228600.0, + "step": 4742 + }, + { + "epoch": 0.6033583513547894, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.039094924926758, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8828492164611816, + "num_tokens": 184268368.0, + "step": 4743 + }, + { + "epoch": 0.60348556163338, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.03990364074707, + "learning_rate": 1e-06, + "loss": 0.407, + "mean_token_accuracy": 0.8631471395492554, + "num_tokens": 184311563.0, + "step": 4744 + }, + { + "epoch": 0.6036127719119705, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.017101287841797, + "learning_rate": 1e-06, + "loss": 0.3664, + "mean_token_accuracy": 0.8761538863182068, + "num_tokens": 184350502.0, + "step": 4745 + }, + { + "epoch": 0.603739982190561, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.022753715515137, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.8772335648536682, + "num_tokens": 184389444.0, + "step": 4746 + }, + { + "epoch": 0.6038671924691515, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.98998498916626, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8768358826637268, + "num_tokens": 184431254.0, + "step": 4747 + }, + { + "epoch": 0.603994402747742, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.04479694366455, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.873778223991394, + "num_tokens": 184473549.0, + "step": 4748 + }, + { + "epoch": 0.6041216130263325, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.044095039367676, + "learning_rate": 1e-06, + "loss": 0.4017, + "mean_token_accuracy": 0.8660790920257568, + "num_tokens": 184514712.0, + "step": 4749 + }, + { + "epoch": 0.604248823304923, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.002304077148438, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8855782747268677, + "num_tokens": 184551610.0, + "step": 4750 + }, + { + "epoch": 0.6043760335835135, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.06164264678955, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8691930174827576, + "num_tokens": 184592183.0, + "step": 4751 + }, + { + "epoch": 0.6045032438621041, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.028426170349121, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.888827383518219, + "num_tokens": 184629381.0, + "step": 4752 + }, + { + "epoch": 0.6046304541406946, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.19582748413086, + "learning_rate": 1e-06, + "loss": 0.4034, + "mean_token_accuracy": 0.8648456335067749, + "num_tokens": 184662434.0, + "step": 4753 + }, + { + "epoch": 0.604757664419285, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.098711967468262, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.8723519444465637, + "num_tokens": 184701060.0, + "step": 4754 + }, + { + "epoch": 0.6048848746978756, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.064278602600098, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.8714181184768677, + "num_tokens": 184741634.0, + "step": 4755 + }, + { + "epoch": 0.6050120849764661, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.08047866821289, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8792910575866699, + "num_tokens": 184777902.0, + "step": 4756 + }, + { + "epoch": 0.6051392952550566, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.07857608795166, + "learning_rate": 1e-06, + "loss": 0.3868, + "mean_token_accuracy": 0.8700365424156189, + "num_tokens": 184816005.0, + "step": 4757 + }, + { + "epoch": 0.6052665055336471, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.0411376953125, + "learning_rate": 1e-06, + "loss": 0.3928, + "mean_token_accuracy": 0.8683587312698364, + "num_tokens": 184855663.0, + "step": 4758 + }, + { + "epoch": 0.6053937158122377, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.964016437530518, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8763710856437683, + "num_tokens": 184898088.0, + "step": 4759 + }, + { + "epoch": 0.6055209260908282, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.12460708618164, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8794187307357788, + "num_tokens": 184932926.0, + "step": 4760 + }, + { + "epoch": 0.6056481363694186, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.321494102478027, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8834099769592285, + "num_tokens": 184971226.0, + "step": 4761 + }, + { + "epoch": 0.6057753466480091, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.118860244750977, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8732656240463257, + "num_tokens": 185010083.0, + "step": 4762 + }, + { + "epoch": 0.6059025569265997, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.992992877960205, + "learning_rate": 1e-06, + "loss": 0.4056, + "mean_token_accuracy": 0.863686203956604, + "num_tokens": 185050034.0, + "step": 4763 + }, + { + "epoch": 0.6060297672051902, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.078625679016113, + "learning_rate": 1e-06, + "loss": 0.3921, + "mean_token_accuracy": 0.8681541085243225, + "num_tokens": 185087239.0, + "step": 4764 + }, + { + "epoch": 0.6061569774837807, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.102731704711914, + "learning_rate": 1e-06, + "loss": 0.4123, + "mean_token_accuracy": 0.8620823621749878, + "num_tokens": 185125155.0, + "step": 4765 + }, + { + "epoch": 0.6062841877623713, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.056796073913574, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8745384216308594, + "num_tokens": 185160899.0, + "step": 4766 + }, + { + "epoch": 0.6064113980409617, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.073344230651855, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8825889825820923, + "num_tokens": 185196781.0, + "step": 4767 + }, + { + "epoch": 0.6065386083195522, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.193450927734375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.062909126281738, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8710739612579346, + "num_tokens": 185231675.0, + "step": 4768 + }, + { + "epoch": 0.6066658185981427, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.026471138000488, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8711544275283813, + "num_tokens": 185272107.0, + "step": 4769 + }, + { + "epoch": 0.6067930288767333, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.102248191833496, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8798682689666748, + "num_tokens": 185307233.0, + "step": 4770 + }, + { + "epoch": 0.6069202391553238, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.10189151763916, + "learning_rate": 1e-06, + "loss": 0.391, + "mean_token_accuracy": 0.8686215281486511, + "num_tokens": 185342098.0, + "step": 4771 + }, + { + "epoch": 0.6070474494339143, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.041747093200684, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8882736563682556, + "num_tokens": 185376298.0, + "step": 4772 + }, + { + "epoch": 0.6071746597125047, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.986447334289551, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.880541205406189, + "num_tokens": 185418161.0, + "step": 4773 + }, + { + "epoch": 0.6073018699910953, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.037635803222656, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8710378408432007, + "num_tokens": 185455071.0, + "step": 4774 + }, + { + "epoch": 0.6074290802696858, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.088122367858887, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8871496915817261, + "num_tokens": 185489525.0, + "step": 4775 + }, + { + "epoch": 0.6075562905482763, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.08792781829834, + "learning_rate": 1e-06, + "loss": 0.4182, + "mean_token_accuracy": 0.8622188568115234, + "num_tokens": 185528681.0, + "step": 4776 + }, + { + "epoch": 0.6076835008268668, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.055980682373047, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8761581778526306, + "num_tokens": 185568556.0, + "step": 4777 + }, + { + "epoch": 0.6078107111054574, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.096153259277344, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.875046968460083, + "num_tokens": 185604754.0, + "step": 4778 + }, + { + "epoch": 0.6079379213840478, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.078258514404297, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8914201855659485, + "num_tokens": 185642114.0, + "step": 4779 + }, + { + "epoch": 0.6080651316626383, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.118300437927246, + "learning_rate": 1e-06, + "loss": 0.4043, + "mean_token_accuracy": 0.8629587888717651, + "num_tokens": 185681017.0, + "step": 4780 + }, + { + "epoch": 0.6081923419412288, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.086874961853027, + "learning_rate": 1e-06, + "loss": 0.3806, + "mean_token_accuracy": 0.8729192614555359, + "num_tokens": 185715270.0, + "step": 4781 + }, + { + "epoch": 0.6083195522198194, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.002484321594238, + "learning_rate": 1e-06, + "loss": 0.3976, + "mean_token_accuracy": 0.8673507571220398, + "num_tokens": 185759049.0, + "step": 4782 + }, + { + "epoch": 0.6084467624984099, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.188966751098633, + "learning_rate": 1e-06, + "loss": 0.4083, + "mean_token_accuracy": 0.8605031371116638, + "num_tokens": 185792404.0, + "step": 4783 + }, + { + "epoch": 0.6085739727770004, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.084189414978027, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8798426389694214, + "num_tokens": 185827987.0, + "step": 4784 + }, + { + "epoch": 0.6087011830555908, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 7.963718891143799, + "learning_rate": 1e-06, + "loss": 0.3757, + "mean_token_accuracy": 0.875867486000061, + "num_tokens": 185871329.0, + "step": 4785 + }, + { + "epoch": 0.6088283933341814, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.130138397216797, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8823140859603882, + "num_tokens": 185902779.0, + "step": 4786 + }, + { + "epoch": 0.6089556036127719, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.198894500732422, + "learning_rate": 1e-06, + "loss": 0.4136, + "mean_token_accuracy": 0.8616552948951721, + "num_tokens": 185936205.0, + "step": 4787 + }, + { + "epoch": 0.6090828138913624, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.112545013427734, + "learning_rate": 1e-06, + "loss": 0.3659, + "mean_token_accuracy": 0.8763289451599121, + "num_tokens": 185971465.0, + "step": 4788 + }, + { + "epoch": 0.609210024169953, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.06656551361084, + "learning_rate": 1e-06, + "loss": 0.4163, + "mean_token_accuracy": 0.8601697683334351, + "num_tokens": 186009490.0, + "step": 4789 + }, + { + "epoch": 0.6093372344485435, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.034159660339355, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8891710042953491, + "num_tokens": 186045047.0, + "step": 4790 + }, + { + "epoch": 0.6094644447271339, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.057743072509766, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8726119995117188, + "num_tokens": 186087713.0, + "step": 4791 + }, + { + "epoch": 0.6095916550057244, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.016867637634277, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8776726126670837, + "num_tokens": 186131266.0, + "step": 4792 + }, + { + "epoch": 0.609718865284315, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.02946662902832, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8764362931251526, + "num_tokens": 186171282.0, + "step": 4793 + }, + { + "epoch": 0.6098460755629055, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.205371856689453e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.088822364807129, + "learning_rate": 1e-06, + "loss": 0.4021, + "mean_token_accuracy": 0.8664054274559021, + "num_tokens": 186212728.0, + "step": 4794 + }, + { + "epoch": 0.609973285841496, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.110552787780762, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.8711356520652771, + "num_tokens": 186248694.0, + "step": 4795 + }, + { + "epoch": 0.6101004961200865, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.091925621032715, + "learning_rate": 1e-06, + "loss": 0.4022, + "mean_token_accuracy": 0.8668996691703796, + "num_tokens": 186286449.0, + "step": 4796 + }, + { + "epoch": 0.610227706398677, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.109248161315918, + "learning_rate": 1e-06, + "loss": 0.4008, + "mean_token_accuracy": 0.8659223914146423, + "num_tokens": 186324863.0, + "step": 4797 + }, + { + "epoch": 0.6103549166772675, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.14215087890625, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8786625266075134, + "num_tokens": 186359457.0, + "step": 4798 + }, + { + "epoch": 0.610482126955858, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.176193237304688, + "learning_rate": 1e-06, + "loss": 0.3757, + "mean_token_accuracy": 0.8723015189170837, + "num_tokens": 186393735.0, + "step": 4799 + }, + { + "epoch": 0.6106093372344485, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.078722953796387, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8796898126602173, + "num_tokens": 186430808.0, + "step": 4800 + }, + { + "epoch": 0.6107365475130391, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.053241729736328, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8912787437438965, + "num_tokens": 186465895.0, + "step": 4801 + }, + { + "epoch": 0.6108637577916296, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.126834869384766, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8741860389709473, + "num_tokens": 186499110.0, + "step": 4802 + }, + { + "epoch": 0.61099096807022, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.075336456298828, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8785033822059631, + "num_tokens": 186539874.0, + "step": 4803 + }, + { + "epoch": 0.6111181783488105, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.961684226989746, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8750693798065186, + "num_tokens": 186582415.0, + "step": 4804 + }, + { + "epoch": 0.6112453886274011, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.07830810546875, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8871254324913025, + "num_tokens": 186617746.0, + "step": 4805 + }, + { + "epoch": 0.6113725989059916, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.107159614562988, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8871281743049622, + "num_tokens": 186652119.0, + "step": 4806 + }, + { + "epoch": 0.6114998091845821, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.111334800720215, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8783555626869202, + "num_tokens": 186688088.0, + "step": 4807 + }, + { + "epoch": 0.6116270194631727, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.022090911865234, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8787849545478821, + "num_tokens": 186731019.0, + "step": 4808 + }, + { + "epoch": 0.6117542297417632, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.964395999908447, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8808801174163818, + "num_tokens": 186772202.0, + "step": 4809 + }, + { + "epoch": 0.6118814400203536, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.14902114868164, + "learning_rate": 1e-06, + "loss": 0.3736, + "mean_token_accuracy": 0.8741766214370728, + "num_tokens": 186807256.0, + "step": 4810 + }, + { + "epoch": 0.6120086502989441, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.051440238952637, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8724017143249512, + "num_tokens": 186846037.0, + "step": 4811 + }, + { + "epoch": 0.6121358605775347, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.105374336242676, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8708217144012451, + "num_tokens": 186884217.0, + "step": 4812 + }, + { + "epoch": 0.6122630708561252, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.11648941040039, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8725014328956604, + "num_tokens": 186921131.0, + "step": 4813 + }, + { + "epoch": 0.6123902811347157, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.035240173339844, + "learning_rate": 1e-06, + "loss": 0.3912, + "mean_token_accuracy": 0.8692310452461243, + "num_tokens": 186964532.0, + "step": 4814 + }, + { + "epoch": 0.6125174914133062, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.162764549255371, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8772691488265991, + "num_tokens": 186999905.0, + "step": 4815 + }, + { + "epoch": 0.6126447016918967, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.078242301940918, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8873208165168762, + "num_tokens": 187035658.0, + "step": 4816 + }, + { + "epoch": 0.6127719119704872, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.978811264038086, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8755728006362915, + "num_tokens": 187079340.0, + "step": 4817 + }, + { + "epoch": 0.6128991222490777, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.085197448730469, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8765773177146912, + "num_tokens": 187118418.0, + "step": 4818 + }, + { + "epoch": 0.6130263325276682, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.059224128723145, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8809221982955933, + "num_tokens": 187156448.0, + "step": 4819 + }, + { + "epoch": 0.6131535428062588, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.043378829956055, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.8744796514511108, + "num_tokens": 187194686.0, + "step": 4820 + }, + { + "epoch": 0.6132807530848493, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.061055183410645, + "learning_rate": 1e-06, + "loss": 0.4098, + "mean_token_accuracy": 0.8650866746902466, + "num_tokens": 187234268.0, + "step": 4821 + }, + { + "epoch": 0.6134079633634397, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.048192024230957, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8782984018325806, + "num_tokens": 187271275.0, + "step": 4822 + }, + { + "epoch": 0.6135351736420303, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.048182487487793, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8863024711608887, + "num_tokens": 187308056.0, + "step": 4823 + }, + { + "epoch": 0.6136623839206208, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.054482460021973, + "learning_rate": 1e-06, + "loss": 0.3846, + "mean_token_accuracy": 0.870475172996521, + "num_tokens": 187350589.0, + "step": 4824 + }, + { + "epoch": 0.6137895941992113, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.09892463684082, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8801757097244263, + "num_tokens": 187385127.0, + "step": 4825 + }, + { + "epoch": 0.6139168044778018, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.039613723754883, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8742566108703613, + "num_tokens": 187427992.0, + "step": 4826 + }, + { + "epoch": 0.6140440147563924, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.056636810302734, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8755046725273132, + "num_tokens": 187463908.0, + "step": 4827 + }, + { + "epoch": 0.6141712250349828, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.06828784942627, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8780724406242371, + "num_tokens": 187500758.0, + "step": 4828 + }, + { + "epoch": 0.6142984353135733, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.975735664367676, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8829115033149719, + "num_tokens": 187546901.0, + "step": 4829 + }, + { + "epoch": 0.6144256455921638, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.117350578308105, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8752845525741577, + "num_tokens": 187583597.0, + "step": 4830 + }, + { + "epoch": 0.6145528558707544, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.025670051574707, + "learning_rate": 1e-06, + "loss": 0.4164, + "mean_token_accuracy": 0.8618749976158142, + "num_tokens": 187627322.0, + "step": 4831 + }, + { + "epoch": 0.6146800661493449, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.09847640991211, + "learning_rate": 1e-06, + "loss": 0.4193, + "mean_token_accuracy": 0.8612382411956787, + "num_tokens": 187666780.0, + "step": 4832 + }, + { + "epoch": 0.6148072764279354, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.071636199951172, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.8703076839447021, + "num_tokens": 187703799.0, + "step": 4833 + }, + { + "epoch": 0.6149344867065258, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.128893852233887, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8817789554595947, + "num_tokens": 187739133.0, + "step": 4834 + }, + { + "epoch": 0.6150616969851164, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.180837631225586, + "learning_rate": 1e-06, + "loss": 0.4267, + "mean_token_accuracy": 0.855092465877533, + "num_tokens": 187774693.0, + "step": 4835 + }, + { + "epoch": 0.6151889072637069, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.9661865234375, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.8738340735435486, + "num_tokens": 187820452.0, + "step": 4836 + }, + { + "epoch": 0.6153161175422974, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.993505001068115, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8790314793586731, + "num_tokens": 187856800.0, + "step": 4837 + }, + { + "epoch": 0.615443327820888, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.094674110412598, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8935145735740662, + "num_tokens": 187891234.0, + "step": 4838 + }, + { + "epoch": 0.6155705380994785, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.057971954345703, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8839892148971558, + "num_tokens": 187926258.0, + "step": 4839 + }, + { + "epoch": 0.6156977483780689, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.081847190856934, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8815009593963623, + "num_tokens": 187961595.0, + "step": 4840 + }, + { + "epoch": 0.6158249586566594, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.072315216064453, + "learning_rate": 1e-06, + "loss": 0.3757, + "mean_token_accuracy": 0.872064471244812, + "num_tokens": 187999229.0, + "step": 4841 + }, + { + "epoch": 0.61595216893525, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.012255668640137, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8855210542678833, + "num_tokens": 188039591.0, + "step": 4842 + }, + { + "epoch": 0.6160793792138405, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.990690231323242, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8779078125953674, + "num_tokens": 188084447.0, + "step": 4843 + }, + { + "epoch": 0.616206589492431, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.99360990524292, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8781285285949707, + "num_tokens": 188127771.0, + "step": 4844 + }, + { + "epoch": 0.6163337997710215, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.2172927856445312e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.107294082641602, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.8681060075759888, + "num_tokens": 188168964.0, + "step": 4845 + }, + { + "epoch": 0.616461010049612, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.179618835449219, + "learning_rate": 1e-06, + "loss": 0.3978, + "mean_token_accuracy": 0.8675253987312317, + "num_tokens": 188204044.0, + "step": 4846 + }, + { + "epoch": 0.6165882203282025, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.177366256713867, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8677535057067871, + "num_tokens": 188238023.0, + "step": 4847 + }, + { + "epoch": 0.616715430606793, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.015334129333496, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8951283693313599, + "num_tokens": 188275729.0, + "step": 4848 + }, + { + "epoch": 0.6168426408853835, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.067289352416992, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8799781799316406, + "num_tokens": 188313198.0, + "step": 4849 + }, + { + "epoch": 0.6169698511639741, + "ewc_loss": 0.0108642578125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.087543487548828, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8870557546615601, + "num_tokens": 188345673.0, + "step": 4850 + }, + { + "epoch": 0.6170970614425646, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.053436279296875, + "learning_rate": 1e-06, + "loss": 0.3744, + "mean_token_accuracy": 0.8692121505737305, + "num_tokens": 188385295.0, + "step": 4851 + }, + { + "epoch": 0.617224271721155, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.05819034576416, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8836572170257568, + "num_tokens": 188427244.0, + "step": 4852 + }, + { + "epoch": 0.6173514819997455, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.175044059753418, + "learning_rate": 1e-06, + "loss": 0.4073, + "mean_token_accuracy": 0.8669065833091736, + "num_tokens": 188462857.0, + "step": 4853 + }, + { + "epoch": 0.6174786922783361, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.014309883117676, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.876188337802887, + "num_tokens": 188507080.0, + "step": 4854 + }, + { + "epoch": 0.6176059025569266, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.07441520690918, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8813592195510864, + "num_tokens": 188541801.0, + "step": 4855 + }, + { + "epoch": 0.6177331128355171, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.057878494262695, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8849591016769409, + "num_tokens": 188579490.0, + "step": 4856 + }, + { + "epoch": 0.6178603231141077, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.011870384216309, + "learning_rate": 1e-06, + "loss": 0.3986, + "mean_token_accuracy": 0.8674678802490234, + "num_tokens": 188621235.0, + "step": 4857 + }, + { + "epoch": 0.6179875333926982, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.118791580200195, + "learning_rate": 1e-06, + "loss": 0.4096, + "mean_token_accuracy": 0.8653448820114136, + "num_tokens": 188663342.0, + "step": 4858 + }, + { + "epoch": 0.6181147436712886, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.991812229156494, + "learning_rate": 1e-06, + "loss": 0.3905, + "mean_token_accuracy": 0.8694959878921509, + "num_tokens": 188709437.0, + "step": 4859 + }, + { + "epoch": 0.6182419539498791, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.000463485717773, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8769080638885498, + "num_tokens": 188751810.0, + "step": 4860 + }, + { + "epoch": 0.6183691642284697, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.061907768249512, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.8733943700790405, + "num_tokens": 188791890.0, + "step": 4861 + }, + { + "epoch": 0.6184963745070602, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.171031951904297, + "learning_rate": 1e-06, + "loss": 0.4018, + "mean_token_accuracy": 0.8614815473556519, + "num_tokens": 188830432.0, + "step": 4862 + }, + { + "epoch": 0.6186235847856507, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.017963409423828, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8750510215759277, + "num_tokens": 188870161.0, + "step": 4863 + }, + { + "epoch": 0.6187507950642412, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.034062385559082, + "learning_rate": 1e-06, + "loss": 0.3922, + "mean_token_accuracy": 0.8699620962142944, + "num_tokens": 188912039.0, + "step": 4864 + }, + { + "epoch": 0.6188780053428317, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.101058006286621, + "learning_rate": 1e-06, + "loss": 0.39, + "mean_token_accuracy": 0.8656724691390991, + "num_tokens": 188948884.0, + "step": 4865 + }, + { + "epoch": 0.6190052156214222, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.05572509765625, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8767747282981873, + "num_tokens": 188983691.0, + "step": 4866 + }, + { + "epoch": 0.6191324259000127, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.001266479492188, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.892823338508606, + "num_tokens": 189021017.0, + "step": 4867 + }, + { + "epoch": 0.6192596361786032, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.043596267700195, + "learning_rate": 1e-06, + "loss": 0.4086, + "mean_token_accuracy": 0.8642407655715942, + "num_tokens": 189063105.0, + "step": 4868 + }, + { + "epoch": 0.6193868464571938, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.156543731689453, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.8731100559234619, + "num_tokens": 189101049.0, + "step": 4869 + }, + { + "epoch": 0.6195140567357843, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.2292137145996094e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.082756996154785, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8830658197402954, + "num_tokens": 189138588.0, + "step": 4870 + }, + { + "epoch": 0.6196412670143747, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.028109550476074, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8867769241333008, + "num_tokens": 189178062.0, + "step": 4871 + }, + { + "epoch": 0.6197684772929652, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.017574310302734, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8888444900512695, + "num_tokens": 189219735.0, + "step": 4872 + }, + { + "epoch": 0.6198956875715558, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.103275299072266, + "learning_rate": 1e-06, + "loss": 0.3643, + "mean_token_accuracy": 0.8768282532691956, + "num_tokens": 189256027.0, + "step": 4873 + }, + { + "epoch": 0.6200228978501463, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.201200485229492, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8775986433029175, + "num_tokens": 189293854.0, + "step": 4874 + }, + { + "epoch": 0.6201501081287368, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.171728134155273, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.8688286542892456, + "num_tokens": 189332442.0, + "step": 4875 + }, + { + "epoch": 0.6202773184073274, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.167699813842773, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8813595771789551, + "num_tokens": 189370261.0, + "step": 4876 + }, + { + "epoch": 0.6204045286859178, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.116016387939453, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.878801703453064, + "num_tokens": 189403528.0, + "step": 4877 + }, + { + "epoch": 0.6205317389645083, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.134929656982422, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8737539052963257, + "num_tokens": 189441076.0, + "step": 4878 + }, + { + "epoch": 0.6206589492430988, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.08046817779541, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.8793158531188965, + "num_tokens": 189478472.0, + "step": 4879 + }, + { + "epoch": 0.6207861595216894, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.115131378173828, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8796426653862, + "num_tokens": 189520339.0, + "step": 4880 + }, + { + "epoch": 0.6209133698002799, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.062178611755371, + "learning_rate": 1e-06, + "loss": 0.4037, + "mean_token_accuracy": 0.8648913502693176, + "num_tokens": 189560331.0, + "step": 4881 + }, + { + "epoch": 0.6210405800788704, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.068425178527832, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.882124125957489, + "num_tokens": 189598834.0, + "step": 4882 + }, + { + "epoch": 0.6211677903574608, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.248302459716797, + "learning_rate": 1e-06, + "loss": 0.3966, + "mean_token_accuracy": 0.8675816059112549, + "num_tokens": 189634693.0, + "step": 4883 + }, + { + "epoch": 0.6212950006360514, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.024950981140137, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8772162199020386, + "num_tokens": 189675725.0, + "step": 4884 + }, + { + "epoch": 0.6214222109146419, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.005810737609863, + "learning_rate": 1e-06, + "loss": 0.3772, + "mean_token_accuracy": 0.8722262382507324, + "num_tokens": 189715212.0, + "step": 4885 + }, + { + "epoch": 0.6215494211932324, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.16141414642334, + "learning_rate": 1e-06, + "loss": 0.3738, + "mean_token_accuracy": 0.8694543838500977, + "num_tokens": 189752952.0, + "step": 4886 + }, + { + "epoch": 0.621676631471823, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.117843627929688, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.884230375289917, + "num_tokens": 189794886.0, + "step": 4887 + }, + { + "epoch": 0.6218038417504135, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.029540061950684, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.871424674987793, + "num_tokens": 189837437.0, + "step": 4888 + }, + { + "epoch": 0.6219310520290039, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2411346435546875e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.063992500305176, + "learning_rate": 1e-06, + "loss": 0.4063, + "mean_token_accuracy": 0.863959550857544, + "num_tokens": 189881351.0, + "step": 4889 + }, + { + "epoch": 0.6220582623075944, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2530555725097656e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.15281867980957, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8779664635658264, + "num_tokens": 189915655.0, + "step": 4890 + }, + { + "epoch": 0.622185472586185, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2530555725097656e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.150245666503906, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8782927393913269, + "num_tokens": 189949930.0, + "step": 4891 + }, + { + "epoch": 0.6223126828647755, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2530555725097656e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.032466888427734, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8702059984207153, + "num_tokens": 189989840.0, + "step": 4892 + }, + { + "epoch": 0.622439893143366, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2530555725097656e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.103974342346191, + "learning_rate": 1e-06, + "loss": 0.422, + "mean_token_accuracy": 0.8598295450210571, + "num_tokens": 190031164.0, + "step": 4893 + }, + { + "epoch": 0.6225671034219565, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2530555725097656e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.084888458251953, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8807199001312256, + "num_tokens": 190067415.0, + "step": 4894 + }, + { + "epoch": 0.622694313700547, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2530555725097656e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.051532745361328, + "learning_rate": 1e-06, + "loss": 0.3784, + "mean_token_accuracy": 0.8719611763954163, + "num_tokens": 190108606.0, + "step": 4895 + }, + { + "epoch": 0.6228215239791375, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2530555725097656e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.101730346679688, + "learning_rate": 1e-06, + "loss": 0.3961, + "mean_token_accuracy": 0.8644421100616455, + "num_tokens": 190146212.0, + "step": 4896 + }, + { + "epoch": 0.622948734257728, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2530555725097656e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.222868919372559, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8861774206161499, + "num_tokens": 190177644.0, + "step": 4897 + }, + { + "epoch": 0.6230759445363185, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.01022720336914, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.884389340877533, + "num_tokens": 190221154.0, + "step": 4898 + }, + { + "epoch": 0.6232031548149091, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.018524169921875, + "learning_rate": 1e-06, + "loss": 0.3919, + "mean_token_accuracy": 0.8678309321403503, + "num_tokens": 190265477.0, + "step": 4899 + }, + { + "epoch": 0.6233303650934996, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.058660507202148, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8856767416000366, + "num_tokens": 190304689.0, + "step": 4900 + }, + { + "epoch": 0.62345757537209, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.983154296875, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8741031885147095, + "num_tokens": 190348253.0, + "step": 4901 + }, + { + "epoch": 0.6235847856506805, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.078595161437988, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8873105049133301, + "num_tokens": 190389298.0, + "step": 4902 + }, + { + "epoch": 0.6237119959292711, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.082963943481445, + "learning_rate": 1e-06, + "loss": 0.3909, + "mean_token_accuracy": 0.8698601722717285, + "num_tokens": 190425825.0, + "step": 4903 + }, + { + "epoch": 0.6238392062078616, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.060205459594727, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8862512111663818, + "num_tokens": 190465178.0, + "step": 4904 + }, + { + "epoch": 0.6239664164864521, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.983248233795166, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8765887022018433, + "num_tokens": 190507545.0, + "step": 4905 + }, + { + "epoch": 0.6240936267650427, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.064482688903809, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8821703195571899, + "num_tokens": 190543981.0, + "step": 4906 + }, + { + "epoch": 0.6242208370436332, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.138944625854492, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.8697216510772705, + "num_tokens": 190578534.0, + "step": 4907 + }, + { + "epoch": 0.6243480473222236, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.061834335327148, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8865803480148315, + "num_tokens": 190617441.0, + "step": 4908 + }, + { + "epoch": 0.6244752576008141, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.066458702087402, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8918233513832092, + "num_tokens": 190652271.0, + "step": 4909 + }, + { + "epoch": 0.6246024678794047, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.072060585021973, + "learning_rate": 1e-06, + "loss": 0.3761, + "mean_token_accuracy": 0.8763070106506348, + "num_tokens": 190692890.0, + "step": 4910 + }, + { + "epoch": 0.6247296781579952, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.111434936523438, + "learning_rate": 1e-06, + "loss": 0.3924, + "mean_token_accuracy": 0.868360161781311, + "num_tokens": 190733478.0, + "step": 4911 + }, + { + "epoch": 0.6248568884365857, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.072760581970215, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8857086300849915, + "num_tokens": 190771240.0, + "step": 4912 + }, + { + "epoch": 0.6249840987151762, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.102396011352539, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8895052075386047, + "num_tokens": 190807681.0, + "step": 4913 + }, + { + "epoch": 0.6251113089937667, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.090651512145996, + "learning_rate": 1e-06, + "loss": 0.3895, + "mean_token_accuracy": 0.8713384866714478, + "num_tokens": 190849232.0, + "step": 4914 + }, + { + "epoch": 0.6252385192723572, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.969152450561523, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8861086368560791, + "num_tokens": 190896238.0, + "step": 4915 + }, + { + "epoch": 0.6253657295509477, + "ewc_loss": 0.01092529296875, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.136393547058105, + "learning_rate": 1e-06, + "loss": 0.3884, + "mean_token_accuracy": 0.8727635145187378, + "num_tokens": 190934852.0, + "step": 4916 + }, + { + "epoch": 0.6254929398295382, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.065502166748047, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8868354558944702, + "num_tokens": 190970633.0, + "step": 4917 + }, + { + "epoch": 0.6256201501081288, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.2649765014648438e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.14928150177002, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8711193799972534, + "num_tokens": 191007988.0, + "step": 4918 + }, + { + "epoch": 0.6257473603867193, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.147220611572266, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8834328651428223, + "num_tokens": 191039827.0, + "step": 4919 + }, + { + "epoch": 0.6258745706653097, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.998431205749512, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8857989311218262, + "num_tokens": 191083902.0, + "step": 4920 + }, + { + "epoch": 0.6260017809439002, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.079445838928223, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8882913589477539, + "num_tokens": 191123428.0, + "step": 4921 + }, + { + "epoch": 0.6261289912224908, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.091611862182617, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8763960599899292, + "num_tokens": 191161775.0, + "step": 4922 + }, + { + "epoch": 0.6262562015010813, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.160910606384277, + "learning_rate": 1e-06, + "loss": 0.3896, + "mean_token_accuracy": 0.8682502508163452, + "num_tokens": 191198218.0, + "step": 4923 + }, + { + "epoch": 0.6263834117796718, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.099751472473145, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8755480647087097, + "num_tokens": 191234872.0, + "step": 4924 + }, + { + "epoch": 0.6265106220582624, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.101194381713867, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8774344325065613, + "num_tokens": 191270337.0, + "step": 4925 + }, + { + "epoch": 0.6266378323368528, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.082660675048828, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8731790781021118, + "num_tokens": 191310951.0, + "step": 4926 + }, + { + "epoch": 0.6267650426154433, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.109029769897461, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.8730894923210144, + "num_tokens": 191347206.0, + "step": 4927 + }, + { + "epoch": 0.6268922528940338, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.17801284790039, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.8676587343215942, + "num_tokens": 191383451.0, + "step": 4928 + }, + { + "epoch": 0.6270194631726244, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.166793823242188, + "learning_rate": 1e-06, + "loss": 0.3772, + "mean_token_accuracy": 0.8704243898391724, + "num_tokens": 191419694.0, + "step": 4929 + }, + { + "epoch": 0.6271466734512149, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.006196022033691, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.8749499320983887, + "num_tokens": 191459958.0, + "step": 4930 + }, + { + "epoch": 0.6272738837298054, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.113532066345215, + "learning_rate": 1e-06, + "loss": 0.3729, + "mean_token_accuracy": 0.877015233039856, + "num_tokens": 191498389.0, + "step": 4931 + }, + { + "epoch": 0.6274010940083958, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.137800216674805, + "learning_rate": 1e-06, + "loss": 0.4439, + "mean_token_accuracy": 0.8537609577178955, + "num_tokens": 191540413.0, + "step": 4932 + }, + { + "epoch": 0.6275283042869864, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.08078670501709, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8786748051643372, + "num_tokens": 191576358.0, + "step": 4933 + }, + { + "epoch": 0.6276555145655769, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.01923656463623, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8768110275268555, + "num_tokens": 191615925.0, + "step": 4934 + }, + { + "epoch": 0.6277827248441674, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.038501739501953, + "learning_rate": 1e-06, + "loss": 0.4183, + "mean_token_accuracy": 0.863581657409668, + "num_tokens": 191660427.0, + "step": 4935 + }, + { + "epoch": 0.627909935122758, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.112399101257324, + "learning_rate": 1e-06, + "loss": 0.3897, + "mean_token_accuracy": 0.8711365461349487, + "num_tokens": 191700208.0, + "step": 4936 + }, + { + "epoch": 0.6280371454013485, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.084035873413086, + "learning_rate": 1e-06, + "loss": 0.3939, + "mean_token_accuracy": 0.8727052211761475, + "num_tokens": 191742699.0, + "step": 4937 + }, + { + "epoch": 0.6281643556799389, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.066636085510254, + "learning_rate": 1e-06, + "loss": 0.3753, + "mean_token_accuracy": 0.8720407485961914, + "num_tokens": 191784154.0, + "step": 4938 + }, + { + "epoch": 0.6282915659585294, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.009603500366211, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8794302940368652, + "num_tokens": 191827084.0, + "step": 4939 + }, + { + "epoch": 0.62841877623712, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.178359985351562, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.878057599067688, + "num_tokens": 191862286.0, + "step": 4940 + }, + { + "epoch": 0.6285459865157105, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.204424858093262, + "learning_rate": 1e-06, + "loss": 0.3894, + "mean_token_accuracy": 0.8685925602912903, + "num_tokens": 191898654.0, + "step": 4941 + }, + { + "epoch": 0.628673196794301, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.058356285095215, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.8712219595909119, + "num_tokens": 191940811.0, + "step": 4942 + }, + { + "epoch": 0.6288004070728915, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.222078323364258, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8857542276382446, + "num_tokens": 191978287.0, + "step": 4943 + }, + { + "epoch": 0.628927617351482, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.010794639587402, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.886996328830719, + "num_tokens": 192021790.0, + "step": 4944 + }, + { + "epoch": 0.6290548276300725, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.101908683776855, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8725665807723999, + "num_tokens": 192065382.0, + "step": 4945 + }, + { + "epoch": 0.629182037908663, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.099040985107422, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8786881566047668, + "num_tokens": 192103659.0, + "step": 4946 + }, + { + "epoch": 0.6293092481872535, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.239004135131836, + "learning_rate": 1e-06, + "loss": 0.4473, + "mean_token_accuracy": 0.8539438247680664, + "num_tokens": 192142872.0, + "step": 4947 + }, + { + "epoch": 0.6294364584658441, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.276897430419922e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.38715934753418, + "learning_rate": 1e-06, + "loss": 0.4027, + "mean_token_accuracy": 0.8648197650909424, + "num_tokens": 192171775.0, + "step": 4948 + }, + { + "epoch": 0.6295636687444346, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.030600547790527, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8816003799438477, + "num_tokens": 192212770.0, + "step": 4949 + }, + { + "epoch": 0.629690879023025, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.993262767791748, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8744235038757324, + "num_tokens": 192250433.0, + "step": 4950 + }, + { + "epoch": 0.6298180893016155, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.092700958251953, + "learning_rate": 1e-06, + "loss": 0.3913, + "mean_token_accuracy": 0.8718197345733643, + "num_tokens": 192288950.0, + "step": 4951 + }, + { + "epoch": 0.6299452995802061, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.043131828308105, + "learning_rate": 1e-06, + "loss": 0.4138, + "mean_token_accuracy": 0.8620056509971619, + "num_tokens": 192335538.0, + "step": 4952 + }, + { + "epoch": 0.6300725098587966, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.098833084106445, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8920317888259888, + "num_tokens": 192369058.0, + "step": 4953 + }, + { + "epoch": 0.6301997201373871, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.106837272644043, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8960426449775696, + "num_tokens": 192404084.0, + "step": 4954 + }, + { + "epoch": 0.6303269304159776, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.137377738952637, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8875055313110352, + "num_tokens": 192440423.0, + "step": 4955 + }, + { + "epoch": 0.6304541406945681, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.113786697387695, + "learning_rate": 1e-06, + "loss": 0.3945, + "mean_token_accuracy": 0.8672716617584229, + "num_tokens": 192481854.0, + "step": 4956 + }, + { + "epoch": 0.6305813509731586, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.133129119873047, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8763552904129028, + "num_tokens": 192516847.0, + "step": 4957 + }, + { + "epoch": 0.6307085612517491, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.10476303100586, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8729946613311768, + "num_tokens": 192556464.0, + "step": 4958 + }, + { + "epoch": 0.6308357715303397, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.180853843688965, + "learning_rate": 1e-06, + "loss": 0.4406, + "mean_token_accuracy": 0.85294508934021, + "num_tokens": 192592549.0, + "step": 4959 + }, + { + "epoch": 0.6309629818089302, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.363507270812988, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8736623525619507, + "num_tokens": 192620397.0, + "step": 4960 + }, + { + "epoch": 0.6310901920875207, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.027813911437988, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8802846074104309, + "num_tokens": 192659722.0, + "step": 4961 + }, + { + "epoch": 0.6312174023661112, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.080945014953613, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8832931518554688, + "num_tokens": 192699578.0, + "step": 4962 + }, + { + "epoch": 0.6313446126447017, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.14593505859375, + "learning_rate": 1e-06, + "loss": 0.392, + "mean_token_accuracy": 0.8705748319625854, + "num_tokens": 192737041.0, + "step": 4963 + }, + { + "epoch": 0.6314718229232922, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.10770034790039, + "learning_rate": 1e-06, + "loss": 0.417, + "mean_token_accuracy": 0.860815703868866, + "num_tokens": 192777900.0, + "step": 4964 + }, + { + "epoch": 0.6315990332018827, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.050795555114746, + "learning_rate": 1e-06, + "loss": 0.3916, + "mean_token_accuracy": 0.8684192895889282, + "num_tokens": 192820806.0, + "step": 4965 + }, + { + "epoch": 0.6317262434804732, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.095951080322266, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8884584903717041, + "num_tokens": 192857361.0, + "step": 4966 + }, + { + "epoch": 0.6318534537590638, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.07437801361084, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8783876895904541, + "num_tokens": 192895969.0, + "step": 4967 + }, + { + "epoch": 0.6319806640376543, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.086371421813965, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8889319896697998, + "num_tokens": 192931508.0, + "step": 4968 + }, + { + "epoch": 0.6321078743162447, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.101346015930176, + "learning_rate": 1e-06, + "loss": 0.3893, + "mean_token_accuracy": 0.8688094019889832, + "num_tokens": 192972299.0, + "step": 4969 + }, + { + "epoch": 0.6322350845948352, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.164708137512207, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8834653496742249, + "num_tokens": 193007962.0, + "step": 4970 + }, + { + "epoch": 0.6323622948734258, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.02207088470459, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8756215572357178, + "num_tokens": 193046590.0, + "step": 4971 + }, + { + "epoch": 0.6324895051520163, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.233545303344727, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8718984723091125, + "num_tokens": 193080305.0, + "step": 4972 + }, + { + "epoch": 0.6326167154306068, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.039134979248047, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8787873983383179, + "num_tokens": 193122401.0, + "step": 4973 + }, + { + "epoch": 0.6327439257091974, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.047195434570312, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8928836584091187, + "num_tokens": 193161133.0, + "step": 4974 + }, + { + "epoch": 0.6328711359877878, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.052770614624023, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8835347890853882, + "num_tokens": 193202309.0, + "step": 4975 + }, + { + "epoch": 0.6329983462663783, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.125082015991211, + "learning_rate": 1e-06, + "loss": 0.3597, + "mean_token_accuracy": 0.8770695924758911, + "num_tokens": 193238483.0, + "step": 4976 + }, + { + "epoch": 0.6331255565449688, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.202832221984863, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8726041316986084, + "num_tokens": 193272832.0, + "step": 4977 + }, + { + "epoch": 0.6332527668235594, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.124971389770508, + "learning_rate": 1e-06, + "loss": 0.4067, + "mean_token_accuracy": 0.8606382608413696, + "num_tokens": 193312561.0, + "step": 4978 + }, + { + "epoch": 0.6333799771021499, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.173151969909668, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8896749019622803, + "num_tokens": 193350537.0, + "step": 4979 + }, + { + "epoch": 0.6335071873807404, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.054646492004395, + "learning_rate": 1e-06, + "loss": 0.3838, + "mean_token_accuracy": 0.8684641122817993, + "num_tokens": 193392014.0, + "step": 4980 + }, + { + "epoch": 0.6336343976593308, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.074127197265625, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8854852914810181, + "num_tokens": 193429400.0, + "step": 4981 + }, + { + "epoch": 0.6337616079379214, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.058156967163086, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.876595675945282, + "num_tokens": 193466483.0, + "step": 4982 + }, + { + "epoch": 0.6338888182165119, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.062769889831543, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8808773159980774, + "num_tokens": 193506952.0, + "step": 4983 + }, + { + "epoch": 0.6340160284951024, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.181468963623047, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.877264142036438, + "num_tokens": 193543014.0, + "step": 4984 + }, + { + "epoch": 0.6341432387736929, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.107255935668945, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.873420000076294, + "num_tokens": 193580575.0, + "step": 4985 + }, + { + "epoch": 0.6342704490522835, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.036324501037598, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8839461803436279, + "num_tokens": 193625652.0, + "step": 4986 + }, + { + "epoch": 0.6343976593308739, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.015613555908203, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8870398998260498, + "num_tokens": 193668549.0, + "step": 4987 + }, + { + "epoch": 0.6345248696094644, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.055068969726562, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8858194947242737, + "num_tokens": 193705414.0, + "step": 4988 + }, + { + "epoch": 0.6346520798880549, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.20931339263916, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8843844532966614, + "num_tokens": 193737352.0, + "step": 4989 + }, + { + "epoch": 0.6347792901666455, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.052278518676758, + "learning_rate": 1e-06, + "loss": 0.3986, + "mean_token_accuracy": 0.8636382818222046, + "num_tokens": 193778760.0, + "step": 4990 + }, + { + "epoch": 0.634906500445236, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.288818359375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.0614013671875, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8915860056877136, + "num_tokens": 193816377.0, + "step": 4991 + }, + { + "epoch": 0.6350337107238265, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.139986038208008, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8685121536254883, + "num_tokens": 193852142.0, + "step": 4992 + }, + { + "epoch": 0.635160921002417, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.070192337036133, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8839890360832214, + "num_tokens": 193888429.0, + "step": 4993 + }, + { + "epoch": 0.6352881312810075, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.011258125305176, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.8770173192024231, + "num_tokens": 193933463.0, + "step": 4994 + }, + { + "epoch": 0.635415341559598, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.133533477783203, + "learning_rate": 1e-06, + "loss": 0.4173, + "mean_token_accuracy": 0.8611732721328735, + "num_tokens": 193974049.0, + "step": 4995 + }, + { + "epoch": 0.6355425518381885, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.13083553314209, + "learning_rate": 1e-06, + "loss": 0.4238, + "mean_token_accuracy": 0.8577994108200073, + "num_tokens": 194012018.0, + "step": 4996 + }, + { + "epoch": 0.6356697621167791, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.082711219787598, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.874966025352478, + "num_tokens": 194049493.0, + "step": 4997 + }, + { + "epoch": 0.6357969723953696, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.119318008422852, + "learning_rate": 1e-06, + "loss": 0.4135, + "mean_token_accuracy": 0.8605514168739319, + "num_tokens": 194090341.0, + "step": 4998 + }, + { + "epoch": 0.63592418267396, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.231829643249512, + "learning_rate": 1e-06, + "loss": 0.3834, + "mean_token_accuracy": 0.8695619106292725, + "num_tokens": 194122800.0, + "step": 4999 + }, + { + "epoch": 0.6360513929525505, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.071353912353516, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.878139853477478, + "num_tokens": 194158552.0, + "step": 5000 + }, + { + "epoch": 0.6361786032311411, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.941896915435791, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8858358263969421, + "num_tokens": 194203119.0, + "step": 5001 + }, + { + "epoch": 0.6363058135097316, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.144549369812012, + "learning_rate": 1e-06, + "loss": 0.3757, + "mean_token_accuracy": 0.8715269565582275, + "num_tokens": 194242053.0, + "step": 5002 + }, + { + "epoch": 0.6364330237883221, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.654277801513672, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.879225492477417, + "num_tokens": 194288456.0, + "step": 5003 + }, + { + "epoch": 0.6365602340669126, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 7.95773983001709, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8754494190216064, + "num_tokens": 194335746.0, + "step": 5004 + }, + { + "epoch": 0.6366874443455031, + "ewc_loss": 0.010986328125, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.64267349243164e-06, + "grad_norm": 8.042665481567383, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8735556602478027, + "num_tokens": 194374053.0, + "step": 5005 + }, + { + "epoch": 0.6368146546240936, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.112035751342773, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.870076596736908, + "num_tokens": 194414077.0, + "step": 5006 + }, + { + "epoch": 0.6369418649026841, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.974390983581543, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8796084523200989, + "num_tokens": 194456553.0, + "step": 5007 + }, + { + "epoch": 0.6370690751812746, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.12554931640625, + "learning_rate": 1e-06, + "loss": 0.3842, + "mean_token_accuracy": 0.870578408241272, + "num_tokens": 194497913.0, + "step": 5008 + }, + { + "epoch": 0.6371962854598652, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.111353874206543, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8890113830566406, + "num_tokens": 194534745.0, + "step": 5009 + }, + { + "epoch": 0.6373234957384557, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.197855949401855, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8744755983352661, + "num_tokens": 194569292.0, + "step": 5010 + }, + { + "epoch": 0.6374507060170462, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.14811897277832, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8756123781204224, + "num_tokens": 194603585.0, + "step": 5011 + }, + { + "epoch": 0.6375779162956366, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.090682983398438, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.8689402341842651, + "num_tokens": 194640741.0, + "step": 5012 + }, + { + "epoch": 0.6377051265742272, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.100353240966797, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8696302771568298, + "num_tokens": 194681301.0, + "step": 5013 + }, + { + "epoch": 0.6378323368528177, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.080245018005371, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.8700560331344604, + "num_tokens": 194720829.0, + "step": 5014 + }, + { + "epoch": 0.6379595471314082, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.13465404510498, + "learning_rate": 1e-06, + "loss": 0.4102, + "mean_token_accuracy": 0.865165650844574, + "num_tokens": 194761798.0, + "step": 5015 + }, + { + "epoch": 0.6380867574099988, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.171366691589355, + "learning_rate": 1e-06, + "loss": 0.4056, + "mean_token_accuracy": 0.8630480766296387, + "num_tokens": 194796516.0, + "step": 5016 + }, + { + "epoch": 0.6382139676885893, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.093174934387207, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8782302141189575, + "num_tokens": 194834488.0, + "step": 5017 + }, + { + "epoch": 0.6383411779671797, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.10251235961914, + "learning_rate": 1e-06, + "loss": 0.3659, + "mean_token_accuracy": 0.8783140182495117, + "num_tokens": 194871110.0, + "step": 5018 + }, + { + "epoch": 0.6384683882457702, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.216501235961914, + "learning_rate": 1e-06, + "loss": 0.3991, + "mean_token_accuracy": 0.8661586046218872, + "num_tokens": 194906788.0, + "step": 5019 + }, + { + "epoch": 0.6385955985243608, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.039828300476074, + "learning_rate": 1e-06, + "loss": 0.3915, + "mean_token_accuracy": 0.8692795038223267, + "num_tokens": 194948565.0, + "step": 5020 + }, + { + "epoch": 0.6387228088029513, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.013327598571777, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8942726850509644, + "num_tokens": 194988625.0, + "step": 5021 + }, + { + "epoch": 0.6388500190815418, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.047038078308105, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8769798278808594, + "num_tokens": 195035829.0, + "step": 5022 + }, + { + "epoch": 0.6389772293601323, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.0864839553833, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.872391939163208, + "num_tokens": 195075154.0, + "step": 5023 + }, + { + "epoch": 0.6391044396387228, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.063197135925293, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8717003464698792, + "num_tokens": 195119561.0, + "step": 5024 + }, + { + "epoch": 0.6392316499173133, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.153450965881348, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8837182521820068, + "num_tokens": 195155673.0, + "step": 5025 + }, + { + "epoch": 0.6393588601959038, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.13358211517334, + "learning_rate": 1e-06, + "loss": 0.4079, + "mean_token_accuracy": 0.8618327379226685, + "num_tokens": 195200384.0, + "step": 5026 + }, + { + "epoch": 0.6394860704744944, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.136762619018555, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.8762561082839966, + "num_tokens": 195235218.0, + "step": 5027 + }, + { + "epoch": 0.6396132807530849, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.099413871765137, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.8716821670532227, + "num_tokens": 195273475.0, + "step": 5028 + }, + { + "epoch": 0.6397404910316754, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.187522888183594, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8829554319381714, + "num_tokens": 195306070.0, + "step": 5029 + }, + { + "epoch": 0.6398677013102658, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.013956069946289, + "learning_rate": 1e-06, + "loss": 0.3664, + "mean_token_accuracy": 0.8751664161682129, + "num_tokens": 195354189.0, + "step": 5030 + }, + { + "epoch": 0.6399949115888564, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.126518249511719, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.873986005783081, + "num_tokens": 195393781.0, + "step": 5031 + }, + { + "epoch": 0.6401221218674469, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.175907135009766, + "learning_rate": 1e-06, + "loss": 0.425, + "mean_token_accuracy": 0.8570890426635742, + "num_tokens": 195432214.0, + "step": 5032 + }, + { + "epoch": 0.6402493321460374, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.112506866455078, + "learning_rate": 1e-06, + "loss": 0.3671, + "mean_token_accuracy": 0.8755019903182983, + "num_tokens": 195470966.0, + "step": 5033 + }, + { + "epoch": 0.6403765424246279, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.131484985351562, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8749105930328369, + "num_tokens": 195510944.0, + "step": 5034 + }, + { + "epoch": 0.6405037527032185, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.099281311035156, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8782850503921509, + "num_tokens": 195551783.0, + "step": 5035 + }, + { + "epoch": 0.6406309629818089, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.105863571166992, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.874966025352478, + "num_tokens": 195592204.0, + "step": 5036 + }, + { + "epoch": 0.6407581732603994, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.262846946716309, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8786234855651855, + "num_tokens": 195621850.0, + "step": 5037 + }, + { + "epoch": 0.6408853835389899, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.109933853149414, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.871595025062561, + "num_tokens": 195658822.0, + "step": 5038 + }, + { + "epoch": 0.6410125938175805, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.084029197692871, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8723549246788025, + "num_tokens": 195698023.0, + "step": 5039 + }, + { + "epoch": 0.641139804096171, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.11767864227295, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8788263201713562, + "num_tokens": 195737868.0, + "step": 5040 + }, + { + "epoch": 0.6412670143747615, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.068754196166992, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8815160989761353, + "num_tokens": 195776192.0, + "step": 5041 + }, + { + "epoch": 0.6413942246533519, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.102303504943848, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8817269802093506, + "num_tokens": 195813468.0, + "step": 5042 + }, + { + "epoch": 0.6415214349319425, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.144356727600098, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8811519742012024, + "num_tokens": 195852643.0, + "step": 5043 + }, + { + "epoch": 0.641648645210533, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.087270736694336, + "learning_rate": 1e-06, + "loss": 0.385, + "mean_token_accuracy": 0.8677430748939514, + "num_tokens": 195891780.0, + "step": 5044 + }, + { + "epoch": 0.6417758554891235, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.125192642211914, + "learning_rate": 1e-06, + "loss": 0.4004, + "mean_token_accuracy": 0.8655635118484497, + "num_tokens": 195931959.0, + "step": 5045 + }, + { + "epoch": 0.641903065767714, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.112749099731445, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.8664274215698242, + "num_tokens": 195968688.0, + "step": 5046 + }, + { + "epoch": 0.6420302760463046, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.044260025024414, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8809735178947449, + "num_tokens": 196011120.0, + "step": 5047 + }, + { + "epoch": 0.642157486324895, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.1102876663208, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.8690762519836426, + "num_tokens": 196048603.0, + "step": 5048 + }, + { + "epoch": 0.6422846966034855, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.058000564575195, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8778014779090881, + "num_tokens": 196091501.0, + "step": 5049 + }, + { + "epoch": 0.6424119068820761, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.180558204650879, + "learning_rate": 1e-06, + "loss": 0.3818, + "mean_token_accuracy": 0.8707545399665833, + "num_tokens": 196134148.0, + "step": 5050 + }, + { + "epoch": 0.6425391171606666, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.08364200592041, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8823309540748596, + "num_tokens": 196170622.0, + "step": 5051 + }, + { + "epoch": 0.6426663274392571, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.204687118530273, + "learning_rate": 1e-06, + "loss": 0.4473, + "mean_token_accuracy": 0.8483433723449707, + "num_tokens": 196208290.0, + "step": 5052 + }, + { + "epoch": 0.6427935377178476, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.08183479309082, + "learning_rate": 1e-06, + "loss": 0.3853, + "mean_token_accuracy": 0.871099054813385, + "num_tokens": 196255887.0, + "step": 5053 + }, + { + "epoch": 0.6429207479964381, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3126602172851562e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.08173942565918, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8753165006637573, + "num_tokens": 196297262.0, + "step": 5054 + }, + { + "epoch": 0.6430479582750286, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3245811462402344e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.996703624725342, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8910325765609741, + "num_tokens": 196336694.0, + "step": 5055 + }, + { + "epoch": 0.6431751685536191, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3245811462402344e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.267778396606445, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.876735270023346, + "num_tokens": 196365808.0, + "step": 5056 + }, + { + "epoch": 0.6433023788322096, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3245811462402344e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.123629570007324, + "learning_rate": 1e-06, + "loss": 0.3882, + "mean_token_accuracy": 0.867883026599884, + "num_tokens": 196407210.0, + "step": 5057 + }, + { + "epoch": 0.6434295891108002, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3245811462402344e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.091793060302734, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.865065336227417, + "num_tokens": 196445848.0, + "step": 5058 + }, + { + "epoch": 0.6435567993893907, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3245811462402344e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.208963394165039, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8767213821411133, + "num_tokens": 196479102.0, + "step": 5059 + }, + { + "epoch": 0.6436840096679812, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.10831356048584, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8823468089103699, + "num_tokens": 196515454.0, + "step": 5060 + }, + { + "epoch": 0.6438112199465716, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.0851411819458, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8805243372917175, + "num_tokens": 196558528.0, + "step": 5061 + }, + { + "epoch": 0.6439384302251622, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.096498489379883, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8736816644668579, + "num_tokens": 196599091.0, + "step": 5062 + }, + { + "epoch": 0.6440656405037527, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.09052562713623, + "learning_rate": 1e-06, + "loss": 0.4054, + "mean_token_accuracy": 0.8660617470741272, + "num_tokens": 196644078.0, + "step": 5063 + }, + { + "epoch": 0.6441928507823432, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.176674842834473, + "learning_rate": 1e-06, + "loss": 0.4296, + "mean_token_accuracy": 0.8544809818267822, + "num_tokens": 196683128.0, + "step": 5064 + }, + { + "epoch": 0.6443200610609338, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.029684066772461, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8793612718582153, + "num_tokens": 196726657.0, + "step": 5065 + }, + { + "epoch": 0.6444472713395243, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.076889991760254, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8762317895889282, + "num_tokens": 196767895.0, + "step": 5066 + }, + { + "epoch": 0.6445744816181147, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.118011474609375, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8882472515106201, + "num_tokens": 196802526.0, + "step": 5067 + }, + { + "epoch": 0.6447016918967052, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.073845863342285, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.8742984533309937, + "num_tokens": 196843957.0, + "step": 5068 + }, + { + "epoch": 0.6448289021752958, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.077983856201172, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.875203013420105, + "num_tokens": 196887650.0, + "step": 5069 + }, + { + "epoch": 0.6449561124538863, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.13808536529541, + "learning_rate": 1e-06, + "loss": 0.3916, + "mean_token_accuracy": 0.8699854016304016, + "num_tokens": 196927018.0, + "step": 5070 + }, + { + "epoch": 0.6450833227324768, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.039894104003906, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8863661289215088, + "num_tokens": 196968249.0, + "step": 5071 + }, + { + "epoch": 0.6452105330110673, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.965064525604248, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8828525543212891, + "num_tokens": 197015588.0, + "step": 5072 + }, + { + "epoch": 0.6453377432896578, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.07528018951416, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8788734674453735, + "num_tokens": 197056633.0, + "step": 5073 + }, + { + "epoch": 0.6454649535682483, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.090066909790039, + "learning_rate": 1e-06, + "loss": 0.3874, + "mean_token_accuracy": 0.870777428150177, + "num_tokens": 197092440.0, + "step": 5074 + }, + { + "epoch": 0.6455921638468388, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.084567070007324, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8874532580375671, + "num_tokens": 197133240.0, + "step": 5075 + }, + { + "epoch": 0.6457193741254293, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.049866676330566, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8770764470100403, + "num_tokens": 197178523.0, + "step": 5076 + }, + { + "epoch": 0.6458465844040199, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.18847370147705, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.8713827729225159, + "num_tokens": 197213675.0, + "step": 5077 + }, + { + "epoch": 0.6459737946826104, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.133160591125488, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8793807029724121, + "num_tokens": 197249050.0, + "step": 5078 + }, + { + "epoch": 0.6461010049612008, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.11328125, + "learning_rate": 1e-06, + "loss": 0.3933, + "mean_token_accuracy": 0.8642903566360474, + "num_tokens": 197288869.0, + "step": 5079 + }, + { + "epoch": 0.6462282152397913, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.206360816955566, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8823539614677429, + "num_tokens": 197322134.0, + "step": 5080 + }, + { + "epoch": 0.6463554255183819, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.119430541992188, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8835065364837646, + "num_tokens": 197357960.0, + "step": 5081 + }, + { + "epoch": 0.6464826357969724, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.057222366333008, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.8741317987442017, + "num_tokens": 197400446.0, + "step": 5082 + }, + { + "epoch": 0.6466098460755629, + "ewc_loss": 0.01104736328125, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.27837085723877, + "learning_rate": 1e-06, + "loss": 0.4437, + "mean_token_accuracy": 0.8503144979476929, + "num_tokens": 197437756.0, + "step": 5083 + }, + { + "epoch": 0.6467370563541535, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.11048698425293, + "learning_rate": 1e-06, + "loss": 0.4106, + "mean_token_accuracy": 0.8615423440933228, + "num_tokens": 197479125.0, + "step": 5084 + }, + { + "epoch": 0.6468642666327439, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.15544605255127, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8712016940116882, + "num_tokens": 197513805.0, + "step": 5085 + }, + { + "epoch": 0.6469914769113344, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.113051414489746, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8797281980514526, + "num_tokens": 197553068.0, + "step": 5086 + }, + { + "epoch": 0.6471186871899249, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.174454689025879, + "learning_rate": 1e-06, + "loss": 0.3937, + "mean_token_accuracy": 0.8671972751617432, + "num_tokens": 197592222.0, + "step": 5087 + }, + { + "epoch": 0.6472458974685155, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.104776382446289, + "learning_rate": 1e-06, + "loss": 0.405, + "mean_token_accuracy": 0.8622552156448364, + "num_tokens": 197635487.0, + "step": 5088 + }, + { + "epoch": 0.647373107747106, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.17147159576416, + "learning_rate": 1e-06, + "loss": 0.4049, + "mean_token_accuracy": 0.8658245801925659, + "num_tokens": 197672347.0, + "step": 5089 + }, + { + "epoch": 0.6475003180256965, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3365020751953125e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.043753623962402, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8807231187820435, + "num_tokens": 197712815.0, + "step": 5090 + }, + { + "epoch": 0.6476275283042869, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.197150230407715, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.869620680809021, + "num_tokens": 197743695.0, + "step": 5091 + }, + { + "epoch": 0.6477547385828775, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.111042976379395, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8807659149169922, + "num_tokens": 197780920.0, + "step": 5092 + }, + { + "epoch": 0.647881948861468, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.167072296142578, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8804507255554199, + "num_tokens": 197816608.0, + "step": 5093 + }, + { + "epoch": 0.6480091591400585, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.093746185302734, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8791009187698364, + "num_tokens": 197853755.0, + "step": 5094 + }, + { + "epoch": 0.648136369418649, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.09010124206543, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.87175053358078, + "num_tokens": 197893264.0, + "step": 5095 + }, + { + "epoch": 0.6482635796972396, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.122050285339355, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8747740983963013, + "num_tokens": 197930912.0, + "step": 5096 + }, + { + "epoch": 0.64839078997583, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.171562194824219, + "learning_rate": 1e-06, + "loss": 0.4114, + "mean_token_accuracy": 0.8658623695373535, + "num_tokens": 197967727.0, + "step": 5097 + }, + { + "epoch": 0.6485180002544205, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.148759841918945, + "learning_rate": 1e-06, + "loss": 0.4044, + "mean_token_accuracy": 0.8655153512954712, + "num_tokens": 198005540.0, + "step": 5098 + }, + { + "epoch": 0.648645210533011, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.1167631149292, + "learning_rate": 1e-06, + "loss": 0.385, + "mean_token_accuracy": 0.8687624931335449, + "num_tokens": 198044006.0, + "step": 5099 + }, + { + "epoch": 0.6487724208116016, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.110857963562012, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8849790692329407, + "num_tokens": 198082117.0, + "step": 5100 + }, + { + "epoch": 0.6488996310901921, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.132397651672363, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8798071146011353, + "num_tokens": 198121024.0, + "step": 5101 + }, + { + "epoch": 0.6490268413687826, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.189001083374023, + "learning_rate": 1e-06, + "loss": 0.4339, + "mean_token_accuracy": 0.8578464984893799, + "num_tokens": 198161282.0, + "step": 5102 + }, + { + "epoch": 0.649154051647373, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.070096015930176, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8784911632537842, + "num_tokens": 198196732.0, + "step": 5103 + }, + { + "epoch": 0.6492812619259636, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.095808982849121, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.891951322555542, + "num_tokens": 198237512.0, + "step": 5104 + }, + { + "epoch": 0.6494084722045541, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.070433616638184, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8778862953186035, + "num_tokens": 198282902.0, + "step": 5105 + }, + { + "epoch": 0.6495356824831446, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.19186019897461, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.8757672309875488, + "num_tokens": 198317097.0, + "step": 5106 + }, + { + "epoch": 0.6496628927617352, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.128561973571777, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.8696543574333191, + "num_tokens": 198357579.0, + "step": 5107 + }, + { + "epoch": 0.6497901030403257, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.088756561279297, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8789420127868652, + "num_tokens": 198398116.0, + "step": 5108 + }, + { + "epoch": 0.6499173133189162, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.197311401367188, + "learning_rate": 1e-06, + "loss": 0.4385, + "mean_token_accuracy": 0.8491827249526978, + "num_tokens": 198436639.0, + "step": 5109 + }, + { + "epoch": 0.6500445235975066, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.207956314086914, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8762006759643555, + "num_tokens": 198475427.0, + "step": 5110 + }, + { + "epoch": 0.6501717338760972, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.102365493774414, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8700726628303528, + "num_tokens": 198516028.0, + "step": 5111 + }, + { + "epoch": 0.6502989441546877, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.168692588806152, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8740460276603699, + "num_tokens": 198549754.0, + "step": 5112 + }, + { + "epoch": 0.6504261544332782, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.073774337768555, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8687516450881958, + "num_tokens": 198595292.0, + "step": 5113 + }, + { + "epoch": 0.6505533647118688, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.151122093200684, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8829408884048462, + "num_tokens": 198629699.0, + "step": 5114 + }, + { + "epoch": 0.6506805749904593, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.128669738769531, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8777387738227844, + "num_tokens": 198665169.0, + "step": 5115 + }, + { + "epoch": 0.6508077852690497, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.273194313049316, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8735371232032776, + "num_tokens": 198694220.0, + "step": 5116 + }, + { + "epoch": 0.6509349955476402, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.179137229919434, + "learning_rate": 1e-06, + "loss": 0.4012, + "mean_token_accuracy": 0.8639256954193115, + "num_tokens": 198729677.0, + "step": 5117 + }, + { + "epoch": 0.6510622058262308, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.104899406433105, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8745148181915283, + "num_tokens": 198767337.0, + "step": 5118 + }, + { + "epoch": 0.6511894161048213, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.16000747680664, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8873382210731506, + "num_tokens": 198801583.0, + "step": 5119 + }, + { + "epoch": 0.6513166263834118, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.092306137084961, + "learning_rate": 1e-06, + "loss": 0.4027, + "mean_token_accuracy": 0.8641295433044434, + "num_tokens": 198844107.0, + "step": 5120 + }, + { + "epoch": 0.6514438366620023, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.134784698486328, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8865149021148682, + "num_tokens": 198881977.0, + "step": 5121 + }, + { + "epoch": 0.6515710469405928, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.098078727722168, + "learning_rate": 1e-06, + "loss": 0.3667, + "mean_token_accuracy": 0.8753129243850708, + "num_tokens": 198923093.0, + "step": 5122 + }, + { + "epoch": 0.6516982572191833, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.096355438232422, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8785206079483032, + "num_tokens": 198966587.0, + "step": 5123 + }, + { + "epoch": 0.6518254674977738, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.101530075073242, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8752732276916504, + "num_tokens": 199005071.0, + "step": 5124 + }, + { + "epoch": 0.6519526777763643, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.14428424835205, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8836020827293396, + "num_tokens": 199040429.0, + "step": 5125 + }, + { + "epoch": 0.6520798880549549, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.088683128356934, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8729413747787476, + "num_tokens": 199083101.0, + "step": 5126 + }, + { + "epoch": 0.6522070983335454, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.107789993286133, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.882492184638977, + "num_tokens": 199125797.0, + "step": 5127 + }, + { + "epoch": 0.6523343086121358, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.10798168182373, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8792508840560913, + "num_tokens": 199166797.0, + "step": 5128 + }, + { + "epoch": 0.6524615188907263, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.125812530517578, + "learning_rate": 1e-06, + "loss": 0.3832, + "mean_token_accuracy": 0.8716704845428467, + "num_tokens": 199203998.0, + "step": 5129 + }, + { + "epoch": 0.6525887291693169, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.213577270507812, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8747243881225586, + "num_tokens": 199240337.0, + "step": 5130 + }, + { + "epoch": 0.6527159394479074, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.080986022949219, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8876573443412781, + "num_tokens": 199276955.0, + "step": 5131 + }, + { + "epoch": 0.6528431497264979, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.083807945251465, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8812034130096436, + "num_tokens": 199322770.0, + "step": 5132 + }, + { + "epoch": 0.6529703600050885, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.031257629394531, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8792821168899536, + "num_tokens": 199365024.0, + "step": 5133 + }, + { + "epoch": 0.6530975702836789, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.145917892456055, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.8728307485580444, + "num_tokens": 199405424.0, + "step": 5134 + }, + { + "epoch": 0.6532247805622694, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.107972145080566, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8782630562782288, + "num_tokens": 199441893.0, + "step": 5135 + }, + { + "epoch": 0.6533519908408599, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.107138633728027, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8803757429122925, + "num_tokens": 199482804.0, + "step": 5136 + }, + { + "epoch": 0.6534792011194505, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.187838554382324, + "learning_rate": 1e-06, + "loss": 0.3963, + "mean_token_accuracy": 0.8692463636398315, + "num_tokens": 199518716.0, + "step": 5137 + }, + { + "epoch": 0.653606411398041, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.13796615600586, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8804266452789307, + "num_tokens": 199556488.0, + "step": 5138 + }, + { + "epoch": 0.6537336216766315, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.163281440734863, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8752920031547546, + "num_tokens": 199594439.0, + "step": 5139 + }, + { + "epoch": 0.6538608319552219, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.08168888092041, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8827240467071533, + "num_tokens": 199633357.0, + "step": 5140 + }, + { + "epoch": 0.6539880422338125, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.050759315490723, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.874485969543457, + "num_tokens": 199684197.0, + "step": 5141 + }, + { + "epoch": 0.654115252512403, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.082256317138672, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8921791315078735, + "num_tokens": 199722109.0, + "step": 5142 + }, + { + "epoch": 0.6542424627909935, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.08201789855957, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8821631073951721, + "num_tokens": 199761868.0, + "step": 5143 + }, + { + "epoch": 0.654369673069584, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.093399047851562, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8861417174339294, + "num_tokens": 199801497.0, + "step": 5144 + }, + { + "epoch": 0.6544968833481746, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.093348503112793, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8904215693473816, + "num_tokens": 199840167.0, + "step": 5145 + }, + { + "epoch": 0.654624093626765, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.11285400390625, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8838541507720947, + "num_tokens": 199879920.0, + "step": 5146 + }, + { + "epoch": 0.6547513039053555, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.115772247314453, + "learning_rate": 1e-06, + "loss": 0.3882, + "mean_token_accuracy": 0.8709923028945923, + "num_tokens": 199922020.0, + "step": 5147 + }, + { + "epoch": 0.654878514183946, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.047781944274902, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8829596638679504, + "num_tokens": 199966117.0, + "step": 5148 + }, + { + "epoch": 0.6550057244625366, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.11606216430664, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8727849125862122, + "num_tokens": 200007119.0, + "step": 5149 + }, + { + "epoch": 0.6551329347411271, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.270637512207031, + "learning_rate": 1e-06, + "loss": 0.3892, + "mean_token_accuracy": 0.866000771522522, + "num_tokens": 200038215.0, + "step": 5150 + }, + { + "epoch": 0.6552601450197176, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.20136833190918, + "learning_rate": 1e-06, + "loss": 0.3865, + "mean_token_accuracy": 0.8710812330245972, + "num_tokens": 200072947.0, + "step": 5151 + }, + { + "epoch": 0.655387355298308, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.203065872192383, + "learning_rate": 1e-06, + "loss": 0.3943, + "mean_token_accuracy": 0.8684531450271606, + "num_tokens": 200109136.0, + "step": 5152 + }, + { + "epoch": 0.6555145655768986, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 7.978596210479736, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8890377283096313, + "num_tokens": 200154014.0, + "step": 5153 + }, + { + "epoch": 0.6556417758554891, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.071011543273926, + "learning_rate": 1e-06, + "loss": 0.4007, + "mean_token_accuracy": 0.865115225315094, + "num_tokens": 200198057.0, + "step": 5154 + }, + { + "epoch": 0.6557689861340796, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.22713851928711, + "learning_rate": 1e-06, + "loss": 0.4056, + "mean_token_accuracy": 0.865498960018158, + "num_tokens": 200235250.0, + "step": 5155 + }, + { + "epoch": 0.6558961964126702, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.102067947387695, + "learning_rate": 1e-06, + "loss": 0.4178, + "mean_token_accuracy": 0.8593481183052063, + "num_tokens": 200277280.0, + "step": 5156 + }, + { + "epoch": 0.6560234066912607, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.18535327911377, + "learning_rate": 1e-06, + "loss": 0.4264, + "mean_token_accuracy": 0.8568607568740845, + "num_tokens": 200314217.0, + "step": 5157 + }, + { + "epoch": 0.6561506169698512, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.119770050048828, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8896574974060059, + "num_tokens": 200347486.0, + "step": 5158 + }, + { + "epoch": 0.6562778272484416, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.164755821228027, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.870341420173645, + "num_tokens": 200382439.0, + "step": 5159 + }, + { + "epoch": 0.6564050375270322, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.233916282653809, + "learning_rate": 1e-06, + "loss": 0.3953, + "mean_token_accuracy": 0.8726319670677185, + "num_tokens": 200417503.0, + "step": 5160 + }, + { + "epoch": 0.6565322478056227, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.117758750915527, + "learning_rate": 1e-06, + "loss": 0.3834, + "mean_token_accuracy": 0.8695542216300964, + "num_tokens": 200458878.0, + "step": 5161 + }, + { + "epoch": 0.6566594580842132, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.148103713989258, + "learning_rate": 1e-06, + "loss": 0.3941, + "mean_token_accuracy": 0.8691294193267822, + "num_tokens": 200498239.0, + "step": 5162 + }, + { + "epoch": 0.6567866683628037, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.234277725219727, + "learning_rate": 1e-06, + "loss": 0.4234, + "mean_token_accuracy": 0.8544193506240845, + "num_tokens": 200538394.0, + "step": 5163 + }, + { + "epoch": 0.6569138786413943, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.040326118469238, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8780418634414673, + "num_tokens": 200576894.0, + "step": 5164 + }, + { + "epoch": 0.6570410889199847, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3484230041503906e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.06918716430664, + "learning_rate": 1e-06, + "loss": 0.413, + "mean_token_accuracy": 0.862619936466217, + "num_tokens": 200620165.0, + "step": 5165 + }, + { + "epoch": 0.6571682991985752, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.138986587524414, + "learning_rate": 1e-06, + "loss": 0.4241, + "mean_token_accuracy": 0.8568721413612366, + "num_tokens": 200661694.0, + "step": 5166 + }, + { + "epoch": 0.6572955094771658, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.169565200805664, + "learning_rate": 1e-06, + "loss": 0.3897, + "mean_token_accuracy": 0.8688749074935913, + "num_tokens": 200701459.0, + "step": 5167 + }, + { + "epoch": 0.6574227197557563, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.127875328063965, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8787891864776611, + "num_tokens": 200738621.0, + "step": 5168 + }, + { + "epoch": 0.6575499300343468, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.168707847595215, + "learning_rate": 1e-06, + "loss": 0.384, + "mean_token_accuracy": 0.8685035705566406, + "num_tokens": 200775985.0, + "step": 5169 + }, + { + "epoch": 0.6576771403129373, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.089865684509277, + "learning_rate": 1e-06, + "loss": 0.4052, + "mean_token_accuracy": 0.8672921657562256, + "num_tokens": 200823631.0, + "step": 5170 + }, + { + "epoch": 0.6578043505915278, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.287921905517578, + "learning_rate": 1e-06, + "loss": 0.3951, + "mean_token_accuracy": 0.8664872050285339, + "num_tokens": 200855510.0, + "step": 5171 + }, + { + "epoch": 0.6579315608701183, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.151803016662598, + "learning_rate": 1e-06, + "loss": 0.4026, + "mean_token_accuracy": 0.8652985692024231, + "num_tokens": 200899049.0, + "step": 5172 + }, + { + "epoch": 0.6580587711487088, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.158394813537598, + "learning_rate": 1e-06, + "loss": 0.3729, + "mean_token_accuracy": 0.875130295753479, + "num_tokens": 200937517.0, + "step": 5173 + }, + { + "epoch": 0.6581859814272993, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.130664825439453, + "learning_rate": 1e-06, + "loss": 0.3888, + "mean_token_accuracy": 0.8684964179992676, + "num_tokens": 200983072.0, + "step": 5174 + }, + { + "epoch": 0.6583131917058899, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.155461311340332, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8930840492248535, + "num_tokens": 201012578.0, + "step": 5175 + }, + { + "epoch": 0.6584404019844804, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.060768127441406, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8782398104667664, + "num_tokens": 201054666.0, + "step": 5176 + }, + { + "epoch": 0.6585676122630708, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.139336585998535, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8899792432785034, + "num_tokens": 201091703.0, + "step": 5177 + }, + { + "epoch": 0.6586948225416613, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.130358695983887, + "learning_rate": 1e-06, + "loss": 0.3816, + "mean_token_accuracy": 0.8705554008483887, + "num_tokens": 201133124.0, + "step": 5178 + }, + { + "epoch": 0.6588220328202519, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.038013458251953, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.897700309753418, + "num_tokens": 201173935.0, + "step": 5179 + }, + { + "epoch": 0.6589492430988424, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.21878433227539, + "learning_rate": 1e-06, + "loss": 0.3859, + "mean_token_accuracy": 0.8708029389381409, + "num_tokens": 201212930.0, + "step": 5180 + }, + { + "epoch": 0.6590764533774329, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.347580909729004, + "learning_rate": 1e-06, + "loss": 0.4009, + "mean_token_accuracy": 0.8667346835136414, + "num_tokens": 201239974.0, + "step": 5181 + }, + { + "epoch": 0.6592036636560235, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.139878273010254, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.8760831356048584, + "num_tokens": 201275257.0, + "step": 5182 + }, + { + "epoch": 0.6593308739346139, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.077468872070312, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8837145566940308, + "num_tokens": 201313859.0, + "step": 5183 + }, + { + "epoch": 0.6594580842132044, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.21591567993164, + "learning_rate": 1e-06, + "loss": 0.3853, + "mean_token_accuracy": 0.8671015501022339, + "num_tokens": 201350786.0, + "step": 5184 + }, + { + "epoch": 0.6595852944917949, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.178109169006348, + "learning_rate": 1e-06, + "loss": 0.4014, + "mean_token_accuracy": 0.8626376390457153, + "num_tokens": 201388285.0, + "step": 5185 + }, + { + "epoch": 0.6597125047703855, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.061505317687988, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8745984435081482, + "num_tokens": 201432831.0, + "step": 5186 + }, + { + "epoch": 0.659839715048976, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.160473823547363, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8806662559509277, + "num_tokens": 201474832.0, + "step": 5187 + }, + { + "epoch": 0.6599669253275665, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.120684623718262, + "learning_rate": 1e-06, + "loss": 0.3953, + "mean_token_accuracy": 0.8682241439819336, + "num_tokens": 201515137.0, + "step": 5188 + }, + { + "epoch": 0.6600941356061569, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.180163383483887, + "learning_rate": 1e-06, + "loss": 0.3817, + "mean_token_accuracy": 0.8732780814170837, + "num_tokens": 201553215.0, + "step": 5189 + }, + { + "epoch": 0.6602213458847475, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.160359382629395, + "learning_rate": 1e-06, + "loss": 0.4028, + "mean_token_accuracy": 0.8662028908729553, + "num_tokens": 201591504.0, + "step": 5190 + }, + { + "epoch": 0.660348556163338, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.12770938873291, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8762478828430176, + "num_tokens": 201628845.0, + "step": 5191 + }, + { + "epoch": 0.6604757664419285, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.107029914855957, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8733460903167725, + "num_tokens": 201665799.0, + "step": 5192 + }, + { + "epoch": 0.660602976720519, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.065093994140625, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8904799222946167, + "num_tokens": 201708108.0, + "step": 5193 + }, + { + "epoch": 0.6607301869991096, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.375166893005371, + "learning_rate": 1e-06, + "loss": 0.3939, + "mean_token_accuracy": 0.8678591847419739, + "num_tokens": 201739607.0, + "step": 5194 + }, + { + "epoch": 0.6608573972777, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.261407852172852, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8787617087364197, + "num_tokens": 201770491.0, + "step": 5195 + }, + { + "epoch": 0.6609846075562905, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 7.96904182434082, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8867103457450867, + "num_tokens": 201814769.0, + "step": 5196 + }, + { + "epoch": 0.661111817834881, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.12830924987793, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8787864446640015, + "num_tokens": 201855097.0, + "step": 5197 + }, + { + "epoch": 0.6612390281134716, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.159330368041992, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8760986328125, + "num_tokens": 201894076.0, + "step": 5198 + }, + { + "epoch": 0.6613662383920621, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.063440322875977, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8811805248260498, + "num_tokens": 201931873.0, + "step": 5199 + }, + { + "epoch": 0.6614934486706526, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.136966705322266, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8751375675201416, + "num_tokens": 201970638.0, + "step": 5200 + }, + { + "epoch": 0.661620658949243, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.061700820922852, + "learning_rate": 1e-06, + "loss": 0.3768, + "mean_token_accuracy": 0.8725709319114685, + "num_tokens": 202014156.0, + "step": 5201 + }, + { + "epoch": 0.6617478692278336, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.125826835632324, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.86952805519104, + "num_tokens": 202055006.0, + "step": 5202 + }, + { + "epoch": 0.6618750795064241, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.087807655334473, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8792856335639954, + "num_tokens": 202095057.0, + "step": 5203 + }, + { + "epoch": 0.6620022897850146, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.268447875976562, + "learning_rate": 1e-06, + "loss": 0.4577, + "mean_token_accuracy": 0.8457674980163574, + "num_tokens": 202132320.0, + "step": 5204 + }, + { + "epoch": 0.6621295000636052, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.16087532043457, + "learning_rate": 1e-06, + "loss": 0.4365, + "mean_token_accuracy": 0.8564580678939819, + "num_tokens": 202174613.0, + "step": 5205 + }, + { + "epoch": 0.6622567103421957, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.053641319274902, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8933091163635254, + "num_tokens": 202218924.0, + "step": 5206 + }, + { + "epoch": 0.6623839206207861, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.102718353271484, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8789721131324768, + "num_tokens": 202261354.0, + "step": 5207 + }, + { + "epoch": 0.6625111308993766, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.086479187011719, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8808699250221252, + "num_tokens": 202298659.0, + "step": 5208 + }, + { + "epoch": 0.6626383411779672, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.056488037109375, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8859785795211792, + "num_tokens": 202340516.0, + "step": 5209 + }, + { + "epoch": 0.6627655514565577, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.157800674438477, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8791203498840332, + "num_tokens": 202378107.0, + "step": 5210 + }, + { + "epoch": 0.6628927617351482, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.15479850769043, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8760732412338257, + "num_tokens": 202413025.0, + "step": 5211 + }, + { + "epoch": 0.6630199720137387, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.124037742614746, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8847010135650635, + "num_tokens": 202450836.0, + "step": 5212 + }, + { + "epoch": 0.6631471822923293, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.158199310302734, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8809189796447754, + "num_tokens": 202489994.0, + "step": 5213 + }, + { + "epoch": 0.6632743925709197, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.057206153869629, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8803697824478149, + "num_tokens": 202529198.0, + "step": 5214 + }, + { + "epoch": 0.6634016028495102, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.3603439331054688e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.106375694274902, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8761357069015503, + "num_tokens": 202567792.0, + "step": 5215 + }, + { + "epoch": 0.6635288131281007, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.107288360595703, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8865852355957031, + "num_tokens": 202610324.0, + "step": 5216 + }, + { + "epoch": 0.6636560234066913, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.094265937805176, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8803133368492126, + "num_tokens": 202648857.0, + "step": 5217 + }, + { + "epoch": 0.6637832336852818, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.222665786743164, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.875879168510437, + "num_tokens": 202689554.0, + "step": 5218 + }, + { + "epoch": 0.6639104439638723, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.092199325561523, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8830543160438538, + "num_tokens": 202729200.0, + "step": 5219 + }, + { + "epoch": 0.6640376542424627, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.143668174743652, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8763667345046997, + "num_tokens": 202765976.0, + "step": 5220 + }, + { + "epoch": 0.6641648645210533, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.11075210571289, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.884040892124176, + "num_tokens": 202803604.0, + "step": 5221 + }, + { + "epoch": 0.6642920747996438, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.121070861816406, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8776923418045044, + "num_tokens": 202842008.0, + "step": 5222 + }, + { + "epoch": 0.6644192850782343, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.133045196533203, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.878029465675354, + "num_tokens": 202881906.0, + "step": 5223 + }, + { + "epoch": 0.6645464953568249, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.151978492736816, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8845423460006714, + "num_tokens": 202917522.0, + "step": 5224 + }, + { + "epoch": 0.6646737056354154, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.144648551940918, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8802797198295593, + "num_tokens": 202957432.0, + "step": 5225 + }, + { + "epoch": 0.6648009159140058, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.14631462097168, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8842039704322815, + "num_tokens": 202993116.0, + "step": 5226 + }, + { + "epoch": 0.6649281261925963, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.019845008850098, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8818105459213257, + "num_tokens": 203038475.0, + "step": 5227 + }, + { + "epoch": 0.6650553364711869, + "ewc_loss": 0.0111083984375, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 8.249382972717285, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8925267457962036, + "num_tokens": 203066442.0, + "step": 5228 + }, + { + "epoch": 0.6651825467497774, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.194897651672363, + "learning_rate": 1e-06, + "loss": 0.3576, + "mean_token_accuracy": 0.8802759647369385, + "num_tokens": 203101284.0, + "step": 5229 + }, + { + "epoch": 0.6653097570283679, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.031014442443848, + "learning_rate": 1e-06, + "loss": 0.3965, + "mean_token_accuracy": 0.8659301400184631, + "num_tokens": 203146541.0, + "step": 5230 + }, + { + "epoch": 0.6654369673069584, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.120590209960938, + "learning_rate": 1e-06, + "loss": 0.3793, + "mean_token_accuracy": 0.8726043701171875, + "num_tokens": 203187580.0, + "step": 5231 + }, + { + "epoch": 0.6655641775855489, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.072381973266602, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8760088682174683, + "num_tokens": 203229988.0, + "step": 5232 + }, + { + "epoch": 0.6656913878641394, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.143778800964355, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8817179203033447, + "num_tokens": 203264428.0, + "step": 5233 + }, + { + "epoch": 0.6658185981427299, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.121906280517578, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8833717107772827, + "num_tokens": 203302728.0, + "step": 5234 + }, + { + "epoch": 0.6659458084213205, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.075301170349121, + "learning_rate": 1e-06, + "loss": 0.4009, + "mean_token_accuracy": 0.8649736046791077, + "num_tokens": 203345446.0, + "step": 5235 + }, + { + "epoch": 0.666073018699911, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.140265464782715, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8852407336235046, + "num_tokens": 203383513.0, + "step": 5236 + }, + { + "epoch": 0.6662002289785015, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.115805625915527, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.876727283000946, + "num_tokens": 203419310.0, + "step": 5237 + }, + { + "epoch": 0.6663274392570919, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.136361122131348, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8748887777328491, + "num_tokens": 203458645.0, + "step": 5238 + }, + { + "epoch": 0.6664546495356825, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.090801239013672, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8850091695785522, + "num_tokens": 203497693.0, + "step": 5239 + }, + { + "epoch": 0.666581859814273, + "ewc_loss": 0.01116943359375, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.142525672912598, + "learning_rate": 1e-06, + "loss": 0.3867, + "mean_token_accuracy": 0.8708215951919556, + "num_tokens": 203536340.0, + "step": 5240 + }, + { + "epoch": 0.6667090700928635, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.14689826965332, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.882605254650116, + "num_tokens": 203572110.0, + "step": 5241 + }, + { + "epoch": 0.666836280371454, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.169116020202637, + "learning_rate": 1e-06, + "loss": 0.4424, + "mean_token_accuracy": 0.849996030330658, + "num_tokens": 203616042.0, + "step": 5242 + }, + { + "epoch": 0.6669634906500446, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.161195755004883, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8794373869895935, + "num_tokens": 203653058.0, + "step": 5243 + }, + { + "epoch": 0.667090700928635, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.142008781433105, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8743680715560913, + "num_tokens": 203691254.0, + "step": 5244 + }, + { + "epoch": 0.6672179112072255, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.178909301757812, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.8769625425338745, + "num_tokens": 203727706.0, + "step": 5245 + }, + { + "epoch": 0.667345121485816, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.184125900268555, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.8738263845443726, + "num_tokens": 203763294.0, + "step": 5246 + }, + { + "epoch": 0.6674723317644066, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.201761245727539, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8915475010871887, + "num_tokens": 203800063.0, + "step": 5247 + }, + { + "epoch": 0.6675995420429971, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.155085563659668, + "learning_rate": 1e-06, + "loss": 0.3915, + "mean_token_accuracy": 0.8658794164657593, + "num_tokens": 203840572.0, + "step": 5248 + }, + { + "epoch": 0.6677267523215876, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.251266479492188, + "learning_rate": 1e-06, + "loss": 0.4023, + "mean_token_accuracy": 0.8641031980514526, + "num_tokens": 203889810.0, + "step": 5249 + }, + { + "epoch": 0.667853962600178, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.156136512756348, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8739581108093262, + "num_tokens": 203929883.0, + "step": 5250 + }, + { + "epoch": 0.6679811728787686, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.11169147491455, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8943442106246948, + "num_tokens": 203969465.0, + "step": 5251 + }, + { + "epoch": 0.6681083831573591, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.188621520996094, + "learning_rate": 1e-06, + "loss": 0.4276, + "mean_token_accuracy": 0.8581978678703308, + "num_tokens": 204008815.0, + "step": 5252 + }, + { + "epoch": 0.6682355934359496, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.183207511901855, + "learning_rate": 1e-06, + "loss": 0.3918, + "mean_token_accuracy": 0.8704853057861328, + "num_tokens": 204045377.0, + "step": 5253 + }, + { + "epoch": 0.6683628037145402, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.107604026794434, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8783777356147766, + "num_tokens": 204084430.0, + "step": 5254 + }, + { + "epoch": 0.6684900139931307, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.098971366882324, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8729739189147949, + "num_tokens": 204126079.0, + "step": 5255 + }, + { + "epoch": 0.6686172242717211, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.144209861755371, + "learning_rate": 1e-06, + "loss": 0.4387, + "mean_token_accuracy": 0.8517550826072693, + "num_tokens": 204171352.0, + "step": 5256 + }, + { + "epoch": 0.6687444345503116, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.037506103515625, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8800266981124878, + "num_tokens": 204211242.0, + "step": 5257 + }, + { + "epoch": 0.6688716448289022, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.208495140075684, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8776466846466064, + "num_tokens": 204245708.0, + "step": 5258 + }, + { + "epoch": 0.6689988551074927, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.092761993408203, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.8786767721176147, + "num_tokens": 204283666.0, + "step": 5259 + }, + { + "epoch": 0.6691260653860832, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.078961372375488, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8792639970779419, + "num_tokens": 204318991.0, + "step": 5260 + }, + { + "epoch": 0.6692532756646737, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.255818367004395, + "learning_rate": 1e-06, + "loss": 0.4035, + "mean_token_accuracy": 0.8637456893920898, + "num_tokens": 204352539.0, + "step": 5261 + }, + { + "epoch": 0.6693804859432643, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.109823226928711, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8849474191665649, + "num_tokens": 204393083.0, + "step": 5262 + }, + { + "epoch": 0.6695076962218547, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.074978828430176, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8738711476325989, + "num_tokens": 204433132.0, + "step": 5263 + }, + { + "epoch": 0.6696349065004452, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.160981178283691, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8876318335533142, + "num_tokens": 204467019.0, + "step": 5264 + }, + { + "epoch": 0.6697621167790357, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.089137077331543, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8889633417129517, + "num_tokens": 204506407.0, + "step": 5265 + }, + { + "epoch": 0.6698893270576263, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.18275260925293, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8864771723747253, + "num_tokens": 204543000.0, + "step": 5266 + }, + { + "epoch": 0.6700165373362168, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.13808822631836, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8750664591789246, + "num_tokens": 204585039.0, + "step": 5267 + }, + { + "epoch": 0.6701437476148073, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.097944259643555, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8856393098831177, + "num_tokens": 204625726.0, + "step": 5268 + }, + { + "epoch": 0.6702709578933977, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.199751853942871, + "learning_rate": 1e-06, + "loss": 0.4144, + "mean_token_accuracy": 0.8599891662597656, + "num_tokens": 204661774.0, + "step": 5269 + }, + { + "epoch": 0.6703981681719883, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.10630989074707, + "learning_rate": 1e-06, + "loss": 0.3933, + "mean_token_accuracy": 0.8657520413398743, + "num_tokens": 204702144.0, + "step": 5270 + }, + { + "epoch": 0.6705253784505788, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.202988624572754, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8811898827552795, + "num_tokens": 204735904.0, + "step": 5271 + }, + { + "epoch": 0.6706525887291693, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.384185791015625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.206082344055176, + "learning_rate": 1e-06, + "loss": 0.4104, + "mean_token_accuracy": 0.8620660305023193, + "num_tokens": 204771321.0, + "step": 5272 + }, + { + "epoch": 0.6707797990077599, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.396106719970703e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.126526832580566, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8739777207374573, + "num_tokens": 204810539.0, + "step": 5273 + }, + { + "epoch": 0.6709070092863504, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.396106719970703e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.197502136230469, + "learning_rate": 1e-06, + "loss": 0.4122, + "mean_token_accuracy": 0.8589828610420227, + "num_tokens": 204848866.0, + "step": 5274 + }, + { + "epoch": 0.6710342195649408, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.396106719970703e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.105509757995605, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8833807110786438, + "num_tokens": 204883591.0, + "step": 5275 + }, + { + "epoch": 0.6711614298435313, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.396106719970703e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.084559440612793, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8787294626235962, + "num_tokens": 204925609.0, + "step": 5276 + }, + { + "epoch": 0.6712886401221219, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.396106719970703e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.150317192077637, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8799782991409302, + "num_tokens": 204963441.0, + "step": 5277 + }, + { + "epoch": 0.6714158504007124, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.396106719970703e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.13314151763916, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.8670506477355957, + "num_tokens": 205009625.0, + "step": 5278 + }, + { + "epoch": 0.6715430606793029, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.396106719970703e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.148109436035156, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8884933590888977, + "num_tokens": 205046151.0, + "step": 5279 + }, + { + "epoch": 0.6716702709578934, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.396106719970703e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.085549354553223, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8860135674476624, + "num_tokens": 205084637.0, + "step": 5280 + }, + { + "epoch": 0.6717974812364839, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.396106719970703e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.15126895904541, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8741241693496704, + "num_tokens": 205124273.0, + "step": 5281 + }, + { + "epoch": 0.6719246915150744, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.396106719970703e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.028443336486816, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8779091238975525, + "num_tokens": 205173274.0, + "step": 5282 + }, + { + "epoch": 0.6720519017936649, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.396106719970703e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.124506950378418, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.881605863571167, + "num_tokens": 205213595.0, + "step": 5283 + }, + { + "epoch": 0.6721791120722554, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.396106719970703e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.232251167297363, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8793428540229797, + "num_tokens": 205246680.0, + "step": 5284 + }, + { + "epoch": 0.672306322350846, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.396106719970703e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.230562210083008, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.8667631149291992, + "num_tokens": 205279910.0, + "step": 5285 + }, + { + "epoch": 0.6724335326294365, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.12449836730957, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.8781823515892029, + "num_tokens": 205318682.0, + "step": 5286 + }, + { + "epoch": 0.6725607429080269, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.097214698791504, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8750618100166321, + "num_tokens": 205357732.0, + "step": 5287 + }, + { + "epoch": 0.6726879531866174, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.062896728515625, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8819510340690613, + "num_tokens": 205404290.0, + "step": 5288 + }, + { + "epoch": 0.672815163465208, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.097329139709473, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8809560537338257, + "num_tokens": 205445829.0, + "step": 5289 + }, + { + "epoch": 0.6729423737437985, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.305237770080566, + "learning_rate": 1e-06, + "loss": 0.4006, + "mean_token_accuracy": 0.864572286605835, + "num_tokens": 205478298.0, + "step": 5290 + }, + { + "epoch": 0.673069584022389, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.103754997253418, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8796336650848389, + "num_tokens": 205516106.0, + "step": 5291 + }, + { + "epoch": 0.6731967943009796, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.167116165161133, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8867214322090149, + "num_tokens": 205548783.0, + "step": 5292 + }, + { + "epoch": 0.67332400457957, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.269960403442383, + "learning_rate": 1e-06, + "loss": 0.3779, + "mean_token_accuracy": 0.8735860586166382, + "num_tokens": 205582062.0, + "step": 5293 + }, + { + "epoch": 0.6734512148581605, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.041999816894531, + "learning_rate": 1e-06, + "loss": 0.3875, + "mean_token_accuracy": 0.8685644865036011, + "num_tokens": 205629705.0, + "step": 5294 + }, + { + "epoch": 0.673578425136751, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.273531913757324, + "learning_rate": 1e-06, + "loss": 0.4331, + "mean_token_accuracy": 0.8652992844581604, + "num_tokens": 205667606.0, + "step": 5295 + }, + { + "epoch": 0.6737056354153416, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.297059059143066, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.887458086013794, + "num_tokens": 205698443.0, + "step": 5296 + }, + { + "epoch": 0.6738328456939321, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.08071517944336, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8816857933998108, + "num_tokens": 205738658.0, + "step": 5297 + }, + { + "epoch": 0.6739600559725226, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.084259986877441, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8794001340866089, + "num_tokens": 205785095.0, + "step": 5298 + }, + { + "epoch": 0.674087266251113, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.144837379455566, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8716081380844116, + "num_tokens": 205823820.0, + "step": 5299 + }, + { + "epoch": 0.6742144765297036, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.215418815612793, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8715523481369019, + "num_tokens": 205861179.0, + "step": 5300 + }, + { + "epoch": 0.6743416868082941, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.152048110961914, + "learning_rate": 1e-06, + "loss": 0.4032, + "mean_token_accuracy": 0.863480806350708, + "num_tokens": 205897220.0, + "step": 5301 + }, + { + "epoch": 0.6744688970868846, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.131949424743652, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8827393054962158, + "num_tokens": 205936366.0, + "step": 5302 + }, + { + "epoch": 0.6745961073654752, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.137812614440918, + "learning_rate": 1e-06, + "loss": 0.3864, + "mean_token_accuracy": 0.8711008429527283, + "num_tokens": 205978715.0, + "step": 5303 + }, + { + "epoch": 0.6747233176440657, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.07701301574707, + "learning_rate": 1e-06, + "loss": 0.3987, + "mean_token_accuracy": 0.8681767582893372, + "num_tokens": 206022747.0, + "step": 5304 + }, + { + "epoch": 0.6748505279226561, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.064403533935547, + "learning_rate": 1e-06, + "loss": 0.3689, + "mean_token_accuracy": 0.875618577003479, + "num_tokens": 206067175.0, + "step": 5305 + }, + { + "epoch": 0.6749777382012466, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.122344017028809, + "learning_rate": 1e-06, + "loss": 0.3798, + "mean_token_accuracy": 0.8759543895721436, + "num_tokens": 206106022.0, + "step": 5306 + }, + { + "epoch": 0.6751049484798372, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.120851516723633, + "learning_rate": 1e-06, + "loss": 0.3771, + "mean_token_accuracy": 0.8733595609664917, + "num_tokens": 206151243.0, + "step": 5307 + }, + { + "epoch": 0.6752321587584277, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.059904098510742, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8731681108474731, + "num_tokens": 206193086.0, + "step": 5308 + }, + { + "epoch": 0.6753593690370182, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.133686065673828, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8753452301025391, + "num_tokens": 206233171.0, + "step": 5309 + }, + { + "epoch": 0.6754865793156087, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.146876335144043, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8880960941314697, + "num_tokens": 206274005.0, + "step": 5310 + }, + { + "epoch": 0.6756137895941993, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.4080276489257812e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.301078796386719, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8801784515380859, + "num_tokens": 206304062.0, + "step": 5311 + }, + { + "epoch": 0.6757409998727897, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4318695068359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.143664360046387, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8770684599876404, + "num_tokens": 206346619.0, + "step": 5312 + }, + { + "epoch": 0.6758682101513802, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.4318695068359375e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.308484077453613, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8795484304428101, + "num_tokens": 206380101.0, + "step": 5313 + }, + { + "epoch": 0.6759954204299707, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4318695068359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.085216522216797, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.8712894320487976, + "num_tokens": 206424251.0, + "step": 5314 + }, + { + "epoch": 0.6761226307085613, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4318695068359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.162640571594238, + "learning_rate": 1e-06, + "loss": 0.412, + "mean_token_accuracy": 0.8621841669082642, + "num_tokens": 206463996.0, + "step": 5315 + }, + { + "epoch": 0.6762498409871518, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4318695068359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.21373176574707, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8862988948822021, + "num_tokens": 206496793.0, + "step": 5316 + }, + { + "epoch": 0.6763770512657423, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4318695068359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.186541557312012, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8810625672340393, + "num_tokens": 206530545.0, + "step": 5317 + }, + { + "epoch": 0.6765042615443327, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4318695068359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.103014945983887, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.875906229019165, + "num_tokens": 206572483.0, + "step": 5318 + }, + { + "epoch": 0.6766314718229233, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4318695068359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.51235294342041, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8780264854431152, + "num_tokens": 206614556.0, + "step": 5319 + }, + { + "epoch": 0.6767586821015138, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.4318695068359375e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.161966323852539, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8886659741401672, + "num_tokens": 206646680.0, + "step": 5320 + }, + { + "epoch": 0.6768858923801043, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4318695068359375e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.176616668701172, + "learning_rate": 1e-06, + "loss": 0.4082, + "mean_token_accuracy": 0.8620204925537109, + "num_tokens": 206681540.0, + "step": 5321 + }, + { + "epoch": 0.6770131026586949, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4318695068359375e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.249859809875488, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.880538821220398, + "num_tokens": 206712861.0, + "step": 5322 + }, + { + "epoch": 0.6771403129372854, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.4318695068359375e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.146711349487305, + "learning_rate": 1e-06, + "loss": 0.4012, + "mean_token_accuracy": 0.8639050722122192, + "num_tokens": 206748589.0, + "step": 5323 + }, + { + "epoch": 0.6772675232158758, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.183282852172852, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8863183856010437, + "num_tokens": 206784450.0, + "step": 5324 + }, + { + "epoch": 0.6773947334944663, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.113198280334473, + "learning_rate": 1e-06, + "loss": 0.3889, + "mean_token_accuracy": 0.8664008378982544, + "num_tokens": 206827903.0, + "step": 5325 + }, + { + "epoch": 0.6775219437730569, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.130400657653809, + "learning_rate": 1e-06, + "loss": 0.3724, + "mean_token_accuracy": 0.8772379159927368, + "num_tokens": 206866136.0, + "step": 5326 + }, + { + "epoch": 0.6776491540516474, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.160910606384277, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.883567214012146, + "num_tokens": 206902507.0, + "step": 5327 + }, + { + "epoch": 0.6777763643302379, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.139605522155762, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8714281916618347, + "num_tokens": 206941964.0, + "step": 5328 + }, + { + "epoch": 0.6779035746088284, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.219525337219238, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8791126012802124, + "num_tokens": 206986062.0, + "step": 5329 + }, + { + "epoch": 0.6780307848874189, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.149052619934082, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8751922249794006, + "num_tokens": 207022328.0, + "step": 5330 + }, + { + "epoch": 0.6781579951660094, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.205124855041504, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8790717124938965, + "num_tokens": 207054553.0, + "step": 5331 + }, + { + "epoch": 0.6782852054445999, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.237460136413574, + "learning_rate": 1e-06, + "loss": 0.4414, + "mean_token_accuracy": 0.854313850402832, + "num_tokens": 207092691.0, + "step": 5332 + }, + { + "epoch": 0.6784124157231904, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.203482627868652, + "learning_rate": 1e-06, + "loss": 0.3995, + "mean_token_accuracy": 0.8639515042304993, + "num_tokens": 207127049.0, + "step": 5333 + }, + { + "epoch": 0.678539626001781, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.078669548034668, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8806314468383789, + "num_tokens": 207166193.0, + "step": 5334 + }, + { + "epoch": 0.6786668362803715, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.115478515625, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8720964193344116, + "num_tokens": 207208207.0, + "step": 5335 + }, + { + "epoch": 0.6787940465589619, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.204617500305176, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.880653977394104, + "num_tokens": 207240097.0, + "step": 5336 + }, + { + "epoch": 0.6789212568375524, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.131673812866211, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8886131048202515, + "num_tokens": 207277540.0, + "step": 5337 + }, + { + "epoch": 0.679048467116143, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.130910873413086, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8832961320877075, + "num_tokens": 207315878.0, + "step": 5338 + }, + { + "epoch": 0.6791756773947335, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.231444358825684, + "learning_rate": 1e-06, + "loss": 0.4184, + "mean_token_accuracy": 0.8594045639038086, + "num_tokens": 207357774.0, + "step": 5339 + }, + { + "epoch": 0.679302887673324, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.24117374420166, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8724810481071472, + "num_tokens": 207394130.0, + "step": 5340 + }, + { + "epoch": 0.6794300979519146, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.106562614440918, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8903590440750122, + "num_tokens": 207431608.0, + "step": 5341 + }, + { + "epoch": 0.679557308230505, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.07580852508545, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8799859285354614, + "num_tokens": 207475499.0, + "step": 5342 + }, + { + "epoch": 0.6796845185090955, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.26220989227295, + "learning_rate": 1e-06, + "loss": 0.4085, + "mean_token_accuracy": 0.864722490310669, + "num_tokens": 207514033.0, + "step": 5343 + }, + { + "epoch": 0.679811728787686, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.231109619140625, + "learning_rate": 1e-06, + "loss": 0.3793, + "mean_token_accuracy": 0.8671135902404785, + "num_tokens": 207546680.0, + "step": 5344 + }, + { + "epoch": 0.6799389390662766, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.174598693847656, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.884406566619873, + "num_tokens": 207583876.0, + "step": 5345 + }, + { + "epoch": 0.6800661493448671, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.144721031188965, + "learning_rate": 1e-06, + "loss": 0.3753, + "mean_token_accuracy": 0.8719951510429382, + "num_tokens": 207624567.0, + "step": 5346 + }, + { + "epoch": 0.6801933596234576, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.083338737487793, + "learning_rate": 1e-06, + "loss": 0.3998, + "mean_token_accuracy": 0.871583104133606, + "num_tokens": 207669506.0, + "step": 5347 + }, + { + "epoch": 0.680320569902048, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.200104713439941, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8855776786804199, + "num_tokens": 207707585.0, + "step": 5348 + }, + { + "epoch": 0.6804477801806386, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.15284252166748, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8738787174224854, + "num_tokens": 207748322.0, + "step": 5349 + }, + { + "epoch": 0.6805749904592291, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.160005569458008, + "learning_rate": 1e-06, + "loss": 0.3786, + "mean_token_accuracy": 0.8704975843429565, + "num_tokens": 207788743.0, + "step": 5350 + }, + { + "epoch": 0.6807022007378196, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.222755432128906, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8764314651489258, + "num_tokens": 207827675.0, + "step": 5351 + }, + { + "epoch": 0.6808294110164101, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.193282127380371, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8793757557868958, + "num_tokens": 207864842.0, + "step": 5352 + }, + { + "epoch": 0.6809566212950007, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.040809631347656, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.889201283454895, + "num_tokens": 207905863.0, + "step": 5353 + }, + { + "epoch": 0.6810838315735911, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.16234302520752, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8781315088272095, + "num_tokens": 207951432.0, + "step": 5354 + }, + { + "epoch": 0.6812110418521816, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.144248008728027, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8766834735870361, + "num_tokens": 207992555.0, + "step": 5355 + }, + { + "epoch": 0.6813382521307721, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.238054275512695, + "learning_rate": 1e-06, + "loss": 0.385, + "mean_token_accuracy": 0.8705437779426575, + "num_tokens": 208030983.0, + "step": 5356 + }, + { + "epoch": 0.6814654624093627, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.253612518310547, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8939691781997681, + "num_tokens": 208062681.0, + "step": 5357 + }, + { + "epoch": 0.6815926726879532, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4437904357910156e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.102200508117676, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8896888494491577, + "num_tokens": 208100922.0, + "step": 5358 + }, + { + "epoch": 0.6817198829665437, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.25796890258789, + "learning_rate": 1e-06, + "loss": 0.3846, + "mean_token_accuracy": 0.8726218938827515, + "num_tokens": 208135106.0, + "step": 5359 + }, + { + "epoch": 0.6818470932451343, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.134316444396973, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8872247338294983, + "num_tokens": 208168717.0, + "step": 5360 + }, + { + "epoch": 0.6819743035237247, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.146349906921387, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8931354284286499, + "num_tokens": 208203428.0, + "step": 5361 + }, + { + "epoch": 0.6821015138023152, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.119545936584473, + "learning_rate": 1e-06, + "loss": 0.3904, + "mean_token_accuracy": 0.8724159002304077, + "num_tokens": 208244883.0, + "step": 5362 + }, + { + "epoch": 0.6822287240809057, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.230088233947754, + "learning_rate": 1e-06, + "loss": 0.4005, + "mean_token_accuracy": 0.8628079891204834, + "num_tokens": 208279173.0, + "step": 5363 + }, + { + "epoch": 0.6823559343594963, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.219460487365723, + "learning_rate": 1e-06, + "loss": 0.3783, + "mean_token_accuracy": 0.873220682144165, + "num_tokens": 208314216.0, + "step": 5364 + }, + { + "epoch": 0.6824831446380868, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.099980354309082, + "learning_rate": 1e-06, + "loss": 0.3944, + "mean_token_accuracy": 0.8671331405639648, + "num_tokens": 208354465.0, + "step": 5365 + }, + { + "epoch": 0.6826103549166773, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.13333511352539, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.8778495788574219, + "num_tokens": 208395097.0, + "step": 5366 + }, + { + "epoch": 0.6827375651952677, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.09599494934082, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8807772397994995, + "num_tokens": 208432309.0, + "step": 5367 + }, + { + "epoch": 0.6828647754738583, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.11251449584961, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8936730623245239, + "num_tokens": 208472264.0, + "step": 5368 + }, + { + "epoch": 0.6829919857524488, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.09407901763916, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.8731614351272583, + "num_tokens": 208514871.0, + "step": 5369 + }, + { + "epoch": 0.6831191960310393, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.096776962280273, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8783594965934753, + "num_tokens": 208557100.0, + "step": 5370 + }, + { + "epoch": 0.6832464063096299, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.088713645935059, + "learning_rate": 1e-06, + "loss": 0.3928, + "mean_token_accuracy": 0.8669100403785706, + "num_tokens": 208599306.0, + "step": 5371 + }, + { + "epoch": 0.6833736165882204, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.142757415771484, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8843483924865723, + "num_tokens": 208638093.0, + "step": 5372 + }, + { + "epoch": 0.6835008268668108, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.154382705688477, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8718826174736023, + "num_tokens": 208674140.0, + "step": 5373 + }, + { + "epoch": 0.6836280371454013, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.098531723022461, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8706328868865967, + "num_tokens": 208714950.0, + "step": 5374 + }, + { + "epoch": 0.6837552474239919, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.055026054382324, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8807547092437744, + "num_tokens": 208755744.0, + "step": 5375 + }, + { + "epoch": 0.6838824577025824, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.165352821350098, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8831067085266113, + "num_tokens": 208793656.0, + "step": 5376 + }, + { + "epoch": 0.6840096679811729, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.12275218963623, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8908119201660156, + "num_tokens": 208830390.0, + "step": 5377 + }, + { + "epoch": 0.6841368782597634, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.07408618927002, + "learning_rate": 1e-06, + "loss": 0.4059, + "mean_token_accuracy": 0.866919219493866, + "num_tokens": 208877918.0, + "step": 5378 + }, + { + "epoch": 0.6842640885383539, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.103556632995605, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.8757891654968262, + "num_tokens": 208918496.0, + "step": 5379 + }, + { + "epoch": 0.6843912988169444, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.059735298156738, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8866133093833923, + "num_tokens": 208959992.0, + "step": 5380 + }, + { + "epoch": 0.6845185090955349, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.312378883361816, + "learning_rate": 1e-06, + "loss": 0.4136, + "mean_token_accuracy": 0.8615230321884155, + "num_tokens": 208992579.0, + "step": 5381 + }, + { + "epoch": 0.6846457193741254, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.186223983764648, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8769538998603821, + "num_tokens": 209030425.0, + "step": 5382 + }, + { + "epoch": 0.684772929652716, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.11439323425293, + "learning_rate": 1e-06, + "loss": 0.3732, + "mean_token_accuracy": 0.8732328414916992, + "num_tokens": 209069495.0, + "step": 5383 + }, + { + "epoch": 0.6849001399313065, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.054595947265625, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8795056343078613, + "num_tokens": 209114331.0, + "step": 5384 + }, + { + "epoch": 0.6850273502098969, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.761882781982422e-06, + "grad_norm": 8.12928295135498, + "learning_rate": 1e-06, + "loss": 0.4023, + "mean_token_accuracy": 0.8633267283439636, + "num_tokens": 209155396.0, + "step": 5385 + }, + { + "epoch": 0.6851545604884874, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.12840461730957, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8841972351074219, + "num_tokens": 209196214.0, + "step": 5386 + }, + { + "epoch": 0.685281770767078, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.13081169128418, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8788048028945923, + "num_tokens": 209230029.0, + "step": 5387 + }, + { + "epoch": 0.6854089810456685, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.228048324584961, + "learning_rate": 1e-06, + "loss": 0.3734, + "mean_token_accuracy": 0.8761249780654907, + "num_tokens": 209265855.0, + "step": 5388 + }, + { + "epoch": 0.685536191324259, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.102967262268066, + "learning_rate": 1e-06, + "loss": 0.3814, + "mean_token_accuracy": 0.872225284576416, + "num_tokens": 209308258.0, + "step": 5389 + }, + { + "epoch": 0.6856634016028496, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.184860229492188, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8734731674194336, + "num_tokens": 209344734.0, + "step": 5390 + }, + { + "epoch": 0.68579061188144, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.153607368469238, + "learning_rate": 1e-06, + "loss": 0.4179, + "mean_token_accuracy": 0.8586152791976929, + "num_tokens": 209389355.0, + "step": 5391 + }, + { + "epoch": 0.6859178221600305, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.101341247558594, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8845217227935791, + "num_tokens": 209429332.0, + "step": 5392 + }, + { + "epoch": 0.686045032438621, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.230498313903809, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8765405416488647, + "num_tokens": 209468916.0, + "step": 5393 + }, + { + "epoch": 0.6861722427172116, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.141440391540527, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.886737585067749, + "num_tokens": 209505999.0, + "step": 5394 + }, + { + "epoch": 0.6862994529958021, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.159564971923828, + "learning_rate": 1e-06, + "loss": 0.3772, + "mean_token_accuracy": 0.8734161257743835, + "num_tokens": 209544885.0, + "step": 5395 + }, + { + "epoch": 0.6864266632743926, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.173365592956543, + "learning_rate": 1e-06, + "loss": 0.3851, + "mean_token_accuracy": 0.8678713440895081, + "num_tokens": 209586357.0, + "step": 5396 + }, + { + "epoch": 0.686553873552983, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.232203483581543, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8754144906997681, + "num_tokens": 209619887.0, + "step": 5397 + }, + { + "epoch": 0.6866810838315736, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.21076774597168, + "learning_rate": 1e-06, + "loss": 0.3818, + "mean_token_accuracy": 0.8699215650558472, + "num_tokens": 209658083.0, + "step": 5398 + }, + { + "epoch": 0.6868082941101641, + "ewc_loss": 0.01129150390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.053296089172363, + "learning_rate": 1e-06, + "loss": 0.3745, + "mean_token_accuracy": 0.873292088508606, + "num_tokens": 209703489.0, + "step": 5399 + }, + { + "epoch": 0.6869355043887546, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.1958646774292, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8806706666946411, + "num_tokens": 209738460.0, + "step": 5400 + }, + { + "epoch": 0.6870627146673451, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.155269622802734, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8740003108978271, + "num_tokens": 209775905.0, + "step": 5401 + }, + { + "epoch": 0.6871899249459357, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.215702056884766, + "learning_rate": 1e-06, + "loss": 0.3916, + "mean_token_accuracy": 0.8676708936691284, + "num_tokens": 209812488.0, + "step": 5402 + }, + { + "epoch": 0.6873171352245261, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.205618858337402, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8763580322265625, + "num_tokens": 209848978.0, + "step": 5403 + }, + { + "epoch": 0.6874443455031166, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.119240760803223, + "learning_rate": 1e-06, + "loss": 0.4112, + "mean_token_accuracy": 0.8597636818885803, + "num_tokens": 209888951.0, + "step": 5404 + }, + { + "epoch": 0.6875715557817071, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.4557113647460938e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.20966911315918, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8821252584457397, + "num_tokens": 209924163.0, + "step": 5405 + }, + { + "epoch": 0.6876987660602977, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.182967185974121, + "learning_rate": 1e-06, + "loss": 0.4185, + "mean_token_accuracy": 0.8616554141044617, + "num_tokens": 209967227.0, + "step": 5406 + }, + { + "epoch": 0.6878259763388882, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.14948558807373, + "learning_rate": 1e-06, + "loss": 0.4082, + "mean_token_accuracy": 0.8646609783172607, + "num_tokens": 210008870.0, + "step": 5407 + }, + { + "epoch": 0.6879531866174787, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.169472694396973, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8954957723617554, + "num_tokens": 210042420.0, + "step": 5408 + }, + { + "epoch": 0.6880803968960693, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.216043472290039, + "learning_rate": 1e-06, + "loss": 0.4038, + "mean_token_accuracy": 0.8612991571426392, + "num_tokens": 210078700.0, + "step": 5409 + }, + { + "epoch": 0.6882076071746597, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.175863265991211, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8762598037719727, + "num_tokens": 210112323.0, + "step": 5410 + }, + { + "epoch": 0.6883348174532502, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.116397857666016, + "learning_rate": 1e-06, + "loss": 0.4001, + "mean_token_accuracy": 0.8662463426589966, + "num_tokens": 210159424.0, + "step": 5411 + }, + { + "epoch": 0.6884620277318407, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.241881370544434, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8906127214431763, + "num_tokens": 210192278.0, + "step": 5412 + }, + { + "epoch": 0.6885892380104313, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.1395263671875, + "learning_rate": 1e-06, + "loss": 0.3949, + "mean_token_accuracy": 0.8645305037498474, + "num_tokens": 210229876.0, + "step": 5413 + }, + { + "epoch": 0.6887164482890218, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.187442779541016, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.880791187286377, + "num_tokens": 210267305.0, + "step": 5414 + }, + { + "epoch": 0.6888436585676123, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.124323844909668, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8799166679382324, + "num_tokens": 210312267.0, + "step": 5415 + }, + { + "epoch": 0.6889708688462027, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.109417915344238, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8731462955474854, + "num_tokens": 210355498.0, + "step": 5416 + }, + { + "epoch": 0.6890980791247933, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.47955322265625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.11646842956543, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8809123039245605, + "num_tokens": 210395277.0, + "step": 5417 + }, + { + "epoch": 0.6892252894033838, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.068156242370605, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8850613832473755, + "num_tokens": 210435108.0, + "step": 5418 + }, + { + "epoch": 0.6893524996819743, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.14659595489502, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.875771164894104, + "num_tokens": 210473438.0, + "step": 5419 + }, + { + "epoch": 0.6894797099605648, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.102339744567871, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8813543915748596, + "num_tokens": 210512586.0, + "step": 5420 + }, + { + "epoch": 0.6896069202391554, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.143065452575684, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8858634233474731, + "num_tokens": 210553166.0, + "step": 5421 + }, + { + "epoch": 0.6897341305177458, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.261414527893066, + "learning_rate": 1e-06, + "loss": 0.4234, + "mean_token_accuracy": 0.8540180921554565, + "num_tokens": 210591223.0, + "step": 5422 + }, + { + "epoch": 0.6898613407963363, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.12869930267334, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8727100491523743, + "num_tokens": 210631613.0, + "step": 5423 + }, + { + "epoch": 0.6899885510749268, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.173775672912598, + "learning_rate": 1e-06, + "loss": 0.3945, + "mean_token_accuracy": 0.8666979074478149, + "num_tokens": 210668192.0, + "step": 5424 + }, + { + "epoch": 0.6901157613535174, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.140244483947754, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.8758542537689209, + "num_tokens": 210709556.0, + "step": 5425 + }, + { + "epoch": 0.6902429716321079, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.186810493469238, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8769351840019226, + "num_tokens": 210747050.0, + "step": 5426 + }, + { + "epoch": 0.6903701819106984, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.080975532531738, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8737794756889343, + "num_tokens": 210790391.0, + "step": 5427 + }, + { + "epoch": 0.6904973921892888, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.226890563964844, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.886135458946228, + "num_tokens": 210824281.0, + "step": 5428 + }, + { + "epoch": 0.6906246024678794, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.07671070098877, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8830874562263489, + "num_tokens": 210866003.0, + "step": 5429 + }, + { + "epoch": 0.6907518127464699, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 10.750033378601074, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.875350832939148, + "num_tokens": 210899249.0, + "step": 5430 + }, + { + "epoch": 0.6908790230250604, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.480082511901855, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8753795027732849, + "num_tokens": 210938542.0, + "step": 5431 + }, + { + "epoch": 0.691006233303651, + "ewc_loss": 0.01123046875, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 8.702278137207031e-06, + "grad_norm": 7.847776889801025, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8711743354797363, + "num_tokens": 210975213.0, + "step": 5432 + }, + { + "epoch": 0.6911334435822415, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.374996185302734, + "learning_rate": 1e-06, + "loss": 0.4066, + "mean_token_accuracy": 0.8631404638290405, + "num_tokens": 211014713.0, + "step": 5433 + }, + { + "epoch": 0.6912606538608319, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.116265296936035, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8888382911682129, + "num_tokens": 211049185.0, + "step": 5434 + }, + { + "epoch": 0.6913878641394224, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.298696517944336, + "learning_rate": 1e-06, + "loss": 0.4169, + "mean_token_accuracy": 0.8600137233734131, + "num_tokens": 211085908.0, + "step": 5435 + }, + { + "epoch": 0.691515074418013, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.349950790405273, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8787635564804077, + "num_tokens": 211117647.0, + "step": 5436 + }, + { + "epoch": 0.6916422846966035, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.124249458312988, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8797473907470703, + "num_tokens": 211158170.0, + "step": 5437 + }, + { + "epoch": 0.691769494975194, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.491474151611328e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.30270767211914, + "learning_rate": 1e-06, + "loss": 0.3986, + "mean_token_accuracy": 0.8689709305763245, + "num_tokens": 211193187.0, + "step": 5438 + }, + { + "epoch": 0.6918967052537845, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.5033950805664062e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.252936363220215, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8819304704666138, + "num_tokens": 211228907.0, + "step": 5439 + }, + { + "epoch": 0.692023915532375, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5033950805664062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.159523010253906, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8796127438545227, + "num_tokens": 211266163.0, + "step": 5440 + }, + { + "epoch": 0.6921511258109655, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5033950805664062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.183995246887207, + "learning_rate": 1e-06, + "loss": 0.3736, + "mean_token_accuracy": 0.8762385845184326, + "num_tokens": 211308555.0, + "step": 5441 + }, + { + "epoch": 0.692278336089556, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5033950805664062e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.20296573638916, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8836385011672974, + "num_tokens": 211341455.0, + "step": 5442 + }, + { + "epoch": 0.6924055463681466, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5033950805664062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.176149368286133, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.9053594470024109, + "num_tokens": 211369736.0, + "step": 5443 + }, + { + "epoch": 0.6925327566467371, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5033950805664062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.195194244384766, + "learning_rate": 1e-06, + "loss": 0.3882, + "mean_token_accuracy": 0.8724068403244019, + "num_tokens": 211406678.0, + "step": 5444 + }, + { + "epoch": 0.6926599669253276, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5033950805664062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.07591724395752, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8865391612052917, + "num_tokens": 211452342.0, + "step": 5445 + }, + { + "epoch": 0.692787177203918, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5033950805664062e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.284849166870117, + "learning_rate": 1e-06, + "loss": 0.3897, + "mean_token_accuracy": 0.8686350584030151, + "num_tokens": 211486164.0, + "step": 5446 + }, + { + "epoch": 0.6929143874825086, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5033950805664062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.890652656555176, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8835918307304382, + "num_tokens": 211525501.0, + "step": 5447 + }, + { + "epoch": 0.6930415977610991, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.5033950805664062e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.21115493774414, + "learning_rate": 1e-06, + "loss": 0.3957, + "mean_token_accuracy": 0.8671707510948181, + "num_tokens": 211561319.0, + "step": 5448 + }, + { + "epoch": 0.6931688080396896, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.218301773071289, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8712977170944214, + "num_tokens": 211593830.0, + "step": 5449 + }, + { + "epoch": 0.6932960183182801, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.25328254699707, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8728441596031189, + "num_tokens": 211626320.0, + "step": 5450 + }, + { + "epoch": 0.6934232285968707, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.086073875427246, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8792864084243774, + "num_tokens": 211666103.0, + "step": 5451 + }, + { + "epoch": 0.6935504388754611, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.226707458496094, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8816313743591309, + "num_tokens": 211703188.0, + "step": 5452 + }, + { + "epoch": 0.6936776491540516, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.172887802124023, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8794205188751221, + "num_tokens": 211736320.0, + "step": 5453 + }, + { + "epoch": 0.6938048594326421, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.29178237915039, + "learning_rate": 1e-06, + "loss": 0.4366, + "mean_token_accuracy": 0.8574682474136353, + "num_tokens": 211771443.0, + "step": 5454 + }, + { + "epoch": 0.6939320697112327, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.127031326293945, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8860328197479248, + "num_tokens": 211811822.0, + "step": 5455 + }, + { + "epoch": 0.6940592799898232, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.126437187194824, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8776000738143921, + "num_tokens": 211852749.0, + "step": 5456 + }, + { + "epoch": 0.6941864902684137, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.162995338439941, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8824640512466431, + "num_tokens": 211892183.0, + "step": 5457 + }, + { + "epoch": 0.6943137005470043, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.19549560546875, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.8764735460281372, + "num_tokens": 211931644.0, + "step": 5458 + }, + { + "epoch": 0.6944409108255947, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.103348731994629, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8890982866287231, + "num_tokens": 211972435.0, + "step": 5459 + }, + { + "epoch": 0.6945681211041852, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.19087028503418, + "learning_rate": 1e-06, + "loss": 0.3884, + "mean_token_accuracy": 0.8698272705078125, + "num_tokens": 212011078.0, + "step": 5460 + }, + { + "epoch": 0.6946953313827757, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.15420150756836, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.8685164451599121, + "num_tokens": 212054304.0, + "step": 5461 + }, + { + "epoch": 0.6948225416613663, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.23807144165039, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8918830752372742, + "num_tokens": 212086144.0, + "step": 5462 + }, + { + "epoch": 0.6949497519399568, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.204046249389648, + "learning_rate": 1e-06, + "loss": 0.4061, + "mean_token_accuracy": 0.8612533807754517, + "num_tokens": 212125804.0, + "step": 5463 + }, + { + "epoch": 0.6950769622185473, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.261177062988281, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8822271823883057, + "num_tokens": 212158864.0, + "step": 5464 + }, + { + "epoch": 0.6952041724971377, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.126946449279785, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8878434896469116, + "num_tokens": 212194484.0, + "step": 5465 + }, + { + "epoch": 0.6953313827757283, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.123661994934082, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8754077553749084, + "num_tokens": 212240071.0, + "step": 5466 + }, + { + "epoch": 0.6954585930543188, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.247749328613281, + "learning_rate": 1e-06, + "loss": 0.3793, + "mean_token_accuracy": 0.8714709281921387, + "num_tokens": 212277698.0, + "step": 5467 + }, + { + "epoch": 0.6955858033329093, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.177345275878906, + "learning_rate": 1e-06, + "loss": 0.4376, + "mean_token_accuracy": 0.8563871383666992, + "num_tokens": 212319762.0, + "step": 5468 + }, + { + "epoch": 0.6957130136114998, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.163820266723633, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8770784139633179, + "num_tokens": 212360131.0, + "step": 5469 + }, + { + "epoch": 0.6958402238900904, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.234399795532227, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.880372166633606, + "num_tokens": 212394680.0, + "step": 5470 + }, + { + "epoch": 0.6959674341686808, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.178975105285645, + "learning_rate": 1e-06, + "loss": 0.416, + "mean_token_accuracy": 0.8637704253196716, + "num_tokens": 212436652.0, + "step": 5471 + }, + { + "epoch": 0.6960946444472713, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.11555290222168, + "learning_rate": 1e-06, + "loss": 0.3962, + "mean_token_accuracy": 0.8674522638320923, + "num_tokens": 212478887.0, + "step": 5472 + }, + { + "epoch": 0.6962218547258618, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.147173881530762, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8763629794120789, + "num_tokens": 212520643.0, + "step": 5473 + }, + { + "epoch": 0.6963490650044524, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.183571815490723, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8846834897994995, + "num_tokens": 212563478.0, + "step": 5474 + }, + { + "epoch": 0.6964762752830429, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.262664794921875, + "learning_rate": 1e-06, + "loss": 0.3671, + "mean_token_accuracy": 0.8783013820648193, + "num_tokens": 212604677.0, + "step": 5475 + }, + { + "epoch": 0.6966034855616334, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.5153160095214844e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.177972793579102, + "learning_rate": 1e-06, + "loss": 0.3848, + "mean_token_accuracy": 0.869713306427002, + "num_tokens": 212642218.0, + "step": 5476 + }, + { + "epoch": 0.6967306958402238, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.207189559936523, + "learning_rate": 1e-06, + "loss": 0.3993, + "mean_token_accuracy": 0.8687199354171753, + "num_tokens": 212682479.0, + "step": 5477 + }, + { + "epoch": 0.6968579061188144, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.135498046875, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8752744197845459, + "num_tokens": 212725236.0, + "step": 5478 + }, + { + "epoch": 0.6969851163974049, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.205120086669922, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.8710938692092896, + "num_tokens": 212762649.0, + "step": 5479 + }, + { + "epoch": 0.6971123266759954, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.25787353515625, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8765074610710144, + "num_tokens": 212799083.0, + "step": 5480 + }, + { + "epoch": 0.697239536954586, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.37000846862793, + "learning_rate": 1e-06, + "loss": 0.3818, + "mean_token_accuracy": 0.8714962601661682, + "num_tokens": 212831331.0, + "step": 5481 + }, + { + "epoch": 0.6973667472331765, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.203591346740723, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.881696879863739, + "num_tokens": 212872262.0, + "step": 5482 + }, + { + "epoch": 0.6974939575117669, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.130264282226562, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8840122222900391, + "num_tokens": 212912128.0, + "step": 5483 + }, + { + "epoch": 0.6976211677903574, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.307059288024902, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8909083604812622, + "num_tokens": 212941000.0, + "step": 5484 + }, + { + "epoch": 0.697748378068948, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.160512924194336, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8731462955474854, + "num_tokens": 212981362.0, + "step": 5485 + }, + { + "epoch": 0.6978755883475385, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.096067428588867, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.891084611415863, + "num_tokens": 213023914.0, + "step": 5486 + }, + { + "epoch": 0.698002798626129, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.22677993774414, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8730834722518921, + "num_tokens": 213060789.0, + "step": 5487 + }, + { + "epoch": 0.6981300089047195, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.136234283447266, + "learning_rate": 1e-06, + "loss": 0.3761, + "mean_token_accuracy": 0.8711380362510681, + "num_tokens": 213106817.0, + "step": 5488 + }, + { + "epoch": 0.69825721918331, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.262569427490234, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8799282908439636, + "num_tokens": 213141074.0, + "step": 5489 + }, + { + "epoch": 0.6983844294619005, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.279900550842285, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8756406307220459, + "num_tokens": 213179388.0, + "step": 5490 + }, + { + "epoch": 0.698511639740491, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.238795280456543, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8722880482673645, + "num_tokens": 213216365.0, + "step": 5491 + }, + { + "epoch": 0.6986388500190815, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.2068510055542, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8792130947113037, + "num_tokens": 213256316.0, + "step": 5492 + }, + { + "epoch": 0.6987660602976721, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.103662490844727, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.881065845489502, + "num_tokens": 213299597.0, + "step": 5493 + }, + { + "epoch": 0.6988932705762626, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.156335830688477, + "learning_rate": 1e-06, + "loss": 0.386, + "mean_token_accuracy": 0.8701608180999756, + "num_tokens": 213335325.0, + "step": 5494 + }, + { + "epoch": 0.699020480854853, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.217838287353516, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8738129138946533, + "num_tokens": 213373766.0, + "step": 5495 + }, + { + "epoch": 0.6991476911334435, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.150298118591309, + "learning_rate": 1e-06, + "loss": 0.3788, + "mean_token_accuracy": 0.871253252029419, + "num_tokens": 213412337.0, + "step": 5496 + }, + { + "epoch": 0.6992749014120341, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.20419979095459, + "learning_rate": 1e-06, + "loss": 0.3896, + "mean_token_accuracy": 0.8698813319206238, + "num_tokens": 213452510.0, + "step": 5497 + }, + { + "epoch": 0.6994021116906246, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.160209655761719, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8812133073806763, + "num_tokens": 213490279.0, + "step": 5498 + }, + { + "epoch": 0.6995293219692151, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.21182918548584, + "learning_rate": 1e-06, + "loss": 0.3753, + "mean_token_accuracy": 0.876896321773529, + "num_tokens": 213528154.0, + "step": 5499 + }, + { + "epoch": 0.6996565322478057, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.156697273254395, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8777496218681335, + "num_tokens": 213567751.0, + "step": 5500 + }, + { + "epoch": 0.6997837425263961, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.120152473449707, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8846734762191772, + "num_tokens": 213608735.0, + "step": 5501 + }, + { + "epoch": 0.6999109528049866, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.239400863647461, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8819306492805481, + "num_tokens": 213644829.0, + "step": 5502 + }, + { + "epoch": 0.7000381630835771, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.06214427947998, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8773307204246521, + "num_tokens": 213689575.0, + "step": 5503 + }, + { + "epoch": 0.7001653733621677, + "ewc_loss": 0.0113525390625, + "ewc_loss_diag": 2.5272369384765625e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.15480899810791, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8772909641265869, + "num_tokens": 213733712.0, + "step": 5504 + }, + { + "epoch": 0.7002925836407582, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.2354097366333, + "learning_rate": 1e-06, + "loss": 0.4065, + "mean_token_accuracy": 0.8658413887023926, + "num_tokens": 213772924.0, + "step": 5505 + }, + { + "epoch": 0.7004197939193487, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.08894157409668, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8843106031417847, + "num_tokens": 213817159.0, + "step": 5506 + }, + { + "epoch": 0.7005470041979391, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.221314430236816, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.8702789545059204, + "num_tokens": 213858187.0, + "step": 5507 + }, + { + "epoch": 0.7006742144765297, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.259773254394531, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8746459484100342, + "num_tokens": 213892477.0, + "step": 5508 + }, + { + "epoch": 0.7008014247551202, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.18998908996582, + "learning_rate": 1e-06, + "loss": 0.3901, + "mean_token_accuracy": 0.8670589923858643, + "num_tokens": 213930341.0, + "step": 5509 + }, + { + "epoch": 0.7009286350337107, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.136481285095215, + "learning_rate": 1e-06, + "loss": 0.3736, + "mean_token_accuracy": 0.8743955492973328, + "num_tokens": 213970488.0, + "step": 5510 + }, + { + "epoch": 0.7010558453123013, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.173589706420898, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8787736892700195, + "num_tokens": 214013805.0, + "step": 5511 + }, + { + "epoch": 0.7011830555908918, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.20468807220459, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8786294460296631, + "num_tokens": 214049177.0, + "step": 5512 + }, + { + "epoch": 0.7013102658694823, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.204437255859375, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.8759064674377441, + "num_tokens": 214085811.0, + "step": 5513 + }, + { + "epoch": 0.7014374761480727, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.105125427246094, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.879871129989624, + "num_tokens": 214130337.0, + "step": 5514 + }, + { + "epoch": 0.7015646864266633, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.245131492614746, + "learning_rate": 1e-06, + "loss": 0.3877, + "mean_token_accuracy": 0.8696123361587524, + "num_tokens": 214164679.0, + "step": 5515 + }, + { + "epoch": 0.7016918967052538, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.155229568481445, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8870874643325806, + "num_tokens": 214203166.0, + "step": 5516 + }, + { + "epoch": 0.7018191069838443, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.210896492004395, + "learning_rate": 1e-06, + "loss": 0.4066, + "mean_token_accuracy": 0.8631267547607422, + "num_tokens": 214244284.0, + "step": 5517 + }, + { + "epoch": 0.7019463172624348, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.20794677734375, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8822737336158752, + "num_tokens": 214279911.0, + "step": 5518 + }, + { + "epoch": 0.7020735275410254, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.181955337524414, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8754101991653442, + "num_tokens": 214316515.0, + "step": 5519 + }, + { + "epoch": 0.7022007378196158, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.219943046569824, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8744138479232788, + "num_tokens": 214352042.0, + "step": 5520 + }, + { + "epoch": 0.7023279480982063, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.16590404510498, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8710694313049316, + "num_tokens": 214387913.0, + "step": 5521 + }, + { + "epoch": 0.7024551583767968, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.22526741027832, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8921259045600891, + "num_tokens": 214419990.0, + "step": 5522 + }, + { + "epoch": 0.7025823686553874, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.065271377563477, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8888041973114014, + "num_tokens": 214460188.0, + "step": 5523 + }, + { + "epoch": 0.7027095789339779, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.209615707397461, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8844249248504639, + "num_tokens": 214496311.0, + "step": 5524 + }, + { + "epoch": 0.7028367892125684, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.131617546081543, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8812137246131897, + "num_tokens": 214540327.0, + "step": 5525 + }, + { + "epoch": 0.7029639994911588, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.221853256225586, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8815972208976746, + "num_tokens": 214577021.0, + "step": 5526 + }, + { + "epoch": 0.7030912097697494, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.142796516418457, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8754916191101074, + "num_tokens": 214620821.0, + "step": 5527 + }, + { + "epoch": 0.7032184200483399, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.103352546691895, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8713905811309814, + "num_tokens": 214662391.0, + "step": 5528 + }, + { + "epoch": 0.7033456303269304, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.188939094543457, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.880486011505127, + "num_tokens": 214701348.0, + "step": 5529 + }, + { + "epoch": 0.703472840605521, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.164325714111328, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8802571296691895, + "num_tokens": 214739096.0, + "step": 5530 + }, + { + "epoch": 0.7036000508841115, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.17238712310791, + "learning_rate": 1e-06, + "loss": 0.3912, + "mean_token_accuracy": 0.8691534399986267, + "num_tokens": 214783463.0, + "step": 5531 + }, + { + "epoch": 0.7037272611627019, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.214764595031738, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8809065222740173, + "num_tokens": 214819253.0, + "step": 5532 + }, + { + "epoch": 0.7038544714412924, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.283841133117676, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8881987929344177, + "num_tokens": 214855870.0, + "step": 5533 + }, + { + "epoch": 0.703981681719883, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.188194274902344, + "learning_rate": 1e-06, + "loss": 0.3869, + "mean_token_accuracy": 0.8689082860946655, + "num_tokens": 214897836.0, + "step": 5534 + }, + { + "epoch": 0.7041088919984735, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.305854797363281, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8765276670455933, + "num_tokens": 214931049.0, + "step": 5535 + }, + { + "epoch": 0.704236102277064, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.185054779052734, + "learning_rate": 1e-06, + "loss": 0.3885, + "mean_token_accuracy": 0.8677366971969604, + "num_tokens": 214971824.0, + "step": 5536 + }, + { + "epoch": 0.7043633125556545, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.158032417297363, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8931877017021179, + "num_tokens": 215014282.0, + "step": 5537 + }, + { + "epoch": 0.704490522834245, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.284982681274414, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8852484822273254, + "num_tokens": 215046934.0, + "step": 5538 + }, + { + "epoch": 0.7046177331128355, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.111492156982422, + "learning_rate": 1e-06, + "loss": 0.381, + "mean_token_accuracy": 0.8723004460334778, + "num_tokens": 215092179.0, + "step": 5539 + }, + { + "epoch": 0.704744943391426, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.181321144104004, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8829404711723328, + "num_tokens": 215137404.0, + "step": 5540 + }, + { + "epoch": 0.7048721536700165, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.346887588500977, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8734766840934753, + "num_tokens": 215176708.0, + "step": 5541 + }, + { + "epoch": 0.7049993639486071, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.190223693847656, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8849511742591858, + "num_tokens": 215216178.0, + "step": 5542 + }, + { + "epoch": 0.7051265742271976, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.21044635772705, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8790620565414429, + "num_tokens": 215251890.0, + "step": 5543 + }, + { + "epoch": 0.705253784505788, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.235697746276855, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.8697563409805298, + "num_tokens": 215290993.0, + "step": 5544 + }, + { + "epoch": 0.7053809947843785, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.073383331298828, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8896622657775879, + "num_tokens": 215333799.0, + "step": 5545 + }, + { + "epoch": 0.7055082050629691, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.199830055236816, + "learning_rate": 1e-06, + "loss": 0.4134, + "mean_token_accuracy": 0.8595969676971436, + "num_tokens": 215373986.0, + "step": 5546 + }, + { + "epoch": 0.7056354153415596, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.23043155670166, + "learning_rate": 1e-06, + "loss": 0.3956, + "mean_token_accuracy": 0.8694642782211304, + "num_tokens": 215416123.0, + "step": 5547 + }, + { + "epoch": 0.7057626256201501, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.074370384216309, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8873155117034912, + "num_tokens": 215454639.0, + "step": 5548 + }, + { + "epoch": 0.7058898358987407, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.30979061126709, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8716773986816406, + "num_tokens": 215496389.0, + "step": 5549 + }, + { + "epoch": 0.7060170461773311, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.196401596069336, + "learning_rate": 1e-06, + "loss": 0.4142, + "mean_token_accuracy": 0.860724687576294, + "num_tokens": 215538464.0, + "step": 5550 + }, + { + "epoch": 0.7061442564559216, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.205464363098145, + "learning_rate": 1e-06, + "loss": 0.425, + "mean_token_accuracy": 0.8560333251953125, + "num_tokens": 215584543.0, + "step": 5551 + }, + { + "epoch": 0.7062714667345121, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.175573348999023, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8772577047348022, + "num_tokens": 215625806.0, + "step": 5552 + }, + { + "epoch": 0.7063986770131027, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.093402862548828, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8810929656028748, + "num_tokens": 215669215.0, + "step": 5553 + }, + { + "epoch": 0.7065258872916932, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.251205444335938, + "learning_rate": 1e-06, + "loss": 0.3849, + "mean_token_accuracy": 0.8701521754264832, + "num_tokens": 215708865.0, + "step": 5554 + }, + { + "epoch": 0.7066530975702837, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.247008323669434, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8716858625411987, + "num_tokens": 215742626.0, + "step": 5555 + }, + { + "epoch": 0.7067803078488741, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.139820098876953, + "learning_rate": 1e-06, + "loss": 0.4027, + "mean_token_accuracy": 0.8700698614120483, + "num_tokens": 215788427.0, + "step": 5556 + }, + { + "epoch": 0.7069075181274647, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.145750999450684, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8737302422523499, + "num_tokens": 215830868.0, + "step": 5557 + }, + { + "epoch": 0.7070347284060552, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.21908950805664, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8782445192337036, + "num_tokens": 215868322.0, + "step": 5558 + }, + { + "epoch": 0.7071619386846457, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.139803886413574, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8706445693969727, + "num_tokens": 215908049.0, + "step": 5559 + }, + { + "epoch": 0.7072891489632362, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.271225929260254, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8757520914077759, + "num_tokens": 215942887.0, + "step": 5560 + }, + { + "epoch": 0.7074163592418268, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.205362319946289, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8836051225662231, + "num_tokens": 215983397.0, + "step": 5561 + }, + { + "epoch": 0.7075435695204173, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5510787963867188e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.102660179138184, + "learning_rate": 1e-06, + "loss": 0.427, + "mean_token_accuracy": 0.8559116125106812, + "num_tokens": 216029875.0, + "step": 5562 + }, + { + "epoch": 0.7076707797990077, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.239850044250488, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8704547882080078, + "num_tokens": 216068734.0, + "step": 5563 + }, + { + "epoch": 0.7077979900775982, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.150504112243652, + "learning_rate": 1e-06, + "loss": 0.3643, + "mean_token_accuracy": 0.8757321834564209, + "num_tokens": 216110202.0, + "step": 5564 + }, + { + "epoch": 0.7079252003561888, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.26655101776123, + "learning_rate": 1e-06, + "loss": 0.4071, + "mean_token_accuracy": 0.8632651567459106, + "num_tokens": 216147198.0, + "step": 5565 + }, + { + "epoch": 0.7080524106347793, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.11982250213623, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.875774621963501, + "num_tokens": 216187405.0, + "step": 5566 + }, + { + "epoch": 0.7081796209133698, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.136852264404297, + "learning_rate": 1e-06, + "loss": 0.3956, + "mean_token_accuracy": 0.8677388429641724, + "num_tokens": 216232974.0, + "step": 5567 + }, + { + "epoch": 0.7083068311919604, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.196159362792969, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8702845573425293, + "num_tokens": 216274596.0, + "step": 5568 + }, + { + "epoch": 0.7084340414705508, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.237723350524902, + "learning_rate": 1e-06, + "loss": 0.3817, + "mean_token_accuracy": 0.8688687086105347, + "num_tokens": 216312431.0, + "step": 5569 + }, + { + "epoch": 0.7085612517491413, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.213302612304688, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8796387910842896, + "num_tokens": 216348305.0, + "step": 5570 + }, + { + "epoch": 0.7086884620277318, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.122238159179688, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8865683674812317, + "num_tokens": 216387630.0, + "step": 5571 + }, + { + "epoch": 0.7088156723063224, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.197479248046875, + "learning_rate": 1e-06, + "loss": 0.399, + "mean_token_accuracy": 0.8682167530059814, + "num_tokens": 216428255.0, + "step": 5572 + }, + { + "epoch": 0.7089428825849129, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.20279598236084, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8835574388504028, + "num_tokens": 216466604.0, + "step": 5573 + }, + { + "epoch": 0.7090700928635034, + "ewc_loss": 0.01141357421875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.25812816619873, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8759198188781738, + "num_tokens": 216500938.0, + "step": 5574 + }, + { + "epoch": 0.7091973031420938, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.24909782409668, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8757510781288147, + "num_tokens": 216538120.0, + "step": 5575 + }, + { + "epoch": 0.7093245134206844, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.162614822387695, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8917858600616455, + "num_tokens": 216576916.0, + "step": 5576 + }, + { + "epoch": 0.7094517236992749, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.130620956420898, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.874252438545227, + "num_tokens": 216618028.0, + "step": 5577 + }, + { + "epoch": 0.7095789339778654, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.234427452087402, + "learning_rate": 1e-06, + "loss": 0.3991, + "mean_token_accuracy": 0.868089497089386, + "num_tokens": 216657749.0, + "step": 5578 + }, + { + "epoch": 0.709706144256456, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.229816436767578, + "learning_rate": 1e-06, + "loss": 0.3963, + "mean_token_accuracy": 0.8687062859535217, + "num_tokens": 216697184.0, + "step": 5579 + }, + { + "epoch": 0.7098333545350465, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.147664070129395, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8795044422149658, + "num_tokens": 216739582.0, + "step": 5580 + }, + { + "epoch": 0.7099605648136369, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.272951126098633, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8817753791809082, + "num_tokens": 216774656.0, + "step": 5581 + }, + { + "epoch": 0.7100877750922274, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.20317554473877, + "learning_rate": 1e-06, + "loss": 0.4097, + "mean_token_accuracy": 0.8644435405731201, + "num_tokens": 216817525.0, + "step": 5582 + }, + { + "epoch": 0.710214985370818, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.28718090057373, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.881361722946167, + "num_tokens": 216851992.0, + "step": 5583 + }, + { + "epoch": 0.7103421956494085, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.283196449279785, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.8745725154876709, + "num_tokens": 216885910.0, + "step": 5584 + }, + { + "epoch": 0.710469405927999, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.127840042114258, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8774834275245667, + "num_tokens": 216930817.0, + "step": 5585 + }, + { + "epoch": 0.7105966162065895, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.259523391723633, + "learning_rate": 1e-06, + "loss": 0.4288, + "mean_token_accuracy": 0.8542518019676208, + "num_tokens": 216971966.0, + "step": 5586 + }, + { + "epoch": 0.71072382648518, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.179408073425293, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8817006349563599, + "num_tokens": 217006145.0, + "step": 5587 + }, + { + "epoch": 0.7108510367637705, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.246241569519043, + "learning_rate": 1e-06, + "loss": 0.3861, + "mean_token_accuracy": 0.8700006008148193, + "num_tokens": 217044160.0, + "step": 5588 + }, + { + "epoch": 0.710978247042361, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.093790054321289, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8808733820915222, + "num_tokens": 217088448.0, + "step": 5589 + }, + { + "epoch": 0.7111054573209515, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.195273399353027, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8834949135780334, + "num_tokens": 217127387.0, + "step": 5590 + }, + { + "epoch": 0.7112326675995421, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.181751251220703, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8808647394180298, + "num_tokens": 217169098.0, + "step": 5591 + }, + { + "epoch": 0.7113598778781326, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.329629898071289, + "learning_rate": 1e-06, + "loss": 0.3844, + "mean_token_accuracy": 0.8671797513961792, + "num_tokens": 217201595.0, + "step": 5592 + }, + { + "epoch": 0.711487088156723, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.213709831237793, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8762747049331665, + "num_tokens": 217238790.0, + "step": 5593 + }, + { + "epoch": 0.7116142984353135, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.171876907348633, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.882128119468689, + "num_tokens": 217272992.0, + "step": 5594 + }, + { + "epoch": 0.7117415087139041, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.346613883972168, + "learning_rate": 1e-06, + "loss": 0.4232, + "mean_token_accuracy": 0.8587559461593628, + "num_tokens": 217309790.0, + "step": 5595 + }, + { + "epoch": 0.7118687189924946, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.167609214782715, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8765493035316467, + "num_tokens": 217345390.0, + "step": 5596 + }, + { + "epoch": 0.7119959292710851, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.194872856140137, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8706928491592407, + "num_tokens": 217383780.0, + "step": 5597 + }, + { + "epoch": 0.7121231395496757, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.157689094543457, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8858596086502075, + "num_tokens": 217424491.0, + "step": 5598 + }, + { + "epoch": 0.7122503498282661, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.29638957977295, + "learning_rate": 1e-06, + "loss": 0.3768, + "mean_token_accuracy": 0.8681708574295044, + "num_tokens": 217457959.0, + "step": 5599 + }, + { + "epoch": 0.7123775601068566, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.102997779846191, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8923715949058533, + "num_tokens": 217496452.0, + "step": 5600 + }, + { + "epoch": 0.7125047703854471, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.336992263793945, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8697749376296997, + "num_tokens": 217529466.0, + "step": 5601 + }, + { + "epoch": 0.7126319806640377, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.280280113220215, + "learning_rate": 1e-06, + "loss": 0.4069, + "mean_token_accuracy": 0.8634504079818726, + "num_tokens": 217565827.0, + "step": 5602 + }, + { + "epoch": 0.7127591909426282, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.12280559539795, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8781989216804504, + "num_tokens": 217606860.0, + "step": 5603 + }, + { + "epoch": 0.7128864012212187, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.134446144104004, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8918182849884033, + "num_tokens": 217651918.0, + "step": 5604 + }, + { + "epoch": 0.7130136114998091, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.173628807067871, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.8734326362609863, + "num_tokens": 217694907.0, + "step": 5605 + }, + { + "epoch": 0.7131408217783997, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.209497451782227, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.878509521484375, + "num_tokens": 217735656.0, + "step": 5606 + }, + { + "epoch": 0.7132680320569902, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.202892303466797, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8780276775360107, + "num_tokens": 217775749.0, + "step": 5607 + }, + { + "epoch": 0.7133952423355807, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.304435729980469, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.8735297918319702, + "num_tokens": 217811636.0, + "step": 5608 + }, + { + "epoch": 0.7135224526141712, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.206165313720703, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8805636167526245, + "num_tokens": 217851500.0, + "step": 5609 + }, + { + "epoch": 0.7136496628927618, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.233649253845215, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8756310939788818, + "num_tokens": 217890497.0, + "step": 5610 + }, + { + "epoch": 0.7137768731713523, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.246955871582031, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.8798759579658508, + "num_tokens": 217926390.0, + "step": 5611 + }, + { + "epoch": 0.7139040834499427, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.14007568359375, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8838422894477844, + "num_tokens": 217966200.0, + "step": 5612 + }, + { + "epoch": 0.7140312937285332, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.418367385864258, + "learning_rate": 1e-06, + "loss": 0.3999, + "mean_token_accuracy": 0.8628596067428589, + "num_tokens": 217997879.0, + "step": 5613 + }, + { + "epoch": 0.7141585040071238, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.286619186401367, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8734526634216309, + "num_tokens": 218035028.0, + "step": 5614 + }, + { + "epoch": 0.7142857142857143, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.156244277954102, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8890091180801392, + "num_tokens": 218070824.0, + "step": 5615 + }, + { + "epoch": 0.7144129245643048, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.108112335205078, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8724684715270996, + "num_tokens": 218113349.0, + "step": 5616 + }, + { + "epoch": 0.7145401348428954, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.17372989654541, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8830037117004395, + "num_tokens": 218152169.0, + "step": 5617 + }, + { + "epoch": 0.7146673451214858, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.0960693359375, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8819942474365234, + "num_tokens": 218197046.0, + "step": 5618 + }, + { + "epoch": 0.7147945554000763, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.211800575256348, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8776180744171143, + "num_tokens": 218235018.0, + "step": 5619 + }, + { + "epoch": 0.7149217656786668, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.207218170166016, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.876404881477356, + "num_tokens": 218274355.0, + "step": 5620 + }, + { + "epoch": 0.7150489759572574, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.167254447937012, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.880821943283081, + "num_tokens": 218314512.0, + "step": 5621 + }, + { + "epoch": 0.7151761862358479, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.162107467651367, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.891730546951294, + "num_tokens": 218354623.0, + "step": 5622 + }, + { + "epoch": 0.7153033965144384, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.273338317871094, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8803974390029907, + "num_tokens": 218394674.0, + "step": 5623 + }, + { + "epoch": 0.7154306067930288, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.246810913085938, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8735754489898682, + "num_tokens": 218429116.0, + "step": 5624 + }, + { + "epoch": 0.7155578170716194, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.192259788513184, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8825414776802063, + "num_tokens": 218463395.0, + "step": 5625 + }, + { + "epoch": 0.7156850273502099, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.218859672546387, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8838053345680237, + "num_tokens": 218496845.0, + "step": 5626 + }, + { + "epoch": 0.7158122376288004, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.174589157104492, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8818202614784241, + "num_tokens": 218534543.0, + "step": 5627 + }, + { + "epoch": 0.715939447907391, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.167303085327148, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8657379150390625, + "num_tokens": 218573962.0, + "step": 5628 + }, + { + "epoch": 0.7160666581859815, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.154426574707031, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8814250230789185, + "num_tokens": 218612413.0, + "step": 5629 + }, + { + "epoch": 0.7161938684645719, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.562999725341797e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.295140266418457, + "learning_rate": 1e-06, + "loss": 0.421, + "mean_token_accuracy": 0.8604718446731567, + "num_tokens": 218651959.0, + "step": 5630 + }, + { + "epoch": 0.7163210787431624, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.250687599182129, + "learning_rate": 1e-06, + "loss": 0.4052, + "mean_token_accuracy": 0.8694198727607727, + "num_tokens": 218690962.0, + "step": 5631 + }, + { + "epoch": 0.716448289021753, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.08044147491455, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8816608190536499, + "num_tokens": 218735232.0, + "step": 5632 + }, + { + "epoch": 0.7165754993003435, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.197518348693848, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.879469633102417, + "num_tokens": 218770455.0, + "step": 5633 + }, + { + "epoch": 0.716702709578934, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.18946647644043, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8777283430099487, + "num_tokens": 218809470.0, + "step": 5634 + }, + { + "epoch": 0.7168299198575245, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.18525218963623, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8864718675613403, + "num_tokens": 218847058.0, + "step": 5635 + }, + { + "epoch": 0.716957130136115, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.254661560058594, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8824565410614014, + "num_tokens": 218879997.0, + "step": 5636 + }, + { + "epoch": 0.7170843404147055, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.190914154052734, + "learning_rate": 1e-06, + "loss": 0.3738, + "mean_token_accuracy": 0.8770694732666016, + "num_tokens": 218921539.0, + "step": 5637 + }, + { + "epoch": 0.717211550693296, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.224535942077637, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8856198191642761, + "num_tokens": 218955056.0, + "step": 5638 + }, + { + "epoch": 0.7173387609718865, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.320270538330078, + "learning_rate": 1e-06, + "loss": 0.4229, + "mean_token_accuracy": 0.8626736402511597, + "num_tokens": 218992787.0, + "step": 5639 + }, + { + "epoch": 0.7174659712504771, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.139122009277344, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8819879293441772, + "num_tokens": 219036430.0, + "step": 5640 + }, + { + "epoch": 0.7175931815290676, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.16750431060791, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8867324590682983, + "num_tokens": 219077375.0, + "step": 5641 + }, + { + "epoch": 0.717720391807658, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.25253963470459, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.871582567691803, + "num_tokens": 219113089.0, + "step": 5642 + }, + { + "epoch": 0.7178476020862485, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.181351661682129, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8804616928100586, + "num_tokens": 219151027.0, + "step": 5643 + }, + { + "epoch": 0.7179748123648391, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.215068817138672, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.878980278968811, + "num_tokens": 219189445.0, + "step": 5644 + }, + { + "epoch": 0.7181020226434296, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.186985969543457, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8893362283706665, + "num_tokens": 219225998.0, + "step": 5645 + }, + { + "epoch": 0.7182292329220201, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.233692169189453, + "learning_rate": 1e-06, + "loss": 0.4021, + "mean_token_accuracy": 0.8688788414001465, + "num_tokens": 219267575.0, + "step": 5646 + }, + { + "epoch": 0.7183564432006107, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.193029403686523, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8831151723861694, + "num_tokens": 219308500.0, + "step": 5647 + }, + { + "epoch": 0.7184836534792011, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.242175102233887, + "learning_rate": 1e-06, + "loss": 0.3927, + "mean_token_accuracy": 0.8683015704154968, + "num_tokens": 219349337.0, + "step": 5648 + }, + { + "epoch": 0.7186108637577916, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.116472244262695, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8788437843322754, + "num_tokens": 219395363.0, + "step": 5649 + }, + { + "epoch": 0.7187380740363821, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.213804244995117, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8809895515441895, + "num_tokens": 219436336.0, + "step": 5650 + }, + { + "epoch": 0.7188652843149727, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.306074142456055, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8801232576370239, + "num_tokens": 219470917.0, + "step": 5651 + }, + { + "epoch": 0.7189924945935632, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.28054141998291, + "learning_rate": 1e-06, + "loss": 0.4131, + "mean_token_accuracy": 0.8607054352760315, + "num_tokens": 219509749.0, + "step": 5652 + }, + { + "epoch": 0.7191197048721537, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.130133628845215, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8868935108184814, + "num_tokens": 219554177.0, + "step": 5653 + }, + { + "epoch": 0.7192469151507441, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.20359992980957, + "learning_rate": 1e-06, + "loss": 0.3798, + "mean_token_accuracy": 0.8723046183586121, + "num_tokens": 219593369.0, + "step": 5654 + }, + { + "epoch": 0.7193741254293347, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.178421020507812, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8752206563949585, + "num_tokens": 219631526.0, + "step": 5655 + }, + { + "epoch": 0.7195013357079252, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.27001953125, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8899456858634949, + "num_tokens": 219665840.0, + "step": 5656 + }, + { + "epoch": 0.7196285459865157, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.574920654296875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.269960403442383, + "learning_rate": 1e-06, + "loss": 0.3897, + "mean_token_accuracy": 0.865809977054596, + "num_tokens": 219706762.0, + "step": 5657 + }, + { + "epoch": 0.7197557562651062, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.586841583251953e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.321572303771973, + "learning_rate": 1e-06, + "loss": 0.438, + "mean_token_accuracy": 0.8587074279785156, + "num_tokens": 219745799.0, + "step": 5658 + }, + { + "epoch": 0.7198829665436968, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.1958646774292, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.87568598985672, + "num_tokens": 219782347.0, + "step": 5659 + }, + { + "epoch": 0.7200101768222873, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.153352737426758, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8870761394500732, + "num_tokens": 219820881.0, + "step": 5660 + }, + { + "epoch": 0.7201373871008777, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.230971336364746, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8780093193054199, + "num_tokens": 219858645.0, + "step": 5661 + }, + { + "epoch": 0.7202645973794682, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.250870704650879, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8879322409629822, + "num_tokens": 219895194.0, + "step": 5662 + }, + { + "epoch": 0.7203918076580588, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.217020988464355, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8850432634353638, + "num_tokens": 219933104.0, + "step": 5663 + }, + { + "epoch": 0.7205190179366493, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.208450317382812, + "learning_rate": 1e-06, + "loss": 0.3738, + "mean_token_accuracy": 0.8753718137741089, + "num_tokens": 219976377.0, + "step": 5664 + }, + { + "epoch": 0.7206462282152398, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.128072738647461, + "learning_rate": 1e-06, + "loss": 0.3889, + "mean_token_accuracy": 0.869140625, + "num_tokens": 220022276.0, + "step": 5665 + }, + { + "epoch": 0.7207734384938304, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.318449020385742, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8708203434944153, + "num_tokens": 220053030.0, + "step": 5666 + }, + { + "epoch": 0.7209006487724208, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.29604721069336, + "learning_rate": 1e-06, + "loss": 0.4043, + "mean_token_accuracy": 0.8689597845077515, + "num_tokens": 220089399.0, + "step": 5667 + }, + { + "epoch": 0.7210278590510113, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.220212936401367, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8753842115402222, + "num_tokens": 220126020.0, + "step": 5668 + }, + { + "epoch": 0.7211550693296018, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.227883338928223, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8906928300857544, + "num_tokens": 220162762.0, + "step": 5669 + }, + { + "epoch": 0.7212822796081924, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.146196365356445, + "learning_rate": 1e-06, + "loss": 0.3894, + "mean_token_accuracy": 0.8697154521942139, + "num_tokens": 220205904.0, + "step": 5670 + }, + { + "epoch": 0.7214094898867829, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.18967342376709, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8783203363418579, + "num_tokens": 220245253.0, + "step": 5671 + }, + { + "epoch": 0.7215367001653734, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.185989379882812, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8764512538909912, + "num_tokens": 220286908.0, + "step": 5672 + }, + { + "epoch": 0.7216639104439638, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.169017791748047, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8883363604545593, + "num_tokens": 220326746.0, + "step": 5673 + }, + { + "epoch": 0.7217911207225544, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.339659690856934, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8739266395568848, + "num_tokens": 220359810.0, + "step": 5674 + }, + { + "epoch": 0.7219183310011449, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.27983283996582, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8742943406105042, + "num_tokens": 220395886.0, + "step": 5675 + }, + { + "epoch": 0.7220455412797354, + "ewc_loss": 0.011474609375, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.178576469421387, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8828519582748413, + "num_tokens": 220433060.0, + "step": 5676 + }, + { + "epoch": 0.7221727515583259, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.29661750793457, + "learning_rate": 1e-06, + "loss": 0.3861, + "mean_token_accuracy": 0.8686378002166748, + "num_tokens": 220471907.0, + "step": 5677 + }, + { + "epoch": 0.7222999618369165, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.227490425109863, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8828445672988892, + "num_tokens": 220503308.0, + "step": 5678 + }, + { + "epoch": 0.7224271721155069, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.32278823852539, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8771811127662659, + "num_tokens": 220536045.0, + "step": 5679 + }, + { + "epoch": 0.7225543823940974, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.5987625122070312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.239286422729492, + "learning_rate": 1e-06, + "loss": 0.3898, + "mean_token_accuracy": 0.8686422109603882, + "num_tokens": 220575977.0, + "step": 5680 + }, + { + "epoch": 0.722681592672688, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.215222358703613, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.881365954875946, + "num_tokens": 220615777.0, + "step": 5681 + }, + { + "epoch": 0.7228088029512785, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.232954978942871, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8697166442871094, + "num_tokens": 220655363.0, + "step": 5682 + }, + { + "epoch": 0.722936013229869, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.1904878616333, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8761365413665771, + "num_tokens": 220693125.0, + "step": 5683 + }, + { + "epoch": 0.7230632235084595, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.21723747253418, + "learning_rate": 1e-06, + "loss": 0.409, + "mean_token_accuracy": 0.8595620393753052, + "num_tokens": 220731117.0, + "step": 5684 + }, + { + "epoch": 0.72319043378705, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.259806632995605, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8810108304023743, + "num_tokens": 220766214.0, + "step": 5685 + }, + { + "epoch": 0.7233176440656405, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.150975227355957, + "learning_rate": 1e-06, + "loss": 0.3847, + "mean_token_accuracy": 0.8694870471954346, + "num_tokens": 220806688.0, + "step": 5686 + }, + { + "epoch": 0.723444854344231, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.183544158935547, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8871929049491882, + "num_tokens": 220849790.0, + "step": 5687 + }, + { + "epoch": 0.7235720646228215, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.211347579956055, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.886803388595581, + "num_tokens": 220889888.0, + "step": 5688 + }, + { + "epoch": 0.7236992749014121, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.186300277709961, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.8768548369407654, + "num_tokens": 220932597.0, + "step": 5689 + }, + { + "epoch": 0.7238264851800026, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.44814682006836, + "learning_rate": 1e-06, + "loss": 0.416, + "mean_token_accuracy": 0.8586469888687134, + "num_tokens": 220973593.0, + "step": 5690 + }, + { + "epoch": 0.723953695458593, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.16897964477539, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.893580436706543, + "num_tokens": 221011445.0, + "step": 5691 + }, + { + "epoch": 0.7240809057371835, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.14717960357666, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.879132866859436, + "num_tokens": 221055761.0, + "step": 5692 + }, + { + "epoch": 0.7242081160157741, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.301407814025879, + "learning_rate": 1e-06, + "loss": 0.416, + "mean_token_accuracy": 0.862365186214447, + "num_tokens": 221092761.0, + "step": 5693 + }, + { + "epoch": 0.7243353262943646, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.213257789611816, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8803749084472656, + "num_tokens": 221130558.0, + "step": 5694 + }, + { + "epoch": 0.7244625365729551, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.283306121826172, + "learning_rate": 1e-06, + "loss": 0.4019, + "mean_token_accuracy": 0.8630527257919312, + "num_tokens": 221166665.0, + "step": 5695 + }, + { + "epoch": 0.7245897468515456, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.155073165893555, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8842300176620483, + "num_tokens": 221205574.0, + "step": 5696 + }, + { + "epoch": 0.7247169571301361, + "ewc_loss": 0.01153564453125, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.237462043762207, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8758949041366577, + "num_tokens": 221241284.0, + "step": 5697 + }, + { + "epoch": 0.7248441674087266, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.273652076721191, + "learning_rate": 1e-06, + "loss": 0.3925, + "mean_token_accuracy": 0.8724796175956726, + "num_tokens": 221277562.0, + "step": 5698 + }, + { + "epoch": 0.7249713776873171, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.221588134765625, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8796417713165283, + "num_tokens": 221311453.0, + "step": 5699 + }, + { + "epoch": 0.7250985879659076, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.198708534240723, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8749883770942688, + "num_tokens": 221349285.0, + "step": 5700 + }, + { + "epoch": 0.7252257982444982, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.286396026611328, + "learning_rate": 1e-06, + "loss": 0.379, + "mean_token_accuracy": 0.8705774545669556, + "num_tokens": 221381003.0, + "step": 5701 + }, + { + "epoch": 0.7253530085230887, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.136279106140137, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8786754608154297, + "num_tokens": 221419679.0, + "step": 5702 + }, + { + "epoch": 0.7254802188016791, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.219758987426758, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8797377347946167, + "num_tokens": 221456688.0, + "step": 5703 + }, + { + "epoch": 0.7256074290802697, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.316344261169434, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.88313227891922, + "num_tokens": 221490400.0, + "step": 5704 + }, + { + "epoch": 0.7257346393588602, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.19227123260498, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8725732564926147, + "num_tokens": 221530141.0, + "step": 5705 + }, + { + "epoch": 0.7258618496374507, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.217318534851074, + "learning_rate": 1e-06, + "loss": 0.4036, + "mean_token_accuracy": 0.8688726425170898, + "num_tokens": 221573515.0, + "step": 5706 + }, + { + "epoch": 0.7259890599160412, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6226043701171875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.2555513381958, + "learning_rate": 1e-06, + "loss": 0.4142, + "mean_token_accuracy": 0.8608461618423462, + "num_tokens": 221614398.0, + "step": 5707 + }, + { + "epoch": 0.7261162701946318, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6226043701171875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.165139198303223, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8899389505386353, + "num_tokens": 221651824.0, + "step": 5708 + }, + { + "epoch": 0.7262434804732223, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6106834411621094e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.17887020111084, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8708938360214233, + "num_tokens": 221689053.0, + "step": 5709 + }, + { + "epoch": 0.7263706907518127, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.1854887008667, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.875272274017334, + "num_tokens": 221729921.0, + "step": 5710 + }, + { + "epoch": 0.7264979010304032, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.089034080505371, + "learning_rate": 1e-06, + "loss": 0.3716, + "mean_token_accuracy": 0.8734022378921509, + "num_tokens": 221769939.0, + "step": 5711 + }, + { + "epoch": 0.7266251113089938, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.167350769042969, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8846158981323242, + "num_tokens": 221812800.0, + "step": 5712 + }, + { + "epoch": 0.7267523215875843, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.224539756774902, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8817892074584961, + "num_tokens": 221846815.0, + "step": 5713 + }, + { + "epoch": 0.7268795318661748, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.152122497558594, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8821347951889038, + "num_tokens": 221888385.0, + "step": 5714 + }, + { + "epoch": 0.7270067421447653, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.261322975158691, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8751756548881531, + "num_tokens": 221925608.0, + "step": 5715 + }, + { + "epoch": 0.7271339524233558, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.207439422607422, + "learning_rate": 1e-06, + "loss": 0.3772, + "mean_token_accuracy": 0.874061107635498, + "num_tokens": 221965138.0, + "step": 5716 + }, + { + "epoch": 0.7272611627019463, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.273619651794434, + "learning_rate": 1e-06, + "loss": 0.4377, + "mean_token_accuracy": 0.8504903316497803, + "num_tokens": 222005213.0, + "step": 5717 + }, + { + "epoch": 0.7273883729805368, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.199860572814941, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8835493326187134, + "num_tokens": 222041290.0, + "step": 5718 + }, + { + "epoch": 0.7275155832591274, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.226126670837402, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8832381367683411, + "num_tokens": 222083900.0, + "step": 5719 + }, + { + "epoch": 0.7276427935377179, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.294088363647461, + "learning_rate": 1e-06, + "loss": 0.4336, + "mean_token_accuracy": 0.8552125692367554, + "num_tokens": 222122135.0, + "step": 5720 + }, + { + "epoch": 0.7277700038163084, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.20051383972168, + "learning_rate": 1e-06, + "loss": 0.3659, + "mean_token_accuracy": 0.8778939843177795, + "num_tokens": 222160544.0, + "step": 5721 + }, + { + "epoch": 0.7278972140948988, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.29960823059082, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8720638751983643, + "num_tokens": 222200648.0, + "step": 5722 + }, + { + "epoch": 0.7280244243734894, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.12610912322998, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8801198601722717, + "num_tokens": 222240296.0, + "step": 5723 + }, + { + "epoch": 0.7281516346520799, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6345252990722656e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.277209281921387, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8726520538330078, + "num_tokens": 222276384.0, + "step": 5724 + }, + { + "epoch": 0.7282788449306704, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.202608108520508, + "learning_rate": 1e-06, + "loss": 0.41, + "mean_token_accuracy": 0.8607285022735596, + "num_tokens": 222317876.0, + "step": 5725 + }, + { + "epoch": 0.7284060552092609, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.190998077392578, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.884389340877533, + "num_tokens": 222355894.0, + "step": 5726 + }, + { + "epoch": 0.7285332654878515, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.172619819641113, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8871049284934998, + "num_tokens": 222393831.0, + "step": 5727 + }, + { + "epoch": 0.7286604757664419, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.218734741210938, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8750776052474976, + "num_tokens": 222433260.0, + "step": 5728 + }, + { + "epoch": 0.7287876860450324, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.2421875, + "learning_rate": 1e-06, + "loss": 0.3745, + "mean_token_accuracy": 0.8726986050605774, + "num_tokens": 222470128.0, + "step": 5729 + }, + { + "epoch": 0.7289148963236229, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.272016525268555, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.882786750793457, + "num_tokens": 222499805.0, + "step": 5730 + }, + { + "epoch": 0.7290421066022135, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.247760772705078, + "learning_rate": 1e-06, + "loss": 0.3955, + "mean_token_accuracy": 0.8658429384231567, + "num_tokens": 222536564.0, + "step": 5731 + }, + { + "epoch": 0.729169316880804, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.16363525390625, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.887243926525116, + "num_tokens": 222575294.0, + "step": 5732 + }, + { + "epoch": 0.7292965271593945, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.172342300415039, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8840737342834473, + "num_tokens": 222614682.0, + "step": 5733 + }, + { + "epoch": 0.7294237374379849, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.368843078613281, + "learning_rate": 1e-06, + "loss": 0.3848, + "mean_token_accuracy": 0.8698709011077881, + "num_tokens": 222647552.0, + "step": 5734 + }, + { + "epoch": 0.7295509477165755, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.132142066955566, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8933987617492676, + "num_tokens": 222686345.0, + "step": 5735 + }, + { + "epoch": 0.729678157995166, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.22840404510498, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8774693608283997, + "num_tokens": 222729041.0, + "step": 5736 + }, + { + "epoch": 0.7298053682737565, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.190690040588379, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8742938041687012, + "num_tokens": 222773850.0, + "step": 5737 + }, + { + "epoch": 0.7299325785523471, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.212194442749023, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8799771666526794, + "num_tokens": 222806652.0, + "step": 5738 + }, + { + "epoch": 0.7300597888309376, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.227926254272461, + "learning_rate": 1e-06, + "loss": 0.3984, + "mean_token_accuracy": 0.8695616722106934, + "num_tokens": 222847536.0, + "step": 5739 + }, + { + "epoch": 0.730186999109528, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.165800094604492, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8751214742660522, + "num_tokens": 222892655.0, + "step": 5740 + }, + { + "epoch": 0.7303142093881185, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.218461990356445, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8780688643455505, + "num_tokens": 222935749.0, + "step": 5741 + }, + { + "epoch": 0.7304414196667091, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.239079475402832, + "learning_rate": 1e-06, + "loss": 0.3944, + "mean_token_accuracy": 0.8702563047409058, + "num_tokens": 222977936.0, + "step": 5742 + }, + { + "epoch": 0.7305686299452996, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.166998863220215, + "learning_rate": 1e-06, + "loss": 0.4087, + "mean_token_accuracy": 0.863190770149231, + "num_tokens": 223023444.0, + "step": 5743 + }, + { + "epoch": 0.7306958402238901, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.21449089050293, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8807660341262817, + "num_tokens": 223059806.0, + "step": 5744 + }, + { + "epoch": 0.7308230505024806, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.290491104125977, + "learning_rate": 1e-06, + "loss": 0.4015, + "mean_token_accuracy": 0.8648195266723633, + "num_tokens": 223098068.0, + "step": 5745 + }, + { + "epoch": 0.7309502607810711, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.322980880737305, + "learning_rate": 1e-06, + "loss": 0.4198, + "mean_token_accuracy": 0.8586089015007019, + "num_tokens": 223131151.0, + "step": 5746 + }, + { + "epoch": 0.7310774710596616, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.18752384185791, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8873029351234436, + "num_tokens": 223165775.0, + "step": 5747 + }, + { + "epoch": 0.7312046813382521, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.24924373626709, + "learning_rate": 1e-06, + "loss": 0.3939, + "mean_token_accuracy": 0.8686739206314087, + "num_tokens": 223201537.0, + "step": 5748 + }, + { + "epoch": 0.7313318916168426, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.23549747467041, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8782927989959717, + "num_tokens": 223238584.0, + "step": 5749 + }, + { + "epoch": 0.7314591018954332, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.205065727233887, + "learning_rate": 1e-06, + "loss": 0.3971, + "mean_token_accuracy": 0.8670510649681091, + "num_tokens": 223277419.0, + "step": 5750 + }, + { + "epoch": 0.7315863121740237, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.16703987121582, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8761590719223022, + "num_tokens": 223316000.0, + "step": 5751 + }, + { + "epoch": 0.7317135224526141, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.260542869567871, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8767826557159424, + "num_tokens": 223350427.0, + "step": 5752 + }, + { + "epoch": 0.7318407327312046, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.20471477508545, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8711841702461243, + "num_tokens": 223388757.0, + "step": 5753 + }, + { + "epoch": 0.7319679430097952, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.216442108154297, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8754127025604248, + "num_tokens": 223433416.0, + "step": 5754 + }, + { + "epoch": 0.7320951532883857, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.431473731994629, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8745684027671814, + "num_tokens": 223476605.0, + "step": 5755 + }, + { + "epoch": 0.7322223635669762, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.275943756103516, + "learning_rate": 1e-06, + "loss": 0.406, + "mean_token_accuracy": 0.8687448501586914, + "num_tokens": 223513878.0, + "step": 5756 + }, + { + "epoch": 0.7323495738455668, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.147336959838867, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.8786428570747375, + "num_tokens": 223554565.0, + "step": 5757 + }, + { + "epoch": 0.7324767841241572, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.219670295715332, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.8716740012168884, + "num_tokens": 223595091.0, + "step": 5758 + }, + { + "epoch": 0.7326039944027477, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.20980167388916, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8786557912826538, + "num_tokens": 223632787.0, + "step": 5759 + }, + { + "epoch": 0.7327312046813382, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.27711296081543, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.881749153137207, + "num_tokens": 223666457.0, + "step": 5760 + }, + { + "epoch": 0.7328584149599288, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.146900177001953, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8806449174880981, + "num_tokens": 223711000.0, + "step": 5761 + }, + { + "epoch": 0.7329856252385193, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.254804611206055, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8765718936920166, + "num_tokens": 223747912.0, + "step": 5762 + }, + { + "epoch": 0.7331128355171098, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.29560661315918, + "learning_rate": 1e-06, + "loss": 0.3864, + "mean_token_accuracy": 0.8676558136940002, + "num_tokens": 223787345.0, + "step": 5763 + }, + { + "epoch": 0.7332400457957003, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.658367156982422e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.19112491607666, + "learning_rate": 1e-06, + "loss": 0.389, + "mean_token_accuracy": 0.8677107691764832, + "num_tokens": 223829571.0, + "step": 5764 + }, + { + "epoch": 0.7333672560742908, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.160510063171387, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8799750208854675, + "num_tokens": 223869386.0, + "step": 5765 + }, + { + "epoch": 0.7334944663528813, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.170073509216309, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.879326343536377, + "num_tokens": 223911221.0, + "step": 5766 + }, + { + "epoch": 0.7336216766314718, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.176691055297852, + "learning_rate": 1e-06, + "loss": 0.3768, + "mean_token_accuracy": 0.8711695671081543, + "num_tokens": 223957250.0, + "step": 5767 + }, + { + "epoch": 0.7337488869100623, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.181564331054688, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8798382878303528, + "num_tokens": 223996614.0, + "step": 5768 + }, + { + "epoch": 0.7338760971886529, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.266961097717285, + "learning_rate": 1e-06, + "loss": 0.4014, + "mean_token_accuracy": 0.8664693236351013, + "num_tokens": 224032610.0, + "step": 5769 + }, + { + "epoch": 0.7340033074672434, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.2481107711792, + "learning_rate": 1e-06, + "loss": 0.402, + "mean_token_accuracy": 0.8655328154563904, + "num_tokens": 224070765.0, + "step": 5770 + }, + { + "epoch": 0.7341305177458338, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.240184783935547, + "learning_rate": 1e-06, + "loss": 0.3968, + "mean_token_accuracy": 0.8721804022789001, + "num_tokens": 224109107.0, + "step": 5771 + }, + { + "epoch": 0.7342577280244243, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.182761192321777, + "learning_rate": 1e-06, + "loss": 0.3886, + "mean_token_accuracy": 0.8732184171676636, + "num_tokens": 224149526.0, + "step": 5772 + }, + { + "epoch": 0.7343849383030149, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.152235984802246, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.8736868500709534, + "num_tokens": 224192711.0, + "step": 5773 + }, + { + "epoch": 0.7345121485816054, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.158743858337402, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8803467750549316, + "num_tokens": 224233985.0, + "step": 5774 + }, + { + "epoch": 0.7346393588601959, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.324926376342773, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8812006711959839, + "num_tokens": 224268044.0, + "step": 5775 + }, + { + "epoch": 0.7347665691387865, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.137252807617188, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8895242810249329, + "num_tokens": 224305104.0, + "step": 5776 + }, + { + "epoch": 0.7348937794173769, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.17193603515625, + "learning_rate": 1e-06, + "loss": 0.379, + "mean_token_accuracy": 0.8714600205421448, + "num_tokens": 224345532.0, + "step": 5777 + }, + { + "epoch": 0.7350209896959674, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.279306411743164, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.875267505645752, + "num_tokens": 224380673.0, + "step": 5778 + }, + { + "epoch": 0.7351481999745579, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.251243591308594, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8778023719787598, + "num_tokens": 224419779.0, + "step": 5779 + }, + { + "epoch": 0.7352754102531485, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.16151237487793, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.893549919128418, + "num_tokens": 224458007.0, + "step": 5780 + }, + { + "epoch": 0.735402620531739, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.19958782196045, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8865005373954773, + "num_tokens": 224497356.0, + "step": 5781 + }, + { + "epoch": 0.7355298308103295, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.240026473999023, + "learning_rate": 1e-06, + "loss": 0.4173, + "mean_token_accuracy": 0.8584448099136353, + "num_tokens": 224537830.0, + "step": 5782 + }, + { + "epoch": 0.7356570410889199, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.284777641296387, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8877642750740051, + "num_tokens": 224573689.0, + "step": 5783 + }, + { + "epoch": 0.7357842513675105, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.178175926208496, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8699768781661987, + "num_tokens": 224615938.0, + "step": 5784 + }, + { + "epoch": 0.735911461646101, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.280017852783203, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8772963881492615, + "num_tokens": 224649653.0, + "step": 5785 + }, + { + "epoch": 0.7360386719246915, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.249128341674805, + "learning_rate": 1e-06, + "loss": 0.4031, + "mean_token_accuracy": 0.8639365434646606, + "num_tokens": 224690502.0, + "step": 5786 + }, + { + "epoch": 0.736165882203282, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.262723922729492, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8718175292015076, + "num_tokens": 224730576.0, + "step": 5787 + }, + { + "epoch": 0.7362930924818726, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.264464378356934, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8820192813873291, + "num_tokens": 224762976.0, + "step": 5788 + }, + { + "epoch": 0.736420302760463, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.10695743560791, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8872790932655334, + "num_tokens": 224809717.0, + "step": 5789 + }, + { + "epoch": 0.7365475130390535, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.27946949005127, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8768205642700195, + "num_tokens": 224848670.0, + "step": 5790 + }, + { + "epoch": 0.736674723317644, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.34540843963623, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.8630450963973999, + "num_tokens": 224883810.0, + "step": 5791 + }, + { + "epoch": 0.7368019335962346, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.352729797363281, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8951859474182129, + "num_tokens": 224921290.0, + "step": 5792 + }, + { + "epoch": 0.7369291438748251, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.40662956237793, + "learning_rate": 1e-06, + "loss": 0.398, + "mean_token_accuracy": 0.8658130168914795, + "num_tokens": 224951446.0, + "step": 5793 + }, + { + "epoch": 0.7370563541534156, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.138228416442871, + "learning_rate": 1e-06, + "loss": 0.4192, + "mean_token_accuracy": 0.8597964644432068, + "num_tokens": 224996530.0, + "step": 5794 + }, + { + "epoch": 0.737183564432006, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.217567443847656, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8778153657913208, + "num_tokens": 225037880.0, + "step": 5795 + }, + { + "epoch": 0.7373107747105966, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.346280097961426, + "learning_rate": 1e-06, + "loss": 0.3892, + "mean_token_accuracy": 0.8690444231033325, + "num_tokens": 225073479.0, + "step": 5796 + }, + { + "epoch": 0.7374379849891871, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.176092147827148, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8801239728927612, + "num_tokens": 225113009.0, + "step": 5797 + }, + { + "epoch": 0.7375651952677776, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.218501091003418, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8736000657081604, + "num_tokens": 225157409.0, + "step": 5798 + }, + { + "epoch": 0.7376924055463682, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.194717407226562, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8757802248001099, + "num_tokens": 225196077.0, + "step": 5799 + }, + { + "epoch": 0.7378196158249587, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.175201416015625, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8855377435684204, + "num_tokens": 225237441.0, + "step": 5800 + }, + { + "epoch": 0.7379468261035491, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.186968803405762, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8849082589149475, + "num_tokens": 225278188.0, + "step": 5801 + }, + { + "epoch": 0.7380740363821396, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.198051452636719, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.887673020362854, + "num_tokens": 225317716.0, + "step": 5802 + }, + { + "epoch": 0.7382012466607302, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.156843185424805, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8805471062660217, + "num_tokens": 225360599.0, + "step": 5803 + }, + { + "epoch": 0.7383284569393207, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.213651657104492, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8771564364433289, + "num_tokens": 225403375.0, + "step": 5804 + }, + { + "epoch": 0.7384556672179112, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.255331039428711, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8844866752624512, + "num_tokens": 225443905.0, + "step": 5805 + }, + { + "epoch": 0.7385828774965018, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.2708101272583, + "learning_rate": 1e-06, + "loss": 0.4095, + "mean_token_accuracy": 0.8620983362197876, + "num_tokens": 225480472.0, + "step": 5806 + }, + { + "epoch": 0.7387100877750922, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.121703147888184, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8913259506225586, + "num_tokens": 225522099.0, + "step": 5807 + }, + { + "epoch": 0.7388372980536827, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.284110069274902, + "learning_rate": 1e-06, + "loss": 0.3872, + "mean_token_accuracy": 0.870874285697937, + "num_tokens": 225559617.0, + "step": 5808 + }, + { + "epoch": 0.7389645083322732, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.196614265441895, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.888675332069397, + "num_tokens": 225600906.0, + "step": 5809 + }, + { + "epoch": 0.7390917186108638, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.202693939208984, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8753598928451538, + "num_tokens": 225642227.0, + "step": 5810 + }, + { + "epoch": 0.7392189288894543, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.277040481567383, + "learning_rate": 1e-06, + "loss": 0.4449, + "mean_token_accuracy": 0.8539820909500122, + "num_tokens": 225684777.0, + "step": 5811 + }, + { + "epoch": 0.7393461391680448, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.197920799255371, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8763185143470764, + "num_tokens": 225724488.0, + "step": 5812 + }, + { + "epoch": 0.7394733494466353, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.332409858703613, + "learning_rate": 1e-06, + "loss": 0.3793, + "mean_token_accuracy": 0.8734092712402344, + "num_tokens": 225758235.0, + "step": 5813 + }, + { + "epoch": 0.7396005597252258, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.188488006591797, + "learning_rate": 1e-06, + "loss": 0.3865, + "mean_token_accuracy": 0.8707587122917175, + "num_tokens": 225804419.0, + "step": 5814 + }, + { + "epoch": 0.7397277700038163, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.191462516784668, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8888511657714844, + "num_tokens": 225840198.0, + "step": 5815 + }, + { + "epoch": 0.7398549802824068, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.252456665039062, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8759667277336121, + "num_tokens": 225878259.0, + "step": 5816 + }, + { + "epoch": 0.7399821905609973, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.197027206420898, + "learning_rate": 1e-06, + "loss": 0.3681, + "mean_token_accuracy": 0.8729056119918823, + "num_tokens": 225920252.0, + "step": 5817 + }, + { + "epoch": 0.7401094008395879, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.187012672424316, + "learning_rate": 1e-06, + "loss": 0.4147, + "mean_token_accuracy": 0.8589264154434204, + "num_tokens": 225962237.0, + "step": 5818 + }, + { + "epoch": 0.7402366111181784, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.25558090209961, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8735289573669434, + "num_tokens": 225999020.0, + "step": 5819 + }, + { + "epoch": 0.7403638213967688, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.157922744750977, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8889780044555664, + "num_tokens": 226037920.0, + "step": 5820 + }, + { + "epoch": 0.7404910316753593, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.233686447143555, + "learning_rate": 1e-06, + "loss": 0.3878, + "mean_token_accuracy": 0.8683629035949707, + "num_tokens": 226079844.0, + "step": 5821 + }, + { + "epoch": 0.7406182419539499, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.284250259399414, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8754885792732239, + "num_tokens": 226113223.0, + "step": 5822 + }, + { + "epoch": 0.7407454522325404, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.249314308166504, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8766868710517883, + "num_tokens": 226147431.0, + "step": 5823 + }, + { + "epoch": 0.7408726625111309, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.20079231262207, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8897301554679871, + "num_tokens": 226184542.0, + "step": 5824 + }, + { + "epoch": 0.7409998727897215, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.258035659790039, + "learning_rate": 1e-06, + "loss": 0.4317, + "mean_token_accuracy": 0.8592326641082764, + "num_tokens": 226226706.0, + "step": 5825 + }, + { + "epoch": 0.7411270830683119, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.682209014892578e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.301664352416992, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8748906850814819, + "num_tokens": 226259968.0, + "step": 5826 + }, + { + "epoch": 0.7412542933469024, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.288265228271484, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.8708348274230957, + "num_tokens": 226298952.0, + "step": 5827 + }, + { + "epoch": 0.7413815036254929, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.174952507019043, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.877971887588501, + "num_tokens": 226340137.0, + "step": 5828 + }, + { + "epoch": 0.7415087139040835, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.16824722290039, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8816225528717041, + "num_tokens": 226382357.0, + "step": 5829 + }, + { + "epoch": 0.741635924182674, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.232446670532227, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8779712915420532, + "num_tokens": 226415678.0, + "step": 5830 + }, + { + "epoch": 0.7417631344612645, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.191849708557129, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8918443322181702, + "num_tokens": 226453839.0, + "step": 5831 + }, + { + "epoch": 0.7418903447398549, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.263715744018555, + "learning_rate": 1e-06, + "loss": 0.3999, + "mean_token_accuracy": 0.8663843870162964, + "num_tokens": 226494043.0, + "step": 5832 + }, + { + "epoch": 0.7420175550184455, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.308917045593262, + "learning_rate": 1e-06, + "loss": 0.3754, + "mean_token_accuracy": 0.8696197271347046, + "num_tokens": 226530857.0, + "step": 5833 + }, + { + "epoch": 0.742144765297036, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.15195369720459, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8817498683929443, + "num_tokens": 226575534.0, + "step": 5834 + }, + { + "epoch": 0.7422719755756265, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.22472095489502, + "learning_rate": 1e-06, + "loss": 0.4294, + "mean_token_accuracy": 0.8552455306053162, + "num_tokens": 226621038.0, + "step": 5835 + }, + { + "epoch": 0.742399185854217, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.432289123535156, + "learning_rate": 1e-06, + "loss": 0.4502, + "mean_token_accuracy": 0.8489906191825867, + "num_tokens": 226654429.0, + "step": 5836 + }, + { + "epoch": 0.7425263961328076, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.15942668914795, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.887348473072052, + "num_tokens": 226693268.0, + "step": 5837 + }, + { + "epoch": 0.742653606411398, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.260193824768066, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8816138505935669, + "num_tokens": 226730005.0, + "step": 5838 + }, + { + "epoch": 0.7427808166899885, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.217659950256348, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8727940320968628, + "num_tokens": 226769107.0, + "step": 5839 + }, + { + "epoch": 0.742908026968579, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.203777313232422, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8771476745605469, + "num_tokens": 226811665.0, + "step": 5840 + }, + { + "epoch": 0.7430352372471696, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.223491668701172, + "learning_rate": 1e-06, + "loss": 0.4013, + "mean_token_accuracy": 0.8652845621109009, + "num_tokens": 226852707.0, + "step": 5841 + }, + { + "epoch": 0.7431624475257601, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.229589462280273, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8937797546386719, + "num_tokens": 226889874.0, + "step": 5842 + }, + { + "epoch": 0.7432896578043506, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.2191743850708, + "learning_rate": 1e-06, + "loss": 0.3868, + "mean_token_accuracy": 0.8684800863265991, + "num_tokens": 226931890.0, + "step": 5843 + }, + { + "epoch": 0.743416868082941, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.252015113830566, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8815792202949524, + "num_tokens": 226969250.0, + "step": 5844 + }, + { + "epoch": 0.7435440783615316, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.185035705566406, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8812252283096313, + "num_tokens": 227013197.0, + "step": 5845 + }, + { + "epoch": 0.7436712886401221, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.16881275177002, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8787181973457336, + "num_tokens": 227054936.0, + "step": 5846 + }, + { + "epoch": 0.7437984989187126, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.177934646606445, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.884821891784668, + "num_tokens": 227100903.0, + "step": 5847 + }, + { + "epoch": 0.7439257091973032, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.361227989196777, + "learning_rate": 1e-06, + "loss": 0.4076, + "mean_token_accuracy": 0.865469217300415, + "num_tokens": 227134586.0, + "step": 5848 + }, + { + "epoch": 0.7440529194758937, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.27998161315918, + "learning_rate": 1e-06, + "loss": 0.4013, + "mean_token_accuracy": 0.8692975044250488, + "num_tokens": 227170456.0, + "step": 5849 + }, + { + "epoch": 0.7441801297544841, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.6941299438476562e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.127355575561523, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8873668909072876, + "num_tokens": 227207689.0, + "step": 5850 + }, + { + "epoch": 0.7443073400330746, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.15188217163086, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8810590505599976, + "num_tokens": 227257110.0, + "step": 5851 + }, + { + "epoch": 0.7444345503116652, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.215786933898926, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8800761699676514, + "num_tokens": 227292231.0, + "step": 5852 + }, + { + "epoch": 0.7445617605902557, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.275382995605469, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8855333924293518, + "num_tokens": 227329682.0, + "step": 5853 + }, + { + "epoch": 0.7446889708688462, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.281537055969238, + "learning_rate": 1e-06, + "loss": 0.3982, + "mean_token_accuracy": 0.8656542301177979, + "num_tokens": 227366699.0, + "step": 5854 + }, + { + "epoch": 0.7448161811474368, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.252470016479492, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8740592002868652, + "num_tokens": 227407393.0, + "step": 5855 + }, + { + "epoch": 0.7449433914260272, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.161027908325195, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8725469708442688, + "num_tokens": 227446588.0, + "step": 5856 + }, + { + "epoch": 0.7450706017046177, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.176182746887207, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8842548727989197, + "num_tokens": 227485372.0, + "step": 5857 + }, + { + "epoch": 0.7451978119832082, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.22170639038086, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8840174078941345, + "num_tokens": 227522483.0, + "step": 5858 + }, + { + "epoch": 0.7453250222617988, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.112154006958008, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8807481527328491, + "num_tokens": 227569077.0, + "step": 5859 + }, + { + "epoch": 0.7454522325403893, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.24299430847168, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8762676119804382, + "num_tokens": 227609462.0, + "step": 5860 + }, + { + "epoch": 0.7455794428189798, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.205622673034668, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8760679960250854, + "num_tokens": 227648095.0, + "step": 5861 + }, + { + "epoch": 0.7457066530975703, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.220274925231934, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8890525102615356, + "num_tokens": 227684215.0, + "step": 5862 + }, + { + "epoch": 0.7458338633761608, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.217905044555664, + "learning_rate": 1e-06, + "loss": 0.4043, + "mean_token_accuracy": 0.8653229475021362, + "num_tokens": 227726545.0, + "step": 5863 + }, + { + "epoch": 0.7459610736547513, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.181714057922363, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8753526210784912, + "num_tokens": 227770651.0, + "step": 5864 + }, + { + "epoch": 0.7460882839333418, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.301849365234375, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8739964962005615, + "num_tokens": 227806601.0, + "step": 5865 + }, + { + "epoch": 0.7462154942119323, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.157261848449707, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8928130865097046, + "num_tokens": 227846866.0, + "step": 5866 + }, + { + "epoch": 0.7463427044905229, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.7060508728027344e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.202147483825684, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8801458477973938, + "num_tokens": 227883156.0, + "step": 5867 + }, + { + "epoch": 0.7464699147691134, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.289324760437012, + "learning_rate": 1e-06, + "loss": 0.4163, + "mean_token_accuracy": 0.8599772453308105, + "num_tokens": 227925052.0, + "step": 5868 + }, + { + "epoch": 0.7465971250477038, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.242751121520996, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8751788139343262, + "num_tokens": 227960345.0, + "step": 5869 + }, + { + "epoch": 0.7467243353262943, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.174703598022461, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8789715766906738, + "num_tokens": 228002397.0, + "step": 5870 + }, + { + "epoch": 0.7468515456048849, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.235644340515137, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8796893358230591, + "num_tokens": 228037826.0, + "step": 5871 + }, + { + "epoch": 0.7469787558834754, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.285574913024902, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8808304667472839, + "num_tokens": 228076738.0, + "step": 5872 + }, + { + "epoch": 0.7471059661620659, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.229094505310059, + "learning_rate": 1e-06, + "loss": 0.3784, + "mean_token_accuracy": 0.8681693077087402, + "num_tokens": 228116402.0, + "step": 5873 + }, + { + "epoch": 0.7472331764406565, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.221195220947266, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8836259841918945, + "num_tokens": 228155791.0, + "step": 5874 + }, + { + "epoch": 0.7473603867192469, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.221081733703613, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.8733822107315063, + "num_tokens": 228197430.0, + "step": 5875 + }, + { + "epoch": 0.7474875969978374, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.253366470336914, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8849493861198425, + "num_tokens": 228231254.0, + "step": 5876 + }, + { + "epoch": 0.7476148072764279, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.240762710571289, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8877497911453247, + "num_tokens": 228265552.0, + "step": 5877 + }, + { + "epoch": 0.7477420175550185, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.180317878723145, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8783444762229919, + "num_tokens": 228307320.0, + "step": 5878 + }, + { + "epoch": 0.747869227833609, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.24681282043457, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.870895266532898, + "num_tokens": 228350040.0, + "step": 5879 + }, + { + "epoch": 0.7479964381121995, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.230981826782227, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8865314722061157, + "num_tokens": 228392008.0, + "step": 5880 + }, + { + "epoch": 0.7481236483907899, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.14637279510498, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8931142091751099, + "num_tokens": 228432045.0, + "step": 5881 + }, + { + "epoch": 0.7482508586693805, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.213915824890137, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8805200457572937, + "num_tokens": 228471648.0, + "step": 5882 + }, + { + "epoch": 0.748378068947971, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.200684547424316, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8755350708961487, + "num_tokens": 228516169.0, + "step": 5883 + }, + { + "epoch": 0.7485052792265615, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.247916221618652, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.8693673610687256, + "num_tokens": 228558602.0, + "step": 5884 + }, + { + "epoch": 0.748632489505152, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.173932075500488, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8836731314659119, + "num_tokens": 228602328.0, + "step": 5885 + }, + { + "epoch": 0.7487596997837426, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.25353717803955, + "learning_rate": 1e-06, + "loss": 0.413, + "mean_token_accuracy": 0.8627021312713623, + "num_tokens": 228640380.0, + "step": 5886 + }, + { + "epoch": 0.748886910062333, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.159662246704102, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.8768404722213745, + "num_tokens": 228682416.0, + "step": 5887 + }, + { + "epoch": 0.7490141203409235, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.273582458496094, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8721076250076294, + "num_tokens": 228717638.0, + "step": 5888 + }, + { + "epoch": 0.749141330619514, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.210875511169434, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8839876651763916, + "num_tokens": 228753745.0, + "step": 5889 + }, + { + "epoch": 0.7492685408981046, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.21504020690918, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.8756660223007202, + "num_tokens": 228791396.0, + "step": 5890 + }, + { + "epoch": 0.7493957511766951, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.174049377441406, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8786457777023315, + "num_tokens": 228830217.0, + "step": 5891 + }, + { + "epoch": 0.7495229614552856, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.233560562133789, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8715417385101318, + "num_tokens": 228873280.0, + "step": 5892 + }, + { + "epoch": 0.749650171733876, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.306474685668945, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8878001570701599, + "num_tokens": 228906256.0, + "step": 5893 + }, + { + "epoch": 0.7497773820124666, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.24011516571045, + "learning_rate": 1e-06, + "loss": 0.3912, + "mean_token_accuracy": 0.867621660232544, + "num_tokens": 228944611.0, + "step": 5894 + }, + { + "epoch": 0.7499045922910571, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.170138359069824, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8941144943237305, + "num_tokens": 228987904.0, + "step": 5895 + }, + { + "epoch": 0.7500318025696476, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.248123168945312, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8822864294052124, + "num_tokens": 229023677.0, + "step": 5896 + }, + { + "epoch": 0.7501590128482382, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.264608383178711, + "learning_rate": 1e-06, + "loss": 0.414, + "mean_token_accuracy": 0.8620741367340088, + "num_tokens": 229060236.0, + "step": 5897 + }, + { + "epoch": 0.7502862231268287, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.211484909057617, + "learning_rate": 1e-06, + "loss": 0.4014, + "mean_token_accuracy": 0.8663372993469238, + "num_tokens": 229100873.0, + "step": 5898 + }, + { + "epoch": 0.7504134334054191, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.203144073486328, + "learning_rate": 1e-06, + "loss": 0.3912, + "mean_token_accuracy": 0.8661431670188904, + "num_tokens": 229144127.0, + "step": 5899 + }, + { + "epoch": 0.7505406436840096, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.27310848236084, + "learning_rate": 1e-06, + "loss": 0.4413, + "mean_token_accuracy": 0.8559466600418091, + "num_tokens": 229182477.0, + "step": 5900 + }, + { + "epoch": 0.7506678539626002, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.415369033813477, + "learning_rate": 1e-06, + "loss": 0.4063, + "mean_token_accuracy": 0.8640119433403015, + "num_tokens": 229214472.0, + "step": 5901 + }, + { + "epoch": 0.7507950642411907, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.323741912841797, + "learning_rate": 1e-06, + "loss": 0.4212, + "mean_token_accuracy": 0.8591363430023193, + "num_tokens": 229249818.0, + "step": 5902 + }, + { + "epoch": 0.7509222745197812, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.187058448791504, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8789881467819214, + "num_tokens": 229293351.0, + "step": 5903 + }, + { + "epoch": 0.7510494847983717, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.204972267150879, + "learning_rate": 1e-06, + "loss": 0.4123, + "mean_token_accuracy": 0.8660637736320496, + "num_tokens": 229335377.0, + "step": 5904 + }, + { + "epoch": 0.7511766950769622, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.133509635925293, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8969374895095825, + "num_tokens": 229382943.0, + "step": 5905 + }, + { + "epoch": 0.7513039053555527, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.191431045532227, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.8689390420913696, + "num_tokens": 229425473.0, + "step": 5906 + }, + { + "epoch": 0.7514311156341432, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.489928245544434, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8825130462646484, + "num_tokens": 229454644.0, + "step": 5907 + }, + { + "epoch": 0.7515583259127337, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.236452102661133, + "learning_rate": 1e-06, + "loss": 0.4046, + "mean_token_accuracy": 0.8648731708526611, + "num_tokens": 229490789.0, + "step": 5908 + }, + { + "epoch": 0.7516855361913243, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.22830867767334, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8772387504577637, + "num_tokens": 229526696.0, + "step": 5909 + }, + { + "epoch": 0.7518127464699148, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.220022201538086, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8789138793945312, + "num_tokens": 229567722.0, + "step": 5910 + }, + { + "epoch": 0.7519399567485053, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7179718017578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.267206192016602, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8772076368331909, + "num_tokens": 229605155.0, + "step": 5911 + }, + { + "epoch": 0.7520671670270958, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.265028953552246, + "learning_rate": 1e-06, + "loss": 0.3847, + "mean_token_accuracy": 0.8696635961532593, + "num_tokens": 229643270.0, + "step": 5912 + }, + { + "epoch": 0.7521943773056863, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.305781364440918, + "learning_rate": 1e-06, + "loss": 0.3856, + "mean_token_accuracy": 0.8706591725349426, + "num_tokens": 229679689.0, + "step": 5913 + }, + { + "epoch": 0.7523215875842768, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.300146102905273, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8770039081573486, + "num_tokens": 229714531.0, + "step": 5914 + }, + { + "epoch": 0.7524487978628673, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.234917640686035, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8802181482315063, + "num_tokens": 229749726.0, + "step": 5915 + }, + { + "epoch": 0.7525760081414579, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.39687442779541, + "learning_rate": 1e-06, + "loss": 0.4074, + "mean_token_accuracy": 0.8610180020332336, + "num_tokens": 229782690.0, + "step": 5916 + }, + { + "epoch": 0.7527032184200484, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.218645095825195, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8754388093948364, + "num_tokens": 229818017.0, + "step": 5917 + }, + { + "epoch": 0.7528304286986388, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.173547744750977, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8884826302528381, + "num_tokens": 229860557.0, + "step": 5918 + }, + { + "epoch": 0.7529576389772293, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.228620529174805, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8692845106124878, + "num_tokens": 229900528.0, + "step": 5919 + }, + { + "epoch": 0.7530848492558199, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.256553649902344, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.884461522102356, + "num_tokens": 229934532.0, + "step": 5920 + }, + { + "epoch": 0.7532120595344104, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.147012710571289, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.880498468875885, + "num_tokens": 229980045.0, + "step": 5921 + }, + { + "epoch": 0.7533392698130009, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.258987426757812, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8777183294296265, + "num_tokens": 230015922.0, + "step": 5922 + }, + { + "epoch": 0.7534664800915915, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.1869535446167, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8854247331619263, + "num_tokens": 230054494.0, + "step": 5923 + }, + { + "epoch": 0.7535936903701819, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.23361873626709, + "learning_rate": 1e-06, + "loss": 0.4143, + "mean_token_accuracy": 0.8630881905555725, + "num_tokens": 230097490.0, + "step": 5924 + }, + { + "epoch": 0.7537209006487724, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 9.413269996643066, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8756217956542969, + "num_tokens": 230129407.0, + "step": 5925 + }, + { + "epoch": 0.7538481109273629, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.376962661743164, + "learning_rate": 1e-06, + "loss": 0.3986, + "mean_token_accuracy": 0.869807243347168, + "num_tokens": 230163076.0, + "step": 5926 + }, + { + "epoch": 0.7539753212059535, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.049581527709961, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8785708546638489, + "num_tokens": 230200861.0, + "step": 5927 + }, + { + "epoch": 0.754102531484544, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.243160247802734, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.888877809047699, + "num_tokens": 230239017.0, + "step": 5928 + }, + { + "epoch": 0.7542297417631345, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.121230125427246, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.897681474685669, + "num_tokens": 230276877.0, + "step": 5929 + }, + { + "epoch": 0.7543569520417249, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.191910743713379, + "learning_rate": 1e-06, + "loss": 0.4001, + "mean_token_accuracy": 0.8641445636749268, + "num_tokens": 230318244.0, + "step": 5930 + }, + { + "epoch": 0.7544841623203155, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.207576751708984, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.8705384731292725, + "num_tokens": 230357677.0, + "step": 5931 + }, + { + "epoch": 0.754611372598906, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.253662109375, + "learning_rate": 1e-06, + "loss": 0.3752, + "mean_token_accuracy": 0.8743501901626587, + "num_tokens": 230394487.0, + "step": 5932 + }, + { + "epoch": 0.7547385828774965, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.252481460571289, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8865511417388916, + "num_tokens": 230434425.0, + "step": 5933 + }, + { + "epoch": 0.754865793156087, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.271145820617676, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8811678886413574, + "num_tokens": 230468127.0, + "step": 5934 + }, + { + "epoch": 0.7549930034346776, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.336151123046875, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8776989579200745, + "num_tokens": 230498264.0, + "step": 5935 + }, + { + "epoch": 0.755120213713268, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.312013626098633, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8786908388137817, + "num_tokens": 230532590.0, + "step": 5936 + }, + { + "epoch": 0.7552474239918585, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.326212882995605, + "learning_rate": 1e-06, + "loss": 0.3918, + "mean_token_accuracy": 0.8679320812225342, + "num_tokens": 230566929.0, + "step": 5937 + }, + { + "epoch": 0.755374634270449, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.313413619995117, + "learning_rate": 1e-06, + "loss": 0.3846, + "mean_token_accuracy": 0.871414303779602, + "num_tokens": 230604061.0, + "step": 5938 + }, + { + "epoch": 0.7555018445490396, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.193822860717773, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.877437174320221, + "num_tokens": 230642585.0, + "step": 5939 + }, + { + "epoch": 0.7556290548276301, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.208033561706543, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8697694540023804, + "num_tokens": 230680416.0, + "step": 5940 + }, + { + "epoch": 0.7557562651062206, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.23575496673584, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.882309079170227, + "num_tokens": 230718493.0, + "step": 5941 + }, + { + "epoch": 0.755883475384811, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7418136596679688e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.14102840423584, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.886417031288147, + "num_tokens": 230759168.0, + "step": 5942 + }, + { + "epoch": 0.7560106856634016, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.333456039428711, + "learning_rate": 1e-06, + "loss": 0.3817, + "mean_token_accuracy": 0.8693063259124756, + "num_tokens": 230797184.0, + "step": 5943 + }, + { + "epoch": 0.7561378959419921, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.174117088317871, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8903141021728516, + "num_tokens": 230835350.0, + "step": 5944 + }, + { + "epoch": 0.7562651062205826, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.238615036010742, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8891407251358032, + "num_tokens": 230877565.0, + "step": 5945 + }, + { + "epoch": 0.7563923164991732, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.349143981933594, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8816827535629272, + "num_tokens": 230909360.0, + "step": 5946 + }, + { + "epoch": 0.7565195267777637, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.181516647338867, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8798384666442871, + "num_tokens": 230949833.0, + "step": 5947 + }, + { + "epoch": 0.7566467370563541, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.320223808288574, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8830541372299194, + "num_tokens": 230984267.0, + "step": 5948 + }, + { + "epoch": 0.7567739473349446, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.286275863647461, + "learning_rate": 1e-06, + "loss": 0.3961, + "mean_token_accuracy": 0.8674004673957825, + "num_tokens": 231021936.0, + "step": 5949 + }, + { + "epoch": 0.7569011576135352, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.32992935180664, + "learning_rate": 1e-06, + "loss": 0.3979, + "mean_token_accuracy": 0.8687001466751099, + "num_tokens": 231058259.0, + "step": 5950 + }, + { + "epoch": 0.7570283678921257, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.13919734954834, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8962438702583313, + "num_tokens": 231096592.0, + "step": 5951 + }, + { + "epoch": 0.7571555781707162, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.251582145690918, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8785035610198975, + "num_tokens": 231139236.0, + "step": 5952 + }, + { + "epoch": 0.7572827884493067, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.403084754943848, + "learning_rate": 1e-06, + "loss": 0.3875, + "mean_token_accuracy": 0.8659718036651611, + "num_tokens": 231171583.0, + "step": 5953 + }, + { + "epoch": 0.7574099987278972, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.26065731048584, + "learning_rate": 1e-06, + "loss": 0.3816, + "mean_token_accuracy": 0.8707331418991089, + "num_tokens": 231208980.0, + "step": 5954 + }, + { + "epoch": 0.7575372090064877, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.245819091796875, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8914450407028198, + "num_tokens": 231246245.0, + "step": 5955 + }, + { + "epoch": 0.7576644192850782, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.189704895019531, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8868774175643921, + "num_tokens": 231282809.0, + "step": 5956 + }, + { + "epoch": 0.7577916295636687, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.232881546020508, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8831337094306946, + "num_tokens": 231323363.0, + "step": 5957 + }, + { + "epoch": 0.7579188398422593, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.174103736877441, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8955385684967041, + "num_tokens": 231361392.0, + "step": 5958 + }, + { + "epoch": 0.7580460501208498, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.252120018005371, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.8739175796508789, + "num_tokens": 231405048.0, + "step": 5959 + }, + { + "epoch": 0.7581732603994403, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.274531364440918, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8769323825836182, + "num_tokens": 231444838.0, + "step": 5960 + }, + { + "epoch": 0.7583004706780307, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.222065925598145, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.8687548637390137, + "num_tokens": 231484542.0, + "step": 5961 + }, + { + "epoch": 0.7584276809566213, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.248024940490723, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.88145911693573, + "num_tokens": 231521499.0, + "step": 5962 + }, + { + "epoch": 0.7585548912352118, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.18721866607666, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8834885358810425, + "num_tokens": 231563434.0, + "step": 5963 + }, + { + "epoch": 0.7586821015138023, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.274943351745605, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8812007904052734, + "num_tokens": 231600122.0, + "step": 5964 + }, + { + "epoch": 0.7588093117923929, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.249208450317383, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8847540616989136, + "num_tokens": 231635495.0, + "step": 5965 + }, + { + "epoch": 0.7589365220709834, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.274185180664062, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.876876711845398, + "num_tokens": 231675168.0, + "step": 5966 + }, + { + "epoch": 0.7590637323495738, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.223320007324219, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8719006776809692, + "num_tokens": 231718531.0, + "step": 5967 + }, + { + "epoch": 0.7591909426281643, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.202211380004883, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8817811012268066, + "num_tokens": 231762207.0, + "step": 5968 + }, + { + "epoch": 0.7593181529067549, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.264058113098145, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8855961561203003, + "num_tokens": 231798311.0, + "step": 5969 + }, + { + "epoch": 0.7594453631853454, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.197867393493652, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8772308826446533, + "num_tokens": 231845475.0, + "step": 5970 + }, + { + "epoch": 0.7595725734639359, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.3135404586792, + "learning_rate": 1e-06, + "loss": 0.4232, + "mean_token_accuracy": 0.8563989996910095, + "num_tokens": 231885113.0, + "step": 5971 + }, + { + "epoch": 0.7596997837425264, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.28293514251709, + "learning_rate": 1e-06, + "loss": 0.3844, + "mean_token_accuracy": 0.8755886554718018, + "num_tokens": 231927429.0, + "step": 5972 + }, + { + "epoch": 0.7598269940211169, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.209650993347168, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8764674663543701, + "num_tokens": 231969149.0, + "step": 5973 + }, + { + "epoch": 0.7599542042997074, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.234389305114746, + "learning_rate": 1e-06, + "loss": 0.4071, + "mean_token_accuracy": 0.8631616234779358, + "num_tokens": 232012236.0, + "step": 5974 + }, + { + "epoch": 0.7600814145782979, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.259018898010254, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8719119429588318, + "num_tokens": 232053317.0, + "step": 5975 + }, + { + "epoch": 0.7602086248568884, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.244877815246582, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8894715905189514, + "num_tokens": 232091563.0, + "step": 5976 + }, + { + "epoch": 0.760335835135479, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.259477615356445, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.8769554495811462, + "num_tokens": 232129857.0, + "step": 5977 + }, + { + "epoch": 0.7604630454140695, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.169110298156738, + "learning_rate": 1e-06, + "loss": 0.3779, + "mean_token_accuracy": 0.875265896320343, + "num_tokens": 232176832.0, + "step": 5978 + }, + { + "epoch": 0.7605902556926599, + "ewc_loss": 0.01165771484375, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.26296329498291, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.876670241355896, + "num_tokens": 232215456.0, + "step": 5979 + }, + { + "epoch": 0.7607174659712505, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.753734588623047e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.204882621765137, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8946853280067444, + "num_tokens": 232258440.0, + "step": 5980 + }, + { + "epoch": 0.760844676249841, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.3045654296875, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8753326535224915, + "num_tokens": 232297447.0, + "step": 5981 + }, + { + "epoch": 0.7609718865284315, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.208930015563965, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8834189176559448, + "num_tokens": 232338159.0, + "step": 5982 + }, + { + "epoch": 0.761099096807022, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.2296724319458, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8863527774810791, + "num_tokens": 232380308.0, + "step": 5983 + }, + { + "epoch": 0.7612263070856126, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.23615550994873, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8766196966171265, + "num_tokens": 232420148.0, + "step": 5984 + }, + { + "epoch": 0.761353517364203, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.212939262390137, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8869569301605225, + "num_tokens": 232461826.0, + "step": 5985 + }, + { + "epoch": 0.7614807276427935, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.222825050354004, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.8734684586524963, + "num_tokens": 232506666.0, + "step": 5986 + }, + { + "epoch": 0.761607937921384, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.306507110595703, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.895094633102417, + "num_tokens": 232537370.0, + "step": 5987 + }, + { + "epoch": 0.7617351481999746, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.25787353515625, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8847174644470215, + "num_tokens": 232574320.0, + "step": 5988 + }, + { + "epoch": 0.7618623584785651, + "ewc_loss": 0.0115966796875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.821487426757812e-06, + "grad_norm": 8.253235816955566, + "learning_rate": 1e-06, + "loss": 0.3761, + "mean_token_accuracy": 0.8732221126556396, + "num_tokens": 232614191.0, + "step": 5989 + }, + { + "epoch": 0.7619895687571556, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.33930492401123, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8818750977516174, + "num_tokens": 232649190.0, + "step": 5990 + }, + { + "epoch": 0.762116779035746, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.351394653320312, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8753451108932495, + "num_tokens": 232682373.0, + "step": 5991 + }, + { + "epoch": 0.7622439893143366, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.224148750305176, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8837358951568604, + "num_tokens": 232722520.0, + "step": 5992 + }, + { + "epoch": 0.7623711995929271, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.22311782836914, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8857458829879761, + "num_tokens": 232762666.0, + "step": 5993 + }, + { + "epoch": 0.7624984098715176, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.2935152053833, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.878461480140686, + "num_tokens": 232804328.0, + "step": 5994 + }, + { + "epoch": 0.7626256201501082, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.21001148223877, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8835529088973999, + "num_tokens": 232843096.0, + "step": 5995 + }, + { + "epoch": 0.7627528304286987, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.325515747070312, + "learning_rate": 1e-06, + "loss": 0.3864, + "mean_token_accuracy": 0.8712975978851318, + "num_tokens": 232881853.0, + "step": 5996 + }, + { + "epoch": 0.7628800407072891, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.195554733276367, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.882391095161438, + "num_tokens": 232925274.0, + "step": 5997 + }, + { + "epoch": 0.7630072509858796, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.881092071533203e-06, + "grad_norm": 8.300741195678711, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8833937048912048, + "num_tokens": 232961108.0, + "step": 5998 + }, + { + "epoch": 0.7631344612644702, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.353362083435059, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8827345371246338, + "num_tokens": 232994177.0, + "step": 5999 + }, + { + "epoch": 0.7632616715430607, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.24549388885498, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8797523379325867, + "num_tokens": 233034887.0, + "step": 6000 + }, + { + "epoch": 0.7633888818216512, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.176127433776855, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8859208226203918, + "num_tokens": 233071995.0, + "step": 6001 + }, + { + "epoch": 0.7635160921002417, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.329085350036621, + "learning_rate": 1e-06, + "loss": 0.3689, + "mean_token_accuracy": 0.8745900392532349, + "num_tokens": 233108543.0, + "step": 6002 + }, + { + "epoch": 0.7636433023788322, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.286995887756348, + "learning_rate": 1e-06, + "loss": 0.4311, + "mean_token_accuracy": 0.8554774522781372, + "num_tokens": 233155067.0, + "step": 6003 + }, + { + "epoch": 0.7637705126574227, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.260660171508789, + "learning_rate": 1e-06, + "loss": 0.3872, + "mean_token_accuracy": 0.8706778287887573, + "num_tokens": 233196921.0, + "step": 6004 + }, + { + "epoch": 0.7638977229360132, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.308463096618652, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.886456310749054, + "num_tokens": 233234367.0, + "step": 6005 + }, + { + "epoch": 0.7640249332146037, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.274162292480469, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8795286417007446, + "num_tokens": 233272411.0, + "step": 6006 + }, + { + "epoch": 0.7641521434931943, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.282621383666992, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8709516525268555, + "num_tokens": 233305485.0, + "step": 6007 + }, + { + "epoch": 0.7642793537717848, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.273093223571777, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8834954500198364, + "num_tokens": 233343515.0, + "step": 6008 + }, + { + "epoch": 0.7644065640503753, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.127897262573242, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8855571746826172, + "num_tokens": 233388402.0, + "step": 6009 + }, + { + "epoch": 0.7645337743289657, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.282600402832031, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8830084800720215, + "num_tokens": 233426515.0, + "step": 6010 + }, + { + "epoch": 0.7646609846075563, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.293580055236816, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8875260353088379, + "num_tokens": 233463201.0, + "step": 6011 + }, + { + "epoch": 0.7647881948861468, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.314756393432617, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8745744228363037, + "num_tokens": 233502601.0, + "step": 6012 + }, + { + "epoch": 0.7649154051647373, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.298498153686523, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8816516399383545, + "num_tokens": 233536587.0, + "step": 6013 + }, + { + "epoch": 0.7650426154433279, + "ewc_loss": 0.01171875, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.273842811584473, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8738473057746887, + "num_tokens": 233574209.0, + "step": 6014 + }, + { + "epoch": 0.7651698257219184, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.316030502319336, + "learning_rate": 1e-06, + "loss": 0.4033, + "mean_token_accuracy": 0.8650472164154053, + "num_tokens": 233612205.0, + "step": 6015 + }, + { + "epoch": 0.7652970360005088, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.334474563598633, + "learning_rate": 1e-06, + "loss": 0.4133, + "mean_token_accuracy": 0.865549623966217, + "num_tokens": 233651001.0, + "step": 6016 + }, + { + "epoch": 0.7654242462790993, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.339213371276855, + "learning_rate": 1e-06, + "loss": 0.3951, + "mean_token_accuracy": 0.8656407594680786, + "num_tokens": 233689201.0, + "step": 6017 + }, + { + "epoch": 0.7655514565576899, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.23029613494873, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8767301440238953, + "num_tokens": 233731062.0, + "step": 6018 + }, + { + "epoch": 0.7656786668362804, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.264054298400879, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8788992166519165, + "num_tokens": 233768566.0, + "step": 6019 + }, + { + "epoch": 0.7658058771148709, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.177115440368652, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.894562840461731, + "num_tokens": 233811008.0, + "step": 6020 + }, + { + "epoch": 0.7659330873934614, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.36121654510498, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8724218010902405, + "num_tokens": 233844900.0, + "step": 6021 + }, + { + "epoch": 0.7660602976720519, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.241005897521973, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8866853713989258, + "num_tokens": 233884461.0, + "step": 6022 + }, + { + "epoch": 0.7661875079506424, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.34971809387207, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8879932165145874, + "num_tokens": 233914633.0, + "step": 6023 + }, + { + "epoch": 0.7663147182292329, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.362798690795898, + "learning_rate": 1e-06, + "loss": 0.3797, + "mean_token_accuracy": 0.8711783289909363, + "num_tokens": 233948791.0, + "step": 6024 + }, + { + "epoch": 0.7664419285078234, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.3132963180542, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8802506923675537, + "num_tokens": 233988495.0, + "step": 6025 + }, + { + "epoch": 0.766569138786414, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.777576446533203e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.225396156311035, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8823479413986206, + "num_tokens": 234027836.0, + "step": 6026 + }, + { + "epoch": 0.7666963490650045, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.294413566589355, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.889327883720398, + "num_tokens": 234065201.0, + "step": 6027 + }, + { + "epoch": 0.7668235593435949, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.200983047485352, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8871136903762817, + "num_tokens": 234104988.0, + "step": 6028 + }, + { + "epoch": 0.7669507696221854, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.201715469360352, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8799198269844055, + "num_tokens": 234145124.0, + "step": 6029 + }, + { + "epoch": 0.767077979900776, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.325077056884766, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8696775436401367, + "num_tokens": 234183647.0, + "step": 6030 + }, + { + "epoch": 0.7672051901793665, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.145761489868164, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8827446103096008, + "num_tokens": 234225698.0, + "step": 6031 + }, + { + "epoch": 0.767332400457957, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.322921752929688, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8748370409011841, + "num_tokens": 234261951.0, + "step": 6032 + }, + { + "epoch": 0.7674596107365476, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.288105964660645, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8801213502883911, + "num_tokens": 234296550.0, + "step": 6033 + }, + { + "epoch": 0.767586821015138, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.348079681396484, + "learning_rate": 1e-06, + "loss": 0.3876, + "mean_token_accuracy": 0.8709684014320374, + "num_tokens": 234333721.0, + "step": 6034 + }, + { + "epoch": 0.7677140312937285, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.324973106384277, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8781988620758057, + "num_tokens": 234363870.0, + "step": 6035 + }, + { + "epoch": 0.767841241572319, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.241564750671387, + "learning_rate": 1e-06, + "loss": 0.3846, + "mean_token_accuracy": 0.8687572479248047, + "num_tokens": 234403850.0, + "step": 6036 + }, + { + "epoch": 0.7679684518509096, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.278875350952148, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8752564191818237, + "num_tokens": 234435563.0, + "step": 6037 + }, + { + "epoch": 0.7680956621295001, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.256004333496094, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8850705027580261, + "num_tokens": 234476010.0, + "step": 6038 + }, + { + "epoch": 0.7682228724080906, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.296570777893066, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8751469254493713, + "num_tokens": 234510314.0, + "step": 6039 + }, + { + "epoch": 0.768350082686681, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.26138973236084, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8756423592567444, + "num_tokens": 234549292.0, + "step": 6040 + }, + { + "epoch": 0.7684772929652716, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.264474868774414, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8728840351104736, + "num_tokens": 234588560.0, + "step": 6041 + }, + { + "epoch": 0.7686045032438621, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.234527587890625, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8792586326599121, + "num_tokens": 234625998.0, + "step": 6042 + }, + { + "epoch": 0.7687317135224526, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.216829299926758, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8842052817344666, + "num_tokens": 234666847.0, + "step": 6043 + }, + { + "epoch": 0.7688589238010431, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.226936340332031, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8866751194000244, + "num_tokens": 234704241.0, + "step": 6044 + }, + { + "epoch": 0.7689861340796337, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.22952651977539, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8827491998672485, + "num_tokens": 234744513.0, + "step": 6045 + }, + { + "epoch": 0.7691133443582241, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.1754732131958, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8819631338119507, + "num_tokens": 234791780.0, + "step": 6046 + }, + { + "epoch": 0.7692405546368146, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.27262020111084, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8881046175956726, + "num_tokens": 234828569.0, + "step": 6047 + }, + { + "epoch": 0.7693677649154052, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.311811447143555, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.8655210733413696, + "num_tokens": 234867826.0, + "step": 6048 + }, + { + "epoch": 0.7694949751939957, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.243420600891113, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8928215503692627, + "num_tokens": 234900633.0, + "step": 6049 + }, + { + "epoch": 0.7696221854725862, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.189281463623047, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8902287483215332, + "num_tokens": 234942606.0, + "step": 6050 + }, + { + "epoch": 0.7697493957511767, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.7894973754882812e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.499777793884277, + "learning_rate": 1e-06, + "loss": 0.4171, + "mean_token_accuracy": 0.8643994331359863, + "num_tokens": 234977815.0, + "step": 6051 + }, + { + "epoch": 0.7698766060297672, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.153300285339355, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8850032091140747, + "num_tokens": 235021608.0, + "step": 6052 + }, + { + "epoch": 0.7700038163083577, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.323715209960938, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8720971941947937, + "num_tokens": 235058224.0, + "step": 6053 + }, + { + "epoch": 0.7701310265869482, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.254485130310059, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8889340758323669, + "num_tokens": 235098014.0, + "step": 6054 + }, + { + "epoch": 0.7702582368655387, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.24587631225586, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8771429061889648, + "num_tokens": 235136250.0, + "step": 6055 + }, + { + "epoch": 0.7703854471441293, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.276444435119629, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.882567286491394, + "num_tokens": 235176179.0, + "step": 6056 + }, + { + "epoch": 0.7705126574227198, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.214644432067871, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8735849857330322, + "num_tokens": 235217604.0, + "step": 6057 + }, + { + "epoch": 0.7706398677013102, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.473551750183105, + "learning_rate": 1e-06, + "loss": 0.4164, + "mean_token_accuracy": 0.8624131083488464, + "num_tokens": 235250072.0, + "step": 6058 + }, + { + "epoch": 0.7707670779799007, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.282299041748047, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8872923254966736, + "num_tokens": 235287066.0, + "step": 6059 + }, + { + "epoch": 0.7708942882584913, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.253922462463379, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8852415084838867, + "num_tokens": 235321328.0, + "step": 6060 + }, + { + "epoch": 0.7710214985370818, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.270776748657227, + "learning_rate": 1e-06, + "loss": 0.4233, + "mean_token_accuracy": 0.8598737716674805, + "num_tokens": 235365728.0, + "step": 6061 + }, + { + "epoch": 0.7711487088156723, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.194318771362305, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8787033557891846, + "num_tokens": 235406853.0, + "step": 6062 + }, + { + "epoch": 0.7712759190942629, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.350956916809082, + "learning_rate": 1e-06, + "loss": 0.3772, + "mean_token_accuracy": 0.8743318319320679, + "num_tokens": 235443806.0, + "step": 6063 + }, + { + "epoch": 0.7714031293728534, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.212590217590332, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8812015652656555, + "num_tokens": 235486977.0, + "step": 6064 + }, + { + "epoch": 0.7715303396514438, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.470173835754395, + "learning_rate": 1e-06, + "loss": 0.4165, + "mean_token_accuracy": 0.8585853576660156, + "num_tokens": 235518538.0, + "step": 6065 + }, + { + "epoch": 0.7716575499300343, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8133392333984375e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.212980270385742, + "learning_rate": 1e-06, + "loss": 0.3838, + "mean_token_accuracy": 0.8719220161437988, + "num_tokens": 235564320.0, + "step": 6066 + }, + { + "epoch": 0.7717847602086249, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8133392333984375e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.258916854858398, + "learning_rate": 1e-06, + "loss": 0.3941, + "mean_token_accuracy": 0.8668546676635742, + "num_tokens": 235604691.0, + "step": 6067 + }, + { + "epoch": 0.7719119704872154, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8133392333984375e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.358083724975586, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8789283633232117, + "num_tokens": 235643739.0, + "step": 6068 + }, + { + "epoch": 0.7720391807658059, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8133392333984375e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.27380084991455, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8791851997375488, + "num_tokens": 235681833.0, + "step": 6069 + }, + { + "epoch": 0.7721663910443964, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8133392333984375e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.232474327087402, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8997775912284851, + "num_tokens": 235724196.0, + "step": 6070 + }, + { + "epoch": 0.7722936013229869, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.250651359558105, + "learning_rate": 1e-06, + "loss": 0.3863, + "mean_token_accuracy": 0.8742559552192688, + "num_tokens": 235765971.0, + "step": 6071 + }, + { + "epoch": 0.7724208116015774, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.367088317871094, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.875940203666687, + "num_tokens": 235802801.0, + "step": 6072 + }, + { + "epoch": 0.7725480218801679, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.363655090332031, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8696302771568298, + "num_tokens": 235839390.0, + "step": 6073 + }, + { + "epoch": 0.7726752321587584, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.273451805114746, + "learning_rate": 1e-06, + "loss": 0.4215, + "mean_token_accuracy": 0.8580920696258545, + "num_tokens": 235880652.0, + "step": 6074 + }, + { + "epoch": 0.772802442437349, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.255412101745605, + "learning_rate": 1e-06, + "loss": 0.3978, + "mean_token_accuracy": 0.8705394864082336, + "num_tokens": 235919912.0, + "step": 6075 + }, + { + "epoch": 0.7729296527159395, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.343491554260254, + "learning_rate": 1e-06, + "loss": 0.3714, + "mean_token_accuracy": 0.8716508150100708, + "num_tokens": 235951635.0, + "step": 6076 + }, + { + "epoch": 0.7730568629945299, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.262069702148438, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8756393790245056, + "num_tokens": 235987928.0, + "step": 6077 + }, + { + "epoch": 0.7731840732731204, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.172718048095703, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8784283399581909, + "num_tokens": 236028958.0, + "step": 6078 + }, + { + "epoch": 0.773311283551711, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.214563369750977, + "learning_rate": 1e-06, + "loss": 0.3754, + "mean_token_accuracy": 0.8726403117179871, + "num_tokens": 236078479.0, + "step": 6079 + }, + { + "epoch": 0.7734384938303015, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8133392333984375e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.285402297973633, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.8721391558647156, + "num_tokens": 236116907.0, + "step": 6080 + }, + { + "epoch": 0.773565704108892, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.218692779541016, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8788620829582214, + "num_tokens": 236157618.0, + "step": 6081 + }, + { + "epoch": 0.7736929143874826, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8133392333984375e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.331146240234375, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8748751878738403, + "num_tokens": 236193587.0, + "step": 6082 + }, + { + "epoch": 0.773820124666073, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8014183044433594e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.308910369873047, + "learning_rate": 1e-06, + "loss": 0.3856, + "mean_token_accuracy": 0.8745984435081482, + "num_tokens": 236231212.0, + "step": 6083 + }, + { + "epoch": 0.7739473349446635, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8133392333984375e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.256646156311035, + "learning_rate": 1e-06, + "loss": 0.3659, + "mean_token_accuracy": 0.8752478361129761, + "num_tokens": 236271248.0, + "step": 6084 + }, + { + "epoch": 0.774074545223254, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8133392333984375e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.267314910888672, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8855429887771606, + "num_tokens": 236305236.0, + "step": 6085 + }, + { + "epoch": 0.7742017555018446, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8133392333984375e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.325723648071289, + "learning_rate": 1e-06, + "loss": 0.381, + "mean_token_accuracy": 0.8722497224807739, + "num_tokens": 236341437.0, + "step": 6086 + }, + { + "epoch": 0.7743289657804351, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8133392333984375e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.416974067687988, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.876356303691864, + "num_tokens": 236369556.0, + "step": 6087 + }, + { + "epoch": 0.7744561760590256, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.714971542358398, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8815865516662598, + "num_tokens": 236409287.0, + "step": 6088 + }, + { + "epoch": 0.774583386337616, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.216341018676758, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8853452205657959, + "num_tokens": 236443938.0, + "step": 6089 + }, + { + "epoch": 0.7747105966162066, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.24507999420166, + "learning_rate": 1e-06, + "loss": 0.4017, + "mean_token_accuracy": 0.8676702380180359, + "num_tokens": 236485933.0, + "step": 6090 + }, + { + "epoch": 0.7748378068947971, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.244189262390137, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.879692554473877, + "num_tokens": 236523440.0, + "step": 6091 + }, + { + "epoch": 0.7749650171733876, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.226641654968262, + "learning_rate": 1e-06, + "loss": 0.3954, + "mean_token_accuracy": 0.8665331602096558, + "num_tokens": 236562173.0, + "step": 6092 + }, + { + "epoch": 0.7750922274519781, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.277832984924316, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.877885639667511, + "num_tokens": 236599738.0, + "step": 6093 + }, + { + "epoch": 0.7752194377305687, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.30085277557373, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8793050050735474, + "num_tokens": 236634858.0, + "step": 6094 + }, + { + "epoch": 0.7753466480091591, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.21235466003418, + "learning_rate": 1e-06, + "loss": 0.3842, + "mean_token_accuracy": 0.8721387386322021, + "num_tokens": 236673685.0, + "step": 6095 + }, + { + "epoch": 0.7754738582877496, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.29495620727539, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8848689794540405, + "num_tokens": 236708174.0, + "step": 6096 + }, + { + "epoch": 0.7756010685663401, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.195502281188965, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8859632015228271, + "num_tokens": 236749331.0, + "step": 6097 + }, + { + "epoch": 0.7757282788449307, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.269065856933594, + "learning_rate": 1e-06, + "loss": 0.399, + "mean_token_accuracy": 0.8681686520576477, + "num_tokens": 236786953.0, + "step": 6098 + }, + { + "epoch": 0.7758554891235212, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.366154670715332, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8824979066848755, + "num_tokens": 236823740.0, + "step": 6099 + }, + { + "epoch": 0.7759826994021117, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.25296401977539, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8787376880645752, + "num_tokens": 236859194.0, + "step": 6100 + }, + { + "epoch": 0.7761099096807021, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.276659965515137, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8660649061203003, + "num_tokens": 236900570.0, + "step": 6101 + }, + { + "epoch": 0.7762371199592927, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.241206169128418, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.881127119064331, + "num_tokens": 236941673.0, + "step": 6102 + }, + { + "epoch": 0.7763643302378832, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.408695220947266, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8803397417068481, + "num_tokens": 236982651.0, + "step": 6103 + }, + { + "epoch": 0.7764915405164737, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.262650489807129, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8722501993179321, + "num_tokens": 237023806.0, + "step": 6104 + }, + { + "epoch": 0.7766187507950643, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.276436805725098, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.8744347095489502, + "num_tokens": 237061764.0, + "step": 6105 + }, + { + "epoch": 0.7767459610736548, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.264537811279297, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8829339742660522, + "num_tokens": 237099157.0, + "step": 6106 + }, + { + "epoch": 0.7768731713522452, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.263755798339844, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8842141628265381, + "num_tokens": 237135014.0, + "step": 6107 + }, + { + "epoch": 0.7770003816308357, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.33293628692627, + "learning_rate": 1e-06, + "loss": 0.3918, + "mean_token_accuracy": 0.870366096496582, + "num_tokens": 237170314.0, + "step": 6108 + }, + { + "epoch": 0.7771275919094263, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.425098419189453, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.8741810321807861, + "num_tokens": 237199617.0, + "step": 6109 + }, + { + "epoch": 0.7772548021880168, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.183741569519043, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.893830418586731, + "num_tokens": 237236007.0, + "step": 6110 + }, + { + "epoch": 0.7773820124666073, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.2490234375, + "learning_rate": 1e-06, + "loss": 0.3927, + "mean_token_accuracy": 0.8661603331565857, + "num_tokens": 237275868.0, + "step": 6111 + }, + { + "epoch": 0.7775092227451978, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.402314186096191, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8770064115524292, + "num_tokens": 237321039.0, + "step": 6112 + }, + { + "epoch": 0.7776364330237884, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.29554557800293, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8812583684921265, + "num_tokens": 237356643.0, + "step": 6113 + }, + { + "epoch": 0.7777636433023788, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.187516212463379, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8835729360580444, + "num_tokens": 237399018.0, + "step": 6114 + }, + { + "epoch": 0.7778908535809693, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.161514282226562, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8836636543273926, + "num_tokens": 237442873.0, + "step": 6115 + }, + { + "epoch": 0.7780180638595598, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.224786758422852, + "learning_rate": 1e-06, + "loss": 0.3886, + "mean_token_accuracy": 0.8694262504577637, + "num_tokens": 237491686.0, + "step": 6116 + }, + { + "epoch": 0.7781452741381504, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.36487102508545, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8810935020446777, + "num_tokens": 237526750.0, + "step": 6117 + }, + { + "epoch": 0.7782724844167409, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.236104965209961, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8824276924133301, + "num_tokens": 237568807.0, + "step": 6118 + }, + { + "epoch": 0.7783996946953314, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.225362777709961, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8818730115890503, + "num_tokens": 237607436.0, + "step": 6119 + }, + { + "epoch": 0.7785269049739219, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.330817222595215, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.8653777837753296, + "num_tokens": 237645047.0, + "step": 6120 + }, + { + "epoch": 0.7786541152525124, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.221441268920898, + "learning_rate": 1e-06, + "loss": 0.3724, + "mean_token_accuracy": 0.8748531937599182, + "num_tokens": 237684631.0, + "step": 6121 + }, + { + "epoch": 0.7787813255311029, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.345296859741211, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8865604400634766, + "num_tokens": 237716065.0, + "step": 6122 + }, + { + "epoch": 0.7789085358096934, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.23209285736084, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8900783061981201, + "num_tokens": 237752210.0, + "step": 6123 + }, + { + "epoch": 0.779035746088284, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.31984806060791, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8885345458984375, + "num_tokens": 237781796.0, + "step": 6124 + }, + { + "epoch": 0.7791629563668745, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.421330451965332, + "learning_rate": 1e-06, + "loss": 0.3944, + "mean_token_accuracy": 0.8660295605659485, + "num_tokens": 237814696.0, + "step": 6125 + }, + { + "epoch": 0.7792901666454649, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.212570190429688, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8907729983329773, + "num_tokens": 237849745.0, + "step": 6126 + }, + { + "epoch": 0.7794173769240554, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.303995132446289, + "learning_rate": 1e-06, + "loss": 0.4157, + "mean_token_accuracy": 0.8658456206321716, + "num_tokens": 237892948.0, + "step": 6127 + }, + { + "epoch": 0.779544587202646, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.274094581604004, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8804219365119934, + "num_tokens": 237932417.0, + "step": 6128 + }, + { + "epoch": 0.7796717974812365, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.256729125976562, + "learning_rate": 1e-06, + "loss": 0.3868, + "mean_token_accuracy": 0.8688226342201233, + "num_tokens": 237974050.0, + "step": 6129 + }, + { + "epoch": 0.779799007759827, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.328088760375977, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8701464533805847, + "num_tokens": 238013416.0, + "step": 6130 + }, + { + "epoch": 0.7799262180384176, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.380854606628418, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8746283054351807, + "num_tokens": 238052233.0, + "step": 6131 + }, + { + "epoch": 0.780053428317008, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8252601623535156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.404716491699219, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8848669528961182, + "num_tokens": 238084842.0, + "step": 6132 + }, + { + "epoch": 0.7801806385955985, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.280638694763184, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8891599178314209, + "num_tokens": 238119678.0, + "step": 6133 + }, + { + "epoch": 0.780307848874189, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.291061401367188, + "learning_rate": 1e-06, + "loss": 0.3936, + "mean_token_accuracy": 0.869331955909729, + "num_tokens": 238155960.0, + "step": 6134 + }, + { + "epoch": 0.7804350591527796, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.242046356201172, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.887235164642334, + "num_tokens": 238197410.0, + "step": 6135 + }, + { + "epoch": 0.7805622694313701, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.202499389648438, + "learning_rate": 1e-06, + "loss": 0.4049, + "mean_token_accuracy": 0.8654208779335022, + "num_tokens": 238240587.0, + "step": 6136 + }, + { + "epoch": 0.7806894797099606, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.243524551391602, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8916956782341003, + "num_tokens": 238280463.0, + "step": 6137 + }, + { + "epoch": 0.780816689988551, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.304850578308105, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8779754638671875, + "num_tokens": 238317332.0, + "step": 6138 + }, + { + "epoch": 0.7809439002671416, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.281617164611816, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8778682947158813, + "num_tokens": 238354068.0, + "step": 6139 + }, + { + "epoch": 0.7810711105457321, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.26722240447998, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.877312421798706, + "num_tokens": 238392622.0, + "step": 6140 + }, + { + "epoch": 0.7811983208243226, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.365178108215332, + "learning_rate": 1e-06, + "loss": 0.3995, + "mean_token_accuracy": 0.8674325942993164, + "num_tokens": 238428701.0, + "step": 6141 + }, + { + "epoch": 0.7813255311029131, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.33519172668457, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8795255422592163, + "num_tokens": 238465348.0, + "step": 6142 + }, + { + "epoch": 0.7814527413815037, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.237350463867188, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8740862011909485, + "num_tokens": 238506447.0, + "step": 6143 + }, + { + "epoch": 0.7815799516600941, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.47413158416748, + "learning_rate": 1e-06, + "loss": 0.3882, + "mean_token_accuracy": 0.8693580031394958, + "num_tokens": 238539733.0, + "step": 6144 + }, + { + "epoch": 0.7817071619386846, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.296273231506348, + "learning_rate": 1e-06, + "loss": 0.3809, + "mean_token_accuracy": 0.8727031946182251, + "num_tokens": 238578951.0, + "step": 6145 + }, + { + "epoch": 0.7818343722172751, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.273518562316895, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8783560395240784, + "num_tokens": 238619494.0, + "step": 6146 + }, + { + "epoch": 0.7819615824958657, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.487329483032227, + "learning_rate": 1e-06, + "loss": 0.4086, + "mean_token_accuracy": 0.8629746437072754, + "num_tokens": 238650901.0, + "step": 6147 + }, + { + "epoch": 0.7820887927744562, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.253519058227539, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8727198243141174, + "num_tokens": 238693372.0, + "step": 6148 + }, + { + "epoch": 0.7822160030530467, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.342557907104492, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.884753942489624, + "num_tokens": 238725319.0, + "step": 6149 + }, + { + "epoch": 0.7823432133316371, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.240009307861328, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8754277229309082, + "num_tokens": 238770276.0, + "step": 6150 + }, + { + "epoch": 0.7824704236102277, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.849102020263672e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.230352401733398, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8806310296058655, + "num_tokens": 238810500.0, + "step": 6151 + }, + { + "epoch": 0.7825976338888182, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.86102294921875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.36487865447998, + "learning_rate": 1e-06, + "loss": 0.3734, + "mean_token_accuracy": 0.8765759468078613, + "num_tokens": 238849956.0, + "step": 6152 + }, + { + "epoch": 0.7827248441674087, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.30589485168457, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8840078115463257, + "num_tokens": 238884138.0, + "step": 6153 + }, + { + "epoch": 0.7828520544459993, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.262627601623535, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8764638900756836, + "num_tokens": 238928321.0, + "step": 6154 + }, + { + "epoch": 0.7829792647245898, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.49325180053711, + "learning_rate": 1e-06, + "loss": 0.4013, + "mean_token_accuracy": 0.864174485206604, + "num_tokens": 238956714.0, + "step": 6155 + }, + { + "epoch": 0.7831064750031802, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.313176155090332, + "learning_rate": 1e-06, + "loss": 0.3925, + "mean_token_accuracy": 0.8653260469436646, + "num_tokens": 238992964.0, + "step": 6156 + }, + { + "epoch": 0.7832336852817707, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.259808540344238, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8750756978988647, + "num_tokens": 239031196.0, + "step": 6157 + }, + { + "epoch": 0.7833608955603613, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.360039710998535, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8771063089370728, + "num_tokens": 239067475.0, + "step": 6158 + }, + { + "epoch": 0.7834881058389518, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.285449981689453, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8788306713104248, + "num_tokens": 239100246.0, + "step": 6159 + }, + { + "epoch": 0.7836153161175423, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.323034286499023, + "learning_rate": 1e-06, + "loss": 0.3896, + "mean_token_accuracy": 0.8686739206314087, + "num_tokens": 239139508.0, + "step": 6160 + }, + { + "epoch": 0.7837425263961328, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.183060646057129, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8761085271835327, + "num_tokens": 239179393.0, + "step": 6161 + }, + { + "epoch": 0.7838697366747234, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.29511547088623, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8792105317115784, + "num_tokens": 239216297.0, + "step": 6162 + }, + { + "epoch": 0.7839969469533138, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.300018310546875, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8854235410690308, + "num_tokens": 239250669.0, + "step": 6163 + }, + { + "epoch": 0.7841241572319043, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.330239295959473, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8765160441398621, + "num_tokens": 239288144.0, + "step": 6164 + }, + { + "epoch": 0.7842513675104948, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.19672966003418, + "learning_rate": 1e-06, + "loss": 0.3689, + "mean_token_accuracy": 0.8756296634674072, + "num_tokens": 239330938.0, + "step": 6165 + }, + { + "epoch": 0.7843785777890854, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.204561233520508, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8947529792785645, + "num_tokens": 239369594.0, + "step": 6166 + }, + { + "epoch": 0.7845057880676759, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.218337059020996, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8901307582855225, + "num_tokens": 239407137.0, + "step": 6167 + }, + { + "epoch": 0.7846329983462664, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.271368026733398, + "learning_rate": 1e-06, + "loss": 0.3938, + "mean_token_accuracy": 0.8659894466400146, + "num_tokens": 239449069.0, + "step": 6168 + }, + { + "epoch": 0.7847602086248568, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.271685600280762, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8817429542541504, + "num_tokens": 239491507.0, + "step": 6169 + }, + { + "epoch": 0.7848874189034474, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.324399948120117, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8816097974777222, + "num_tokens": 239531587.0, + "step": 6170 + }, + { + "epoch": 0.7850146291820379, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.301739692687988, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.8684486746788025, + "num_tokens": 239569116.0, + "step": 6171 + }, + { + "epoch": 0.7851418394606284, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.218401908874512, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8776832818984985, + "num_tokens": 239614383.0, + "step": 6172 + }, + { + "epoch": 0.785269049739219, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.292686462402344, + "learning_rate": 1e-06, + "loss": 0.3788, + "mean_token_accuracy": 0.8714109659194946, + "num_tokens": 239649967.0, + "step": 6173 + }, + { + "epoch": 0.7853962600178095, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.289007186889648, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8874772787094116, + "num_tokens": 239690063.0, + "step": 6174 + }, + { + "epoch": 0.7855234702963999, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.279120445251465, + "learning_rate": 1e-06, + "loss": 0.4116, + "mean_token_accuracy": 0.8627546429634094, + "num_tokens": 239732382.0, + "step": 6175 + }, + { + "epoch": 0.7856506805749904, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.265288352966309, + "learning_rate": 1e-06, + "loss": 0.368, + "mean_token_accuracy": 0.8769117593765259, + "num_tokens": 239774068.0, + "step": 6176 + }, + { + "epoch": 0.785777890853581, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.21672534942627, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8843347430229187, + "num_tokens": 239813162.0, + "step": 6177 + }, + { + "epoch": 0.7859051011321715, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.524992942810059, + "learning_rate": 1e-06, + "loss": 0.3868, + "mean_token_accuracy": 0.8744301795959473, + "num_tokens": 239843425.0, + "step": 6178 + }, + { + "epoch": 0.786032311410762, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.270920753479004, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8859319686889648, + "num_tokens": 239880217.0, + "step": 6179 + }, + { + "epoch": 0.7861595216893525, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.337873458862305, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8765872120857239, + "num_tokens": 239921837.0, + "step": 6180 + }, + { + "epoch": 0.786286731967943, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.306205749511719, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.885917067527771, + "num_tokens": 239956426.0, + "step": 6181 + }, + { + "epoch": 0.7864139422465335, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.39002799987793, + "learning_rate": 1e-06, + "loss": 0.3879, + "mean_token_accuracy": 0.8694819211959839, + "num_tokens": 239991124.0, + "step": 6182 + }, + { + "epoch": 0.786541152525124, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.203848838806152, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8914594650268555, + "num_tokens": 240026115.0, + "step": 6183 + }, + { + "epoch": 0.7866683628037145, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.156017303466797, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8880683779716492, + "num_tokens": 240068725.0, + "step": 6184 + }, + { + "epoch": 0.7867955730823051, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.329025268554688, + "learning_rate": 1e-06, + "loss": 0.3976, + "mean_token_accuracy": 0.867392897605896, + "num_tokens": 240106468.0, + "step": 6185 + }, + { + "epoch": 0.7869227833608956, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.35388469696045, + "learning_rate": 1e-06, + "loss": 0.3977, + "mean_token_accuracy": 0.8662756681442261, + "num_tokens": 240145804.0, + "step": 6186 + }, + { + "epoch": 0.787049993639486, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.312664031982422, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8744904398918152, + "num_tokens": 240183384.0, + "step": 6187 + }, + { + "epoch": 0.7871772039180766, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.280351638793945, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8806105852127075, + "num_tokens": 240223930.0, + "step": 6188 + }, + { + "epoch": 0.7873044141966671, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.32253646850586, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8908898830413818, + "num_tokens": 240259395.0, + "step": 6189 + }, + { + "epoch": 0.7874316244752576, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.151175498962402, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8801246881484985, + "num_tokens": 240299243.0, + "step": 6190 + }, + { + "epoch": 0.7875588347538481, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.256359100341797, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8836057782173157, + "num_tokens": 240335008.0, + "step": 6191 + }, + { + "epoch": 0.7876860450324387, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.295466423034668, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8828953504562378, + "num_tokens": 240375000.0, + "step": 6192 + }, + { + "epoch": 0.7878132553110291, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.265868186950684, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8859680891036987, + "num_tokens": 240413786.0, + "step": 6193 + }, + { + "epoch": 0.7879404655896196, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.246087074279785, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.8752579689025879, + "num_tokens": 240457711.0, + "step": 6194 + }, + { + "epoch": 0.7880676758682101, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.226381301879883, + "learning_rate": 1e-06, + "loss": 0.3733, + "mean_token_accuracy": 0.8753130435943604, + "num_tokens": 240499085.0, + "step": 6195 + }, + { + "epoch": 0.7881948861468007, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.362250328063965, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8758261799812317, + "num_tokens": 240531417.0, + "step": 6196 + }, + { + "epoch": 0.7883220964253912, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.278825759887695, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8788118362426758, + "num_tokens": 240569545.0, + "step": 6197 + }, + { + "epoch": 0.7884493067039817, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.301846504211426, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8802619576454163, + "num_tokens": 240604148.0, + "step": 6198 + }, + { + "epoch": 0.7885765169825721, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.221718788146973, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8851561546325684, + "num_tokens": 240643739.0, + "step": 6199 + }, + { + "epoch": 0.7887037272611627, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.872943878173828e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.298151016235352, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8881053924560547, + "num_tokens": 240679457.0, + "step": 6200 + }, + { + "epoch": 0.7888309375397532, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.2544584274292, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8907462358474731, + "num_tokens": 240719494.0, + "step": 6201 + }, + { + "epoch": 0.7889581478183437, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.268404960632324, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8773263692855835, + "num_tokens": 240759688.0, + "step": 6202 + }, + { + "epoch": 0.7890853580969343, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.166738510131836, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8905968070030212, + "num_tokens": 240801024.0, + "step": 6203 + }, + { + "epoch": 0.7892125683755248, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.256500244140625, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.885287344455719, + "num_tokens": 240839207.0, + "step": 6204 + }, + { + "epoch": 0.7893397786541152, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.3615140914917, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8867915272712708, + "num_tokens": 240875223.0, + "step": 6205 + }, + { + "epoch": 0.7894669889327057, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.272414207458496, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8856818675994873, + "num_tokens": 240918224.0, + "step": 6206 + }, + { + "epoch": 0.7895941992112963, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.230452537536621, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8923050165176392, + "num_tokens": 240961313.0, + "step": 6207 + }, + { + "epoch": 0.7897214094898868, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.260841369628906, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8852982521057129, + "num_tokens": 240997205.0, + "step": 6208 + }, + { + "epoch": 0.7898486197684773, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.327213287353516, + "learning_rate": 1e-06, + "loss": 0.4202, + "mean_token_accuracy": 0.8623818755149841, + "num_tokens": 241036870.0, + "step": 6209 + }, + { + "epoch": 0.7899758300470678, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.301549911499023, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8824251890182495, + "num_tokens": 241078482.0, + "step": 6210 + }, + { + "epoch": 0.7901030403256584, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.346966743469238, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.882106602191925, + "num_tokens": 241113367.0, + "step": 6211 + }, + { + "epoch": 0.7902302506042488, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.282754898071289, + "learning_rate": 1e-06, + "loss": 0.3927, + "mean_token_accuracy": 0.8671567440032959, + "num_tokens": 241151493.0, + "step": 6212 + }, + { + "epoch": 0.7903574608828393, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.328561782836914, + "learning_rate": 1e-06, + "loss": 0.3815, + "mean_token_accuracy": 0.8674138188362122, + "num_tokens": 241192454.0, + "step": 6213 + }, + { + "epoch": 0.7904846711614298, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.409502983093262, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8777559995651245, + "num_tokens": 241222899.0, + "step": 6214 + }, + { + "epoch": 0.7906118814400204, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.20662784576416, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8948577642440796, + "num_tokens": 241264076.0, + "step": 6215 + }, + { + "epoch": 0.7907390917186109, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.252849578857422, + "learning_rate": 1e-06, + "loss": 0.3732, + "mean_token_accuracy": 0.8729594945907593, + "num_tokens": 241306967.0, + "step": 6216 + }, + { + "epoch": 0.7908663019972014, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.304200172424316, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8860822319984436, + "num_tokens": 241345526.0, + "step": 6217 + }, + { + "epoch": 0.7909935122757918, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.25892162322998, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8788213729858398, + "num_tokens": 241388854.0, + "step": 6218 + }, + { + "epoch": 0.7911207225543824, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.209878921508789, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8852400779724121, + "num_tokens": 241434096.0, + "step": 6219 + }, + { + "epoch": 0.7912479328329729, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.421351432800293, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.8694257140159607, + "num_tokens": 241468307.0, + "step": 6220 + }, + { + "epoch": 0.7913751431115634, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.23012638092041, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8869736194610596, + "num_tokens": 241509868.0, + "step": 6221 + }, + { + "epoch": 0.791502353390154, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.410073280334473, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8742648363113403, + "num_tokens": 241542778.0, + "step": 6222 + }, + { + "epoch": 0.7916295636687445, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.244580268859863, + "learning_rate": 1e-06, + "loss": 0.4013, + "mean_token_accuracy": 0.8658185601234436, + "num_tokens": 241590794.0, + "step": 6223 + }, + { + "epoch": 0.7917567739473349, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.271478652954102, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8774569034576416, + "num_tokens": 241633411.0, + "step": 6224 + }, + { + "epoch": 0.7918839842259254, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.223723411560059, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8866395354270935, + "num_tokens": 241675701.0, + "step": 6225 + }, + { + "epoch": 0.792011194504516, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.334526062011719, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8869802951812744, + "num_tokens": 241710958.0, + "step": 6226 + }, + { + "epoch": 0.7921384047831065, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.269636154174805, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8868919610977173, + "num_tokens": 241751937.0, + "step": 6227 + }, + { + "epoch": 0.792265615061697, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.199822425842285, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8781707286834717, + "num_tokens": 241794923.0, + "step": 6228 + }, + { + "epoch": 0.7923928253402875, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.481868743896484, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8860682249069214, + "num_tokens": 241834476.0, + "step": 6229 + }, + { + "epoch": 0.792520035618878, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.349260330200195, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.8733542561531067, + "num_tokens": 241874780.0, + "step": 6230 + }, + { + "epoch": 0.7926472458974685, + "ewc_loss": 0.0118408203125, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.200133323669434, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8998018503189087, + "num_tokens": 241912109.0, + "step": 6231 + }, + { + "epoch": 0.792774456176059, + "ewc_loss": 0.01177978515625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.46823501586914, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8766908645629883, + "num_tokens": 241941138.0, + "step": 6232 + }, + { + "epoch": 0.7929016664546495, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.223862648010254, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8731065988540649, + "num_tokens": 241982964.0, + "step": 6233 + }, + { + "epoch": 0.7930288767332401, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.426858901977539, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.8734124898910522, + "num_tokens": 242013992.0, + "step": 6234 + }, + { + "epoch": 0.7931560870118306, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.292546272277832, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.884702205657959, + "num_tokens": 242049812.0, + "step": 6235 + }, + { + "epoch": 0.793283297290421, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.181901931762695, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8810920119285583, + "num_tokens": 242094636.0, + "step": 6236 + }, + { + "epoch": 0.7934105075690115, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.304317474365234, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8832464218139648, + "num_tokens": 242133604.0, + "step": 6237 + }, + { + "epoch": 0.7935377178476021, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.354667663574219, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8750292062759399, + "num_tokens": 242168663.0, + "step": 6238 + }, + { + "epoch": 0.7936649281261926, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.275338172912598, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8838004469871521, + "num_tokens": 242209614.0, + "step": 6239 + }, + { + "epoch": 0.7937921384047831, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.363948822021484, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8776897192001343, + "num_tokens": 242242854.0, + "step": 6240 + }, + { + "epoch": 0.7939193486833737, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.307456016540527, + "learning_rate": 1e-06, + "loss": 0.4045, + "mean_token_accuracy": 0.8628351092338562, + "num_tokens": 242282156.0, + "step": 6241 + }, + { + "epoch": 0.7940465589619641, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.215192794799805, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8774574995040894, + "num_tokens": 242321790.0, + "step": 6242 + }, + { + "epoch": 0.7941737692405546, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.250570297241211, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8849916458129883, + "num_tokens": 242361125.0, + "step": 6243 + }, + { + "epoch": 0.7943009795191451, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.285064697265625, + "learning_rate": 1e-06, + "loss": 0.4016, + "mean_token_accuracy": 0.8625229001045227, + "num_tokens": 242400412.0, + "step": 6244 + }, + { + "epoch": 0.7944281897977357, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.266542434692383, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.876930832862854, + "num_tokens": 242437045.0, + "step": 6245 + }, + { + "epoch": 0.7945554000763262, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.298530578613281, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8821449875831604, + "num_tokens": 242472820.0, + "step": 6246 + }, + { + "epoch": 0.7946826103549167, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.231225967407227, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.8696185350418091, + "num_tokens": 242515302.0, + "step": 6247 + }, + { + "epoch": 0.7948098206335071, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.8848648071289062e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.312653541564941, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8759019374847412, + "num_tokens": 242557515.0, + "step": 6248 + }, + { + "epoch": 0.7949370309120977, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9087066650390625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.373664855957031, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.8663138151168823, + "num_tokens": 242595098.0, + "step": 6249 + }, + { + "epoch": 0.7950642411906882, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9087066650390625e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.272586822509766, + "learning_rate": 1e-06, + "loss": 0.3982, + "mean_token_accuracy": 0.8630068302154541, + "num_tokens": 242637651.0, + "step": 6250 + }, + { + "epoch": 0.7951914514692787, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9087066650390625e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.318252563476562, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8838208913803101, + "num_tokens": 242673648.0, + "step": 6251 + }, + { + "epoch": 0.7953186617478692, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9087066650390625e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.27720832824707, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8882010579109192, + "num_tokens": 242708306.0, + "step": 6252 + }, + { + "epoch": 0.7954458720264598, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.240260124206543, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8802675604820251, + "num_tokens": 242749083.0, + "step": 6253 + }, + { + "epoch": 0.7955730823050502, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.334822654724121, + "learning_rate": 1e-06, + "loss": 0.3978, + "mean_token_accuracy": 0.8644335269927979, + "num_tokens": 242786040.0, + "step": 6254 + }, + { + "epoch": 0.7957002925836407, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.174002647399902, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8804653286933899, + "num_tokens": 242828857.0, + "step": 6255 + }, + { + "epoch": 0.7958275028622313, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.30897331237793, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8808673024177551, + "num_tokens": 242863278.0, + "step": 6256 + }, + { + "epoch": 0.7959547131408218, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.387076377868652, + "learning_rate": 1e-06, + "loss": 0.3885, + "mean_token_accuracy": 0.8694348335266113, + "num_tokens": 242896909.0, + "step": 6257 + }, + { + "epoch": 0.7960819234194123, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.301198959350586, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.8751661777496338, + "num_tokens": 242935167.0, + "step": 6258 + }, + { + "epoch": 0.7962091336980028, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.228910446166992, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.879989743232727, + "num_tokens": 242977607.0, + "step": 6259 + }, + { + "epoch": 0.7963363439765934, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.308735847473145, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8850702047348022, + "num_tokens": 243016158.0, + "step": 6260 + }, + { + "epoch": 0.7964635542551838, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.28943920135498, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.8700727820396423, + "num_tokens": 243056116.0, + "step": 6261 + }, + { + "epoch": 0.7965907645337743, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.389914512634277, + "learning_rate": 1e-06, + "loss": 0.3971, + "mean_token_accuracy": 0.8656559586524963, + "num_tokens": 243091430.0, + "step": 6262 + }, + { + "epoch": 0.7967179748123648, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.260122299194336, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8830156922340393, + "num_tokens": 243126124.0, + "step": 6263 + }, + { + "epoch": 0.7968451850909554, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.305804252624512, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8789827823638916, + "num_tokens": 243163722.0, + "step": 6264 + }, + { + "epoch": 0.7969723953695459, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.389076232910156, + "learning_rate": 1e-06, + "loss": 0.3965, + "mean_token_accuracy": 0.8677988052368164, + "num_tokens": 243200696.0, + "step": 6265 + }, + { + "epoch": 0.7970996056481364, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.349396705627441, + "learning_rate": 1e-06, + "loss": 0.396, + "mean_token_accuracy": 0.8630008697509766, + "num_tokens": 243238930.0, + "step": 6266 + }, + { + "epoch": 0.7972268159267268, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.261494636535645, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8731076717376709, + "num_tokens": 243281310.0, + "step": 6267 + }, + { + "epoch": 0.7973540262053174, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.234885215759277, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8786583542823792, + "num_tokens": 243321423.0, + "step": 6268 + }, + { + "epoch": 0.7974812364839079, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.330438613891602, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8880104422569275, + "num_tokens": 243357799.0, + "step": 6269 + }, + { + "epoch": 0.7976084467624984, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.333073616027832, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8665669560432434, + "num_tokens": 243394387.0, + "step": 6270 + }, + { + "epoch": 0.797735657041089, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.311491966247559, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.883012056350708, + "num_tokens": 243426952.0, + "step": 6271 + }, + { + "epoch": 0.7978628673196795, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.325950622558594, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.885767936706543, + "num_tokens": 243464749.0, + "step": 6272 + }, + { + "epoch": 0.7979900775982699, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.280738830566406, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8789446353912354, + "num_tokens": 243503283.0, + "step": 6273 + }, + { + "epoch": 0.7981172878768604, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.260899543762207, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8755974769592285, + "num_tokens": 243543116.0, + "step": 6274 + }, + { + "epoch": 0.798244498155451, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.25612735748291, + "learning_rate": 1e-06, + "loss": 0.3922, + "mean_token_accuracy": 0.867301881313324, + "num_tokens": 243584053.0, + "step": 6275 + }, + { + "epoch": 0.7983717084340415, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.319042205810547, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8794883489608765, + "num_tokens": 243622168.0, + "step": 6276 + }, + { + "epoch": 0.798498918712632, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.250802040100098, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8946578502655029, + "num_tokens": 243663533.0, + "step": 6277 + }, + { + "epoch": 0.7986261289912225, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.34818172454834, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.874053955078125, + "num_tokens": 243701512.0, + "step": 6278 + }, + { + "epoch": 0.798753339269813, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.186092376708984, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8829843997955322, + "num_tokens": 243745729.0, + "step": 6279 + }, + { + "epoch": 0.7988805495484035, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.357868194580078, + "learning_rate": 1e-06, + "loss": 0.4167, + "mean_token_accuracy": 0.8630586862564087, + "num_tokens": 243786950.0, + "step": 6280 + }, + { + "epoch": 0.799007759826994, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.37109088897705, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8810387253761292, + "num_tokens": 243819494.0, + "step": 6281 + }, + { + "epoch": 0.7991349701055845, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.389379501342773, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.8684543967247009, + "num_tokens": 243852886.0, + "step": 6282 + }, + { + "epoch": 0.7992621803841751, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.312262535095215, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8923866152763367, + "num_tokens": 243885737.0, + "step": 6283 + }, + { + "epoch": 0.7993893906627656, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.341306686401367, + "learning_rate": 1e-06, + "loss": 0.3909, + "mean_token_accuracy": 0.865176796913147, + "num_tokens": 243917691.0, + "step": 6284 + }, + { + "epoch": 0.799516600941356, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.326333999633789, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8783296346664429, + "num_tokens": 243956688.0, + "step": 6285 + }, + { + "epoch": 0.7996438112199465, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.243563652038574, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8822002410888672, + "num_tokens": 243993535.0, + "step": 6286 + }, + { + "epoch": 0.7997710214985371, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.228976249694824, + "learning_rate": 1e-06, + "loss": 0.4018, + "mean_token_accuracy": 0.8642976880073547, + "num_tokens": 244039590.0, + "step": 6287 + }, + { + "epoch": 0.7998982317771276, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.319158554077148, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8784466981887817, + "num_tokens": 244076829.0, + "step": 6288 + }, + { + "epoch": 0.8000254420557181, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.228236198425293, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8848162889480591, + "num_tokens": 244118024.0, + "step": 6289 + }, + { + "epoch": 0.8001526523343087, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.38931655883789, + "learning_rate": 1e-06, + "loss": 0.3887, + "mean_token_accuracy": 0.8667522668838501, + "num_tokens": 244154899.0, + "step": 6290 + }, + { + "epoch": 0.8002798626128991, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.229351997375488, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8826913833618164, + "num_tokens": 244195177.0, + "step": 6291 + }, + { + "epoch": 0.8004070728914896, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.377240180969238, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.873356819152832, + "num_tokens": 244229279.0, + "step": 6292 + }, + { + "epoch": 0.8005342831700801, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.278369903564453, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.885028600692749, + "num_tokens": 244263920.0, + "step": 6293 + }, + { + "epoch": 0.8006614934486707, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.261022567749023, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8744183778762817, + "num_tokens": 244301671.0, + "step": 6294 + }, + { + "epoch": 0.8007887037272612, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.409262657165527, + "learning_rate": 1e-06, + "loss": 0.4217, + "mean_token_accuracy": 0.860589325428009, + "num_tokens": 244337984.0, + "step": 6295 + }, + { + "epoch": 0.8009159140058517, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.307520866394043, + "learning_rate": 1e-06, + "loss": 0.4124, + "mean_token_accuracy": 0.8580553531646729, + "num_tokens": 244373598.0, + "step": 6296 + }, + { + "epoch": 0.8010431242844421, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.349004745483398, + "learning_rate": 1e-06, + "loss": 0.3888, + "mean_token_accuracy": 0.8730099201202393, + "num_tokens": 244409699.0, + "step": 6297 + }, + { + "epoch": 0.8011703345630327, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.395868301391602, + "learning_rate": 1e-06, + "loss": 0.4112, + "mean_token_accuracy": 0.8562479019165039, + "num_tokens": 244443165.0, + "step": 6298 + }, + { + "epoch": 0.8012975448416232, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.272842407226562, + "learning_rate": 1e-06, + "loss": 0.3954, + "mean_token_accuracy": 0.8682036399841309, + "num_tokens": 244482532.0, + "step": 6299 + }, + { + "epoch": 0.8014247551202137, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.241602897644043, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8899730443954468, + "num_tokens": 244520897.0, + "step": 6300 + }, + { + "epoch": 0.8015519653988042, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.284177780151367, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8702712059020996, + "num_tokens": 244559664.0, + "step": 6301 + }, + { + "epoch": 0.8016791756773948, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.307085990905762, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.8833569288253784, + "num_tokens": 244594274.0, + "step": 6302 + }, + { + "epoch": 0.8018063859559852, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.382573127746582, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8797332048416138, + "num_tokens": 244628070.0, + "step": 6303 + }, + { + "epoch": 0.8019335962345757, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.178711891174316, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8933807611465454, + "num_tokens": 244669216.0, + "step": 6304 + }, + { + "epoch": 0.8020608065131662, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.251852989196777, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.892396867275238, + "num_tokens": 244708712.0, + "step": 6305 + }, + { + "epoch": 0.8021880167917568, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.339890480041504, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8798130750656128, + "num_tokens": 244749357.0, + "step": 6306 + }, + { + "epoch": 0.8023152270703473, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9206275939941406e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.505511283874512, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.8780968189239502, + "num_tokens": 244782155.0, + "step": 6307 + }, + { + "epoch": 0.8024424373489378, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.334600448608398, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.8696133494377136, + "num_tokens": 244820424.0, + "step": 6308 + }, + { + "epoch": 0.8025696476275284, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.310480117797852, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8749852180480957, + "num_tokens": 244853363.0, + "step": 6309 + }, + { + "epoch": 0.8026968579061188, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.289779663085938, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8747669458389282, + "num_tokens": 244889556.0, + "step": 6310 + }, + { + "epoch": 0.8028240681847093, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.317066192626953, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8847615718841553, + "num_tokens": 244926183.0, + "step": 6311 + }, + { + "epoch": 0.8029512784632998, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.425047874450684, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8771254420280457, + "num_tokens": 244960108.0, + "step": 6312 + }, + { + "epoch": 0.8030784887418904, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.251208305358887, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8693016171455383, + "num_tokens": 245003756.0, + "step": 6313 + }, + { + "epoch": 0.8032056990204809, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.268112182617188, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8835713863372803, + "num_tokens": 245043162.0, + "step": 6314 + }, + { + "epoch": 0.8033329092990714, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.252628326416016, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8862302303314209, + "num_tokens": 245081740.0, + "step": 6315 + }, + { + "epoch": 0.8034601195776618, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.36217212677002, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8740757703781128, + "num_tokens": 245121658.0, + "step": 6316 + }, + { + "epoch": 0.8035873298562524, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.299413681030273, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8875589370727539, + "num_tokens": 245157076.0, + "step": 6317 + }, + { + "epoch": 0.8037145401348429, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.267338752746582, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8855921030044556, + "num_tokens": 245198635.0, + "step": 6318 + }, + { + "epoch": 0.8038417504134334, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.381641387939453, + "learning_rate": 1e-06, + "loss": 0.4026, + "mean_token_accuracy": 0.8667446374893188, + "num_tokens": 245237638.0, + "step": 6319 + }, + { + "epoch": 0.803968960692024, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.48669719696045, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8745027780532837, + "num_tokens": 245276244.0, + "step": 6320 + }, + { + "epoch": 0.8040961709706145, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.356298446655273, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.872852623462677, + "num_tokens": 245311579.0, + "step": 6321 + }, + { + "epoch": 0.8042233812492049, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.259479522705078, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8822492957115173, + "num_tokens": 245349488.0, + "step": 6322 + }, + { + "epoch": 0.8043505915277954, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.418034553527832, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8796867728233337, + "num_tokens": 245382656.0, + "step": 6323 + }, + { + "epoch": 0.804477801806386, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.374738693237305, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8714029788970947, + "num_tokens": 245416954.0, + "step": 6324 + }, + { + "epoch": 0.8046050120849765, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.31718921661377, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.870517909526825, + "num_tokens": 245456934.0, + "step": 6325 + }, + { + "epoch": 0.804732222363567, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.304121017456055, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.869977593421936, + "num_tokens": 245496796.0, + "step": 6326 + }, + { + "epoch": 0.8048594326421575, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.280909538269043, + "learning_rate": 1e-06, + "loss": 0.4136, + "mean_token_accuracy": 0.8623659610748291, + "num_tokens": 245539137.0, + "step": 6327 + }, + { + "epoch": 0.804986642920748, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.28610897064209, + "learning_rate": 1e-06, + "loss": 0.4244, + "mean_token_accuracy": 0.8607577681541443, + "num_tokens": 245586717.0, + "step": 6328 + }, + { + "epoch": 0.8051138531993385, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.273785591125488, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8772463798522949, + "num_tokens": 245629288.0, + "step": 6329 + }, + { + "epoch": 0.805241063477929, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.307106971740723, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8784140348434448, + "num_tokens": 245665369.0, + "step": 6330 + }, + { + "epoch": 0.8053682737565195, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9325485229492188e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.3427152633667, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8812143802642822, + "num_tokens": 245702116.0, + "step": 6331 + }, + { + "epoch": 0.8054954840351101, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.2805757522583, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8797737956047058, + "num_tokens": 245738989.0, + "step": 6332 + }, + { + "epoch": 0.8056226943137006, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.324285507202148, + "learning_rate": 1e-06, + "loss": 0.4072, + "mean_token_accuracy": 0.861151397228241, + "num_tokens": 245780374.0, + "step": 6333 + }, + { + "epoch": 0.805749904592291, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.276394844055176, + "learning_rate": 1e-06, + "loss": 0.4055, + "mean_token_accuracy": 0.8607968091964722, + "num_tokens": 245821309.0, + "step": 6334 + }, + { + "epoch": 0.8058771148708815, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.242879867553711, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8859331607818604, + "num_tokens": 245858033.0, + "step": 6335 + }, + { + "epoch": 0.8060043251494721, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.231115341186523, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8778542876243591, + "num_tokens": 245901498.0, + "step": 6336 + }, + { + "epoch": 0.8061315354280626, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.24394702911377, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8806438446044922, + "num_tokens": 245937953.0, + "step": 6337 + }, + { + "epoch": 0.8062587457066531, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.273506164550781, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8791571259498596, + "num_tokens": 245979812.0, + "step": 6338 + }, + { + "epoch": 0.8063859559852437, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.295126914978027, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8853884935379028, + "num_tokens": 246018035.0, + "step": 6339 + }, + { + "epoch": 0.8065131662638341, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.247611045837402, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8797687292098999, + "num_tokens": 246059048.0, + "step": 6340 + }, + { + "epoch": 0.8066403765424246, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.268657684326172, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8815130591392517, + "num_tokens": 246100515.0, + "step": 6341 + }, + { + "epoch": 0.8067675868210151, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.357142448425293, + "learning_rate": 1e-06, + "loss": 0.3969, + "mean_token_accuracy": 0.8649153113365173, + "num_tokens": 246141861.0, + "step": 6342 + }, + { + "epoch": 0.8068947970996057, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.188684463500977, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8917998671531677, + "num_tokens": 246190098.0, + "step": 6343 + }, + { + "epoch": 0.8070220073781962, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.321121215820312, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8727260828018188, + "num_tokens": 246230812.0, + "step": 6344 + }, + { + "epoch": 0.8071492176567867, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.275473594665527, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8749855756759644, + "num_tokens": 246277248.0, + "step": 6345 + }, + { + "epoch": 0.8072764279353771, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.326863288879395, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8932061791419983, + "num_tokens": 246312110.0, + "step": 6346 + }, + { + "epoch": 0.8074036382139677, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.341228485107422, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8779940605163574, + "num_tokens": 246351492.0, + "step": 6347 + }, + { + "epoch": 0.8075308484925582, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.444635391235352, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.871225893497467, + "num_tokens": 246382871.0, + "step": 6348 + }, + { + "epoch": 0.8076580587711487, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.319640159606934, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8853710293769836, + "num_tokens": 246418698.0, + "step": 6349 + }, + { + "epoch": 0.8077852690497392, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.264018058776855, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8769797086715698, + "num_tokens": 246458876.0, + "step": 6350 + }, + { + "epoch": 0.8079124793283298, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.322012901306152, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8817922472953796, + "num_tokens": 246501934.0, + "step": 6351 + }, + { + "epoch": 0.8080396896069202, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.302328109741211, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8815964460372925, + "num_tokens": 246544187.0, + "step": 6352 + }, + { + "epoch": 0.8081668998855107, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.293166160583496, + "learning_rate": 1e-06, + "loss": 0.391, + "mean_token_accuracy": 0.8673480749130249, + "num_tokens": 246584319.0, + "step": 6353 + }, + { + "epoch": 0.8082941101641012, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.348374366760254, + "learning_rate": 1e-06, + "loss": 0.3881, + "mean_token_accuracy": 0.8699986934661865, + "num_tokens": 246623054.0, + "step": 6354 + }, + { + "epoch": 0.8084213204426918, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.255240440368652, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8926715850830078, + "num_tokens": 246664545.0, + "step": 6355 + }, + { + "epoch": 0.8085485307212823, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.274927139282227, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.880223274230957, + "num_tokens": 246705156.0, + "step": 6356 + }, + { + "epoch": 0.8086757409998728, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.347345352172852, + "learning_rate": 1e-06, + "loss": 0.413, + "mean_token_accuracy": 0.8573733568191528, + "num_tokens": 246746811.0, + "step": 6357 + }, + { + "epoch": 0.8088029512784632, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.288135528564453, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8843994140625, + "num_tokens": 246786341.0, + "step": 6358 + }, + { + "epoch": 0.8089301615570538, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.32724380493164, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8806280493736267, + "num_tokens": 246823543.0, + "step": 6359 + }, + { + "epoch": 0.8090573718356443, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.295784950256348, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8768329620361328, + "num_tokens": 246858564.0, + "step": 6360 + }, + { + "epoch": 0.8091845821142348, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.264822006225586, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8826903104782104, + "num_tokens": 246901246.0, + "step": 6361 + }, + { + "epoch": 0.8093117923928254, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.329100608825684, + "learning_rate": 1e-06, + "loss": 0.3935, + "mean_token_accuracy": 0.865619421005249, + "num_tokens": 246941968.0, + "step": 6362 + }, + { + "epoch": 0.8094390026714159, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.27419662475586, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8828544616699219, + "num_tokens": 246984573.0, + "step": 6363 + }, + { + "epoch": 0.8095662129500064, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.29969596862793, + "learning_rate": 1e-06, + "loss": 0.3716, + "mean_token_accuracy": 0.8748213648796082, + "num_tokens": 247024472.0, + "step": 6364 + }, + { + "epoch": 0.8096934232285968, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.279132843017578, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8897254467010498, + "num_tokens": 247063449.0, + "step": 6365 + }, + { + "epoch": 0.8098206335071874, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.304058074951172, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8907254934310913, + "num_tokens": 247096196.0, + "step": 6366 + }, + { + "epoch": 0.8099478437857779, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.944469451904297e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.250757217407227, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8884371519088745, + "num_tokens": 247135183.0, + "step": 6367 + }, + { + "epoch": 0.8100750540643684, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.956390380859375e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.243382453918457, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8924298286437988, + "num_tokens": 247175052.0, + "step": 6368 + }, + { + "epoch": 0.8102022643429589, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.956390380859375e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.200757026672363, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8844863176345825, + "num_tokens": 247218518.0, + "step": 6369 + }, + { + "epoch": 0.8103294746215495, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.956390380859375e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.399831771850586, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.887380838394165, + "num_tokens": 247255979.0, + "step": 6370 + }, + { + "epoch": 0.8104566849001399, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.956390380859375e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.266587257385254, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8815619349479675, + "num_tokens": 247297044.0, + "step": 6371 + }, + { + "epoch": 0.8105838951787304, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.956390380859375e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.250690460205078, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8768985271453857, + "num_tokens": 247339972.0, + "step": 6372 + }, + { + "epoch": 0.810711105457321, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.298501968383789, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8723567128181458, + "num_tokens": 247382546.0, + "step": 6373 + }, + { + "epoch": 0.8108383157359115, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.348687171936035, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8710311651229858, + "num_tokens": 247418588.0, + "step": 6374 + }, + { + "epoch": 0.810965526014502, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.346524238586426, + "learning_rate": 1e-06, + "loss": 0.4112, + "mean_token_accuracy": 0.8630303144454956, + "num_tokens": 247458315.0, + "step": 6375 + }, + { + "epoch": 0.8110927362930925, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.31517219543457, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8742362260818481, + "num_tokens": 247495003.0, + "step": 6376 + }, + { + "epoch": 0.811219946571683, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.364285469055176, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8784812688827515, + "num_tokens": 247532066.0, + "step": 6377 + }, + { + "epoch": 0.8113471568502735, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.300524711608887, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8861985206604004, + "num_tokens": 247571742.0, + "step": 6378 + }, + { + "epoch": 0.811474367128864, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.305708885192871, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8825719356536865, + "num_tokens": 247610122.0, + "step": 6379 + }, + { + "epoch": 0.8116015774074545, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.36609172821045, + "learning_rate": 1e-06, + "loss": 0.4013, + "mean_token_accuracy": 0.8729075193405151, + "num_tokens": 247647411.0, + "step": 6380 + }, + { + "epoch": 0.8117287876860451, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.335199356079102, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8777579069137573, + "num_tokens": 247683859.0, + "step": 6381 + }, + { + "epoch": 0.8118559979646356, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.462614059448242, + "learning_rate": 1e-06, + "loss": 0.3941, + "mean_token_accuracy": 0.8676227927207947, + "num_tokens": 247717271.0, + "step": 6382 + }, + { + "epoch": 0.811983208243226, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.372058868408203, + "learning_rate": 1e-06, + "loss": 0.4186, + "mean_token_accuracy": 0.8593514561653137, + "num_tokens": 247757851.0, + "step": 6383 + }, + { + "epoch": 0.8121104185218165, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.26734447479248, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8938981890678406, + "num_tokens": 247792543.0, + "step": 6384 + }, + { + "epoch": 0.8122376288004071, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.32646656036377, + "learning_rate": 1e-06, + "loss": 0.4065, + "mean_token_accuracy": 0.8685325384140015, + "num_tokens": 247833572.0, + "step": 6385 + }, + { + "epoch": 0.8123648390789976, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.318604469299316, + "learning_rate": 1e-06, + "loss": 0.3761, + "mean_token_accuracy": 0.8729649782180786, + "num_tokens": 247871954.0, + "step": 6386 + }, + { + "epoch": 0.8124920493575881, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.296110153198242, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8734652996063232, + "num_tokens": 247914474.0, + "step": 6387 + }, + { + "epoch": 0.8126192596361786, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.304832458496094, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.869817316532135, + "num_tokens": 247958107.0, + "step": 6388 + }, + { + "epoch": 0.8127464699147691, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.26125717163086, + "learning_rate": 1e-06, + "loss": 0.3733, + "mean_token_accuracy": 0.8734915256500244, + "num_tokens": 248005163.0, + "step": 6389 + }, + { + "epoch": 0.8128736801933596, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.263738632202148, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8787401914596558, + "num_tokens": 248048048.0, + "step": 6390 + }, + { + "epoch": 0.8130008904719501, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.275335311889648, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8725348711013794, + "num_tokens": 248086037.0, + "step": 6391 + }, + { + "epoch": 0.8131281007505406, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.397183418273926, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8850915431976318, + "num_tokens": 248120534.0, + "step": 6392 + }, + { + "epoch": 0.8132553110291312, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.297454833984375, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8770246505737305, + "num_tokens": 248160990.0, + "step": 6393 + }, + { + "epoch": 0.8133825213077217, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.290460586547852, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8813256621360779, + "num_tokens": 248200085.0, + "step": 6394 + }, + { + "epoch": 0.8135097315863121, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.266165733337402, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8732221126556396, + "num_tokens": 248244596.0, + "step": 6395 + }, + { + "epoch": 0.8136369418649027, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.33889102935791, + "learning_rate": 1e-06, + "loss": 0.368, + "mean_token_accuracy": 0.873563289642334, + "num_tokens": 248281519.0, + "step": 6396 + }, + { + "epoch": 0.8137641521434932, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.519979476928711, + "learning_rate": 1e-06, + "loss": 0.4115, + "mean_token_accuracy": 0.8584800362586975, + "num_tokens": 248311502.0, + "step": 6397 + }, + { + "epoch": 0.8138913624220837, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.27508544921875, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8831475973129272, + "num_tokens": 248350609.0, + "step": 6398 + }, + { + "epoch": 0.8140185727006742, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.321596145629883, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8757527470588684, + "num_tokens": 248387736.0, + "step": 6399 + }, + { + "epoch": 0.8141457829792648, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.288811683654785, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8813062310218811, + "num_tokens": 248427472.0, + "step": 6400 + }, + { + "epoch": 0.8142729932578552, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.402304649353027, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.880754292011261, + "num_tokens": 248458202.0, + "step": 6401 + }, + { + "epoch": 0.8144002035364457, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.306414604187012, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8940914869308472, + "num_tokens": 248493424.0, + "step": 6402 + }, + { + "epoch": 0.8145274138150362, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.248481750488281, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.8681451082229614, + "num_tokens": 248534485.0, + "step": 6403 + }, + { + "epoch": 0.8146546240936268, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.272417068481445, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8785423040390015, + "num_tokens": 248575825.0, + "step": 6404 + }, + { + "epoch": 0.8147818343722173, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.322839736938477, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8761657476425171, + "num_tokens": 248616828.0, + "step": 6405 + }, + { + "epoch": 0.8149090446508078, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.270195960998535, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.888931155204773, + "num_tokens": 248653155.0, + "step": 6406 + }, + { + "epoch": 0.8150362549293982, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.33585262298584, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8751217126846313, + "num_tokens": 248689279.0, + "step": 6407 + }, + { + "epoch": 0.8151634652079888, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.313175201416016, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.886547863483429, + "num_tokens": 248731531.0, + "step": 6408 + }, + { + "epoch": 0.8152906754865793, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.257304191589355, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8713303804397583, + "num_tokens": 248776859.0, + "step": 6409 + }, + { + "epoch": 0.8154178857651698, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.401859283447266, + "learning_rate": 1e-06, + "loss": 0.4332, + "mean_token_accuracy": 0.8592492341995239, + "num_tokens": 248815114.0, + "step": 6410 + }, + { + "epoch": 0.8155450960437604, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.233186721801758, + "learning_rate": 1e-06, + "loss": 0.3893, + "mean_token_accuracy": 0.8704501986503601, + "num_tokens": 248856396.0, + "step": 6411 + }, + { + "epoch": 0.8156723063223509, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.392776489257812, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8647644519805908, + "num_tokens": 248890668.0, + "step": 6412 + }, + { + "epoch": 0.8157995166009414, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.37607192993164, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8815582394599915, + "num_tokens": 248928372.0, + "step": 6413 + }, + { + "epoch": 0.8159267268795318, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.267788887023926, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8849881887435913, + "num_tokens": 248963834.0, + "step": 6414 + }, + { + "epoch": 0.8160539371581224, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.432247161865234, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8922209739685059, + "num_tokens": 248993436.0, + "step": 6415 + }, + { + "epoch": 0.8161811474367129, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.289734840393066, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8724677562713623, + "num_tokens": 249035047.0, + "step": 6416 + }, + { + "epoch": 0.8163083577153034, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.315561294555664, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.8722181916236877, + "num_tokens": 249076621.0, + "step": 6417 + }, + { + "epoch": 0.8164355679938939, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.319356918334961, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8738236427307129, + "num_tokens": 249115341.0, + "step": 6418 + }, + { + "epoch": 0.8165627782724845, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.309026718139648, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8900967240333557, + "num_tokens": 249151987.0, + "step": 6419 + }, + { + "epoch": 0.8166899885510749, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.261738777160645, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8789049386978149, + "num_tokens": 249194718.0, + "step": 6420 + }, + { + "epoch": 0.8168171988296654, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.379498481750488, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8729705810546875, + "num_tokens": 249230436.0, + "step": 6421 + }, + { + "epoch": 0.8169444091082559, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.285943984985352, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8892478942871094, + "num_tokens": 249273151.0, + "step": 6422 + }, + { + "epoch": 0.8170716193868465, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.283466339111328, + "learning_rate": 1e-06, + "loss": 0.402, + "mean_token_accuracy": 0.8625208735466003, + "num_tokens": 249317136.0, + "step": 6423 + }, + { + "epoch": 0.817198829665437, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.346733093261719, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8780879974365234, + "num_tokens": 249358443.0, + "step": 6424 + }, + { + "epoch": 0.8173260399440275, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.315011024475098, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8745809197425842, + "num_tokens": 249399230.0, + "step": 6425 + }, + { + "epoch": 0.8174532502226179, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.449751853942871, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8711072206497192, + "num_tokens": 249430884.0, + "step": 6426 + }, + { + "epoch": 0.8175804605012085, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.241469383239746, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8852908611297607, + "num_tokens": 249474312.0, + "step": 6427 + }, + { + "epoch": 0.817707670779799, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.261062622070312, + "learning_rate": 1e-06, + "loss": 0.3948, + "mean_token_accuracy": 0.8670423626899719, + "num_tokens": 249520907.0, + "step": 6428 + }, + { + "epoch": 0.8178348810583895, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.35484790802002, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8800687789916992, + "num_tokens": 249558650.0, + "step": 6429 + }, + { + "epoch": 0.8179620913369801, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.207372665405273, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8948802947998047, + "num_tokens": 249597459.0, + "step": 6430 + }, + { + "epoch": 0.8180893016155706, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.346414566040039, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8904472589492798, + "num_tokens": 249633634.0, + "step": 6431 + }, + { + "epoch": 0.818216511894161, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.285887718200684, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8826754093170166, + "num_tokens": 249672768.0, + "step": 6432 + }, + { + "epoch": 0.8183437221727515, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.28080940246582, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8885856866836548, + "num_tokens": 249708686.0, + "step": 6433 + }, + { + "epoch": 0.8184709324513421, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.309206008911133, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8871220350265503, + "num_tokens": 249749280.0, + "step": 6434 + }, + { + "epoch": 0.8185981427299326, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.346632957458496, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.873228907585144, + "num_tokens": 249790132.0, + "step": 6435 + }, + { + "epoch": 0.8187253530085231, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.3099946975708, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8823136687278748, + "num_tokens": 249826415.0, + "step": 6436 + }, + { + "epoch": 0.8188525632871136, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.315474510192871, + "learning_rate": 1e-06, + "loss": 0.4047, + "mean_token_accuracy": 0.8627046942710876, + "num_tokens": 249867627.0, + "step": 6437 + }, + { + "epoch": 0.8189797735657041, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.275997161865234, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8796652555465698, + "num_tokens": 249911046.0, + "step": 6438 + }, + { + "epoch": 0.8191069838442946, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.316529273986816, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8889189958572388, + "num_tokens": 249950121.0, + "step": 6439 + }, + { + "epoch": 0.8192341941228851, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.247925758361816, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8749328851699829, + "num_tokens": 249997027.0, + "step": 6440 + }, + { + "epoch": 0.8193614044014756, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.35468864440918, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8756422996520996, + "num_tokens": 250036510.0, + "step": 6441 + }, + { + "epoch": 0.8194886146800662, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.319072723388672, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8803557753562927, + "num_tokens": 250072538.0, + "step": 6442 + }, + { + "epoch": 0.8196158249586567, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.226689338684082, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8931862115859985, + "num_tokens": 250111021.0, + "step": 6443 + }, + { + "epoch": 0.8197430352372471, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.296485900878906, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8801104426383972, + "num_tokens": 250149293.0, + "step": 6444 + }, + { + "epoch": 0.8198702455158376, + "ewc_loss": 0.01190185546875, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 8.940696716308594e-06, + "grad_norm": 8.243488311767578, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8713036179542542, + "num_tokens": 250198619.0, + "step": 6445 + }, + { + "epoch": 0.8199974557944282, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.343104362487793, + "learning_rate": 1e-06, + "loss": 0.4191, + "mean_token_accuracy": 0.859332263469696, + "num_tokens": 250242777.0, + "step": 6446 + }, + { + "epoch": 0.8201246660730187, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.351139068603516, + "learning_rate": 1e-06, + "loss": 0.3847, + "mean_token_accuracy": 0.8753232955932617, + "num_tokens": 250278021.0, + "step": 6447 + }, + { + "epoch": 0.8202518763516092, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.225924491882324, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.8804616928100586, + "num_tokens": 250319497.0, + "step": 6448 + }, + { + "epoch": 0.8203790866301998, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.365513801574707, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8842706680297852, + "num_tokens": 250356639.0, + "step": 6449 + }, + { + "epoch": 0.8205062969087902, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.2676362991333, + "learning_rate": 1e-06, + "loss": 0.3858, + "mean_token_accuracy": 0.8699846267700195, + "num_tokens": 250400933.0, + "step": 6450 + }, + { + "epoch": 0.8206335071873807, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.253472328186035, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8945972919464111, + "num_tokens": 250441962.0, + "step": 6451 + }, + { + "epoch": 0.8207607174659712, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.247114181518555, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8797632455825806, + "num_tokens": 250483984.0, + "step": 6452 + }, + { + "epoch": 0.8208879277445618, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.42308235168457, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8756012916564941, + "num_tokens": 250519677.0, + "step": 6453 + }, + { + "epoch": 0.8210151380231523, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.264833450317383, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.8723533153533936, + "num_tokens": 250565774.0, + "step": 6454 + }, + { + "epoch": 0.8211423483017428, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.363399505615234, + "learning_rate": 1e-06, + "loss": 0.3881, + "mean_token_accuracy": 0.8706054091453552, + "num_tokens": 250608562.0, + "step": 6455 + }, + { + "epoch": 0.8212695585803332, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.32919692993164, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8765460848808289, + "num_tokens": 250644799.0, + "step": 6456 + }, + { + "epoch": 0.8213967688589238, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.238816261291504, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8649692535400391, + "num_tokens": 250688738.0, + "step": 6457 + }, + { + "epoch": 0.8215239791375143, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.434033393859863, + "learning_rate": 1e-06, + "loss": 0.4117, + "mean_token_accuracy": 0.8630793690681458, + "num_tokens": 250724989.0, + "step": 6458 + }, + { + "epoch": 0.8216511894161048, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.217727661132812, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8810423612594604, + "num_tokens": 250769821.0, + "step": 6459 + }, + { + "epoch": 0.8217783996946953, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.400740623474121, + "learning_rate": 1e-06, + "loss": 0.4012, + "mean_token_accuracy": 0.8660071492195129, + "num_tokens": 250810637.0, + "step": 6460 + }, + { + "epoch": 0.8219056099732859, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.281327247619629, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.8774856328964233, + "num_tokens": 250851409.0, + "step": 6461 + }, + { + "epoch": 0.8220328202518764, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.288639068603516, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8741029500961304, + "num_tokens": 250891981.0, + "step": 6462 + }, + { + "epoch": 0.8221600305304668, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.377204895019531, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.8788965940475464, + "num_tokens": 250929482.0, + "step": 6463 + }, + { + "epoch": 0.8222872408090574, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.269357681274414, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8813670873641968, + "num_tokens": 250963915.0, + "step": 6464 + }, + { + "epoch": 0.8224144510876479, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.343840599060059, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8764827847480774, + "num_tokens": 251000492.0, + "step": 6465 + }, + { + "epoch": 0.8225416613662384, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.306605339050293, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.8797614574432373, + "num_tokens": 251036499.0, + "step": 6466 + }, + { + "epoch": 0.8226688716448289, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.266277313232422, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8757572174072266, + "num_tokens": 251077822.0, + "step": 6467 + }, + { + "epoch": 0.8227960819234195, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.265641212463379, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8805602788925171, + "num_tokens": 251117085.0, + "step": 6468 + }, + { + "epoch": 0.8229232922020099, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.303608894348145, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8818924427032471, + "num_tokens": 251154714.0, + "step": 6469 + }, + { + "epoch": 0.8230505024806004, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.306083679199219, + "learning_rate": 1e-06, + "loss": 0.3863, + "mean_token_accuracy": 0.8691366314888, + "num_tokens": 251197371.0, + "step": 6470 + }, + { + "epoch": 0.8231777127591909, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.324284553527832, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8841860294342041, + "num_tokens": 251237643.0, + "step": 6471 + }, + { + "epoch": 0.8233049230377815, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.37285041809082, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8724139332771301, + "num_tokens": 251278592.0, + "step": 6472 + }, + { + "epoch": 0.823432133316372, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.269664764404297, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8733588457107544, + "num_tokens": 251327277.0, + "step": 6473 + }, + { + "epoch": 0.8235593435949625, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.436630249023438, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8824842572212219, + "num_tokens": 251360255.0, + "step": 6474 + }, + { + "epoch": 0.8236865538735529, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.34041690826416, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.88138747215271, + "num_tokens": 251399454.0, + "step": 6475 + }, + { + "epoch": 0.8238137641521435, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.273383140563965, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8908392190933228, + "num_tokens": 251438079.0, + "step": 6476 + }, + { + "epoch": 0.823940974430734, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.36998462677002, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8768050074577332, + "num_tokens": 251476264.0, + "step": 6477 + }, + { + "epoch": 0.8240681847093245, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.338447570800781, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8747929334640503, + "num_tokens": 251514658.0, + "step": 6478 + }, + { + "epoch": 0.824195394987915, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.31503963470459, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8839285969734192, + "num_tokens": 251554744.0, + "step": 6479 + }, + { + "epoch": 0.8243226052665056, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.361164093017578, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8729965090751648, + "num_tokens": 251593150.0, + "step": 6480 + }, + { + "epoch": 0.824449815545096, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.224462509155273, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.888027548789978, + "num_tokens": 251636713.0, + "step": 6481 + }, + { + "epoch": 0.8245770258236865, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.353941917419434, + "learning_rate": 1e-06, + "loss": 0.4148, + "mean_token_accuracy": 0.8628076314926147, + "num_tokens": 251679198.0, + "step": 6482 + }, + { + "epoch": 0.824704236102277, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.378785133361816, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.878379225730896, + "num_tokens": 251715795.0, + "step": 6483 + }, + { + "epoch": 0.8248314463808676, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.361113548278809, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.8770469427108765, + "num_tokens": 251756648.0, + "step": 6484 + }, + { + "epoch": 0.8249586566594581, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.273627281188965, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8877604603767395, + "num_tokens": 251796917.0, + "step": 6485 + }, + { + "epoch": 0.8250858669380486, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.302726745605469, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8864206075668335, + "num_tokens": 251840598.0, + "step": 6486 + }, + { + "epoch": 0.8252130772166391, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.246886253356934, + "learning_rate": 1e-06, + "loss": 0.3734, + "mean_token_accuracy": 0.8766881823539734, + "num_tokens": 251884099.0, + "step": 6487 + }, + { + "epoch": 0.8253402874952296, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.387145042419434, + "learning_rate": 1e-06, + "loss": 0.4165, + "mean_token_accuracy": 0.8568729758262634, + "num_tokens": 251924869.0, + "step": 6488 + }, + { + "epoch": 0.8254674977738201, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.968311309814453e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.347087860107422, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8848596811294556, + "num_tokens": 251958156.0, + "step": 6489 + }, + { + "epoch": 0.8255947080524106, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.277464866638184, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8814680576324463, + "num_tokens": 251998494.0, + "step": 6490 + }, + { + "epoch": 0.8257219183310012, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.2706937789917, + "learning_rate": 1e-06, + "loss": 0.3643, + "mean_token_accuracy": 0.8754953145980835, + "num_tokens": 252041676.0, + "step": 6491 + }, + { + "epoch": 0.8258491286095917, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.375044822692871, + "learning_rate": 1e-06, + "loss": 0.386, + "mean_token_accuracy": 0.8696130514144897, + "num_tokens": 252078230.0, + "step": 6492 + }, + { + "epoch": 0.8259763388881821, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.30601978302002, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8836076259613037, + "num_tokens": 252118765.0, + "step": 6493 + }, + { + "epoch": 0.8261035491667726, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.285350799560547, + "learning_rate": 1e-06, + "loss": 0.4271, + "mean_token_accuracy": 0.8601138591766357, + "num_tokens": 252166867.0, + "step": 6494 + }, + { + "epoch": 0.8262307594453632, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.364087104797363, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8878719210624695, + "num_tokens": 252203257.0, + "step": 6495 + }, + { + "epoch": 0.8263579697239537, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.263317108154297, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8792951703071594, + "num_tokens": 252245073.0, + "step": 6496 + }, + { + "epoch": 0.8264851800025442, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.358452796936035, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8836615085601807, + "num_tokens": 252283441.0, + "step": 6497 + }, + { + "epoch": 0.8266123902811348, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.331734657287598, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8880859613418579, + "num_tokens": 252320436.0, + "step": 6498 + }, + { + "epoch": 0.8267396005597252, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.285484313964844, + "learning_rate": 1e-06, + "loss": 0.4127, + "mean_token_accuracy": 0.863707959651947, + "num_tokens": 252360694.0, + "step": 6499 + }, + { + "epoch": 0.8268668108383157, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.398202896118164, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8730885982513428, + "num_tokens": 252397745.0, + "step": 6500 + }, + { + "epoch": 0.8269940211169062, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.204133033752441, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8792617321014404, + "num_tokens": 252444891.0, + "step": 6501 + }, + { + "epoch": 0.8271212313954968, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.341383934020996, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8761370182037354, + "num_tokens": 252483616.0, + "step": 6502 + }, + { + "epoch": 0.8272484416740873, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.384428024291992, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.870015025138855, + "num_tokens": 252519966.0, + "step": 6503 + }, + { + "epoch": 0.8273756519526778, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.35285472869873, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8818312883377075, + "num_tokens": 252556085.0, + "step": 6504 + }, + { + "epoch": 0.8275028622312682, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.357464790344238, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8767409324645996, + "num_tokens": 252594026.0, + "step": 6505 + }, + { + "epoch": 0.8276300725098588, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.364727020263672, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8853684663772583, + "num_tokens": 252626402.0, + "step": 6506 + }, + { + "epoch": 0.8277572827884493, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.297465324401855, + "learning_rate": 1e-06, + "loss": 0.3859, + "mean_token_accuracy": 0.8690606355667114, + "num_tokens": 252666064.0, + "step": 6507 + }, + { + "epoch": 0.8278844930670398, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.507970809936523, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8740174770355225, + "num_tokens": 252705579.0, + "step": 6508 + }, + { + "epoch": 0.8280117033456303, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.22234058380127, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8840007781982422, + "num_tokens": 252748011.0, + "step": 6509 + }, + { + "epoch": 0.8281389136242209, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.422497749328613, + "learning_rate": 1e-06, + "loss": 0.4214, + "mean_token_accuracy": 0.8627339005470276, + "num_tokens": 252784877.0, + "step": 6510 + }, + { + "epoch": 0.8282661239028114, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.304704666137695, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8814582228660583, + "num_tokens": 252825727.0, + "step": 6511 + }, + { + "epoch": 0.8283933341814018, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.319597244262695, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8809118270874023, + "num_tokens": 252862830.0, + "step": 6512 + }, + { + "epoch": 0.8285205444599923, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.415079116821289, + "learning_rate": 1e-06, + "loss": 0.4073, + "mean_token_accuracy": 0.8633596897125244, + "num_tokens": 252898743.0, + "step": 6513 + }, + { + "epoch": 0.8286477547385829, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.361087799072266, + "learning_rate": 1e-06, + "loss": 0.3784, + "mean_token_accuracy": 0.8732355833053589, + "num_tokens": 252934143.0, + "step": 6514 + }, + { + "epoch": 0.8287749650171734, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.264374732971191, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8910719752311707, + "num_tokens": 252974931.0, + "step": 6515 + }, + { + "epoch": 0.8289021752957639, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.274733543395996, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8891806602478027, + "num_tokens": 253014690.0, + "step": 6516 + }, + { + "epoch": 0.8290293855743545, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.431292533874512, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.88152015209198, + "num_tokens": 253046358.0, + "step": 6517 + }, + { + "epoch": 0.8291565958529449, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.323716163635254, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8798353672027588, + "num_tokens": 253082483.0, + "step": 6518 + }, + { + "epoch": 0.8292838061315354, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.346158027648926, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8808703422546387, + "num_tokens": 253121179.0, + "step": 6519 + }, + { + "epoch": 0.8294110164101259, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.264608383178711, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8759247064590454, + "num_tokens": 253165610.0, + "step": 6520 + }, + { + "epoch": 0.8295382266887165, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.385844230651855, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8791472911834717, + "num_tokens": 253200550.0, + "step": 6521 + }, + { + "epoch": 0.829665436967307, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.39015007019043, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8792086839675903, + "num_tokens": 253238613.0, + "step": 6522 + }, + { + "epoch": 0.8297926472458975, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.364325523376465, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8823931217193604, + "num_tokens": 253276608.0, + "step": 6523 + }, + { + "epoch": 0.8299198575244879, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.253546714782715, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8800723552703857, + "num_tokens": 253317931.0, + "step": 6524 + }, + { + "epoch": 0.8300470678030785, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.38202953338623, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.880966067314148, + "num_tokens": 253358171.0, + "step": 6525 + }, + { + "epoch": 0.830174278081669, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.361466407775879, + "learning_rate": 1e-06, + "loss": 0.381, + "mean_token_accuracy": 0.8706832528114319, + "num_tokens": 253394915.0, + "step": 6526 + }, + { + "epoch": 0.8303014883602595, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.457792282104492, + "learning_rate": 1e-06, + "loss": 0.4484, + "mean_token_accuracy": 0.8520380258560181, + "num_tokens": 253435279.0, + "step": 6527 + }, + { + "epoch": 0.83042869863885, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.285938262939453, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.870201587677002, + "num_tokens": 253478624.0, + "step": 6528 + }, + { + "epoch": 0.8305559089174406, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.2681303024292, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8790668845176697, + "num_tokens": 253516343.0, + "step": 6529 + }, + { + "epoch": 0.830683119196031, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.431982040405273, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8837924003601074, + "num_tokens": 253548058.0, + "step": 6530 + }, + { + "epoch": 0.8308103294746215, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.29345417022705, + "learning_rate": 1e-06, + "loss": 0.3776, + "mean_token_accuracy": 0.8711965084075928, + "num_tokens": 253587771.0, + "step": 6531 + }, + { + "epoch": 0.830937539753212, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.374503135681152, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8841321468353271, + "num_tokens": 253626934.0, + "step": 6532 + }, + { + "epoch": 0.8310647500318026, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9802322387695312e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.295453071594238, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8738198280334473, + "num_tokens": 253666753.0, + "step": 6533 + }, + { + "epoch": 0.8311919603103931, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9921531677246094e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.337423324584961, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.8746439814567566, + "num_tokens": 253708040.0, + "step": 6534 + }, + { + "epoch": 0.8313191705889836, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9921531677246094e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.329668998718262, + "learning_rate": 1e-06, + "loss": 0.3851, + "mean_token_accuracy": 0.8730169534683228, + "num_tokens": 253752319.0, + "step": 6535 + }, + { + "epoch": 0.831446380867574, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9921531677246094e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.246394157409668, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8916504383087158, + "num_tokens": 253794948.0, + "step": 6536 + }, + { + "epoch": 0.8315735911461646, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9921531677246094e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.38862133026123, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8822617530822754, + "num_tokens": 253830026.0, + "step": 6537 + }, + { + "epoch": 0.8317008014247551, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 2.9921531677246094e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.308133125305176, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8791770935058594, + "num_tokens": 253869477.0, + "step": 6538 + }, + { + "epoch": 0.8318280117033456, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9921531677246094e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.388055801391602, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8760959506034851, + "num_tokens": 253910177.0, + "step": 6539 + }, + { + "epoch": 0.8319552219819362, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9921531677246094e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.280067443847656, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.883065938949585, + "num_tokens": 253948924.0, + "step": 6540 + }, + { + "epoch": 0.8320824322605267, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 2.9921531677246094e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.355573654174805, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8803989887237549, + "num_tokens": 253984455.0, + "step": 6541 + }, + { + "epoch": 0.8322096425391171, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.281522750854492, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8901576995849609, + "num_tokens": 254025982.0, + "step": 6542 + }, + { + "epoch": 0.8323368528177076, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.259708404541016, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.887984037399292, + "num_tokens": 254069529.0, + "step": 6543 + }, + { + "epoch": 0.8324640630962982, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.31908130645752, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8880972862243652, + "num_tokens": 254109333.0, + "step": 6544 + }, + { + "epoch": 0.8325912733748887, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.362576484680176, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8804259300231934, + "num_tokens": 254146698.0, + "step": 6545 + }, + { + "epoch": 0.8327184836534792, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.34432601928711, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.877244234085083, + "num_tokens": 254186833.0, + "step": 6546 + }, + { + "epoch": 0.8328456939320698, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.326154708862305, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8900134563446045, + "num_tokens": 254220502.0, + "step": 6547 + }, + { + "epoch": 0.8329729042106602, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.424558639526367, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8777123689651489, + "num_tokens": 254258563.0, + "step": 6548 + }, + { + "epoch": 0.8331001144892507, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.361776351928711, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8785601854324341, + "num_tokens": 254300260.0, + "step": 6549 + }, + { + "epoch": 0.8332273247678412, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.464818954467773, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.8663716316223145, + "num_tokens": 254332558.0, + "step": 6550 + }, + { + "epoch": 0.8333545350464318, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.4005126953125, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8720902800559998, + "num_tokens": 254370953.0, + "step": 6551 + }, + { + "epoch": 0.8334817453250223, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.340348243713379, + "learning_rate": 1e-06, + "loss": 0.3892, + "mean_token_accuracy": 0.8696898818016052, + "num_tokens": 254413548.0, + "step": 6552 + }, + { + "epoch": 0.8336089556036128, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.623396873474121, + "learning_rate": 1e-06, + "loss": 0.3961, + "mean_token_accuracy": 0.8636382818222046, + "num_tokens": 254444952.0, + "step": 6553 + }, + { + "epoch": 0.8337361658822032, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.295226097106934, + "learning_rate": 1e-06, + "loss": 0.4058, + "mean_token_accuracy": 0.8604736328125, + "num_tokens": 254487377.0, + "step": 6554 + }, + { + "epoch": 0.8338633761607938, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.325166702270508, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8859056234359741, + "num_tokens": 254526578.0, + "step": 6555 + }, + { + "epoch": 0.8339905864393843, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.33043098449707, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8853079080581665, + "num_tokens": 254564098.0, + "step": 6556 + }, + { + "epoch": 0.8341177967179748, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.228540420532227, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8818469643592834, + "num_tokens": 254605210.0, + "step": 6557 + }, + { + "epoch": 0.8342450069965653, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.437896728515625, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8814975023269653, + "num_tokens": 254639916.0, + "step": 6558 + }, + { + "epoch": 0.8343722172751559, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.260889053344727, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8845785856246948, + "num_tokens": 254677050.0, + "step": 6559 + }, + { + "epoch": 0.8344994275537464, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.335299491882324, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8800031542778015, + "num_tokens": 254712612.0, + "step": 6560 + }, + { + "epoch": 0.8346266378323368, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.314169883728027, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.879152774810791, + "num_tokens": 254754931.0, + "step": 6561 + }, + { + "epoch": 0.8347538481109273, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.444196701049805, + "learning_rate": 1e-06, + "loss": 0.4, + "mean_token_accuracy": 0.8669710159301758, + "num_tokens": 254788956.0, + "step": 6562 + }, + { + "epoch": 0.8348810583895179, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.335931777954102, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.876732587814331, + "num_tokens": 254829861.0, + "step": 6563 + }, + { + "epoch": 0.8350082686681084, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.270760536193848, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.8760518431663513, + "num_tokens": 254874613.0, + "step": 6564 + }, + { + "epoch": 0.8351354789466989, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.394947052001953, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8768678903579712, + "num_tokens": 254916296.0, + "step": 6565 + }, + { + "epoch": 0.8352626892252895, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.410240173339844, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.879856288433075, + "num_tokens": 254955628.0, + "step": 6566 + }, + { + "epoch": 0.8353898995038799, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.295221328735352, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8773240447044373, + "num_tokens": 254998948.0, + "step": 6567 + }, + { + "epoch": 0.8355171097824704, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.356053352355957, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.889046311378479, + "num_tokens": 255033174.0, + "step": 6568 + }, + { + "epoch": 0.8356443200610609, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.402292251586914, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.8801812529563904, + "num_tokens": 255071519.0, + "step": 6569 + }, + { + "epoch": 0.8357715303396515, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.373023986816406, + "learning_rate": 1e-06, + "loss": 0.4205, + "mean_token_accuracy": 0.8616669178009033, + "num_tokens": 255109826.0, + "step": 6570 + }, + { + "epoch": 0.835898740618242, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.326615333557129, + "learning_rate": 1e-06, + "loss": 0.3832, + "mean_token_accuracy": 0.8707420825958252, + "num_tokens": 255155966.0, + "step": 6571 + }, + { + "epoch": 0.8360259508968325, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.385592460632324, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8896387815475464, + "num_tokens": 255196073.0, + "step": 6572 + }, + { + "epoch": 0.8361531611754229, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.460230827331543, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8782175779342651, + "num_tokens": 255231674.0, + "step": 6573 + }, + { + "epoch": 0.8362803714540135, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.33449649810791, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8759868144989014, + "num_tokens": 255273345.0, + "step": 6574 + }, + { + "epoch": 0.836407581732604, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.260075569152832, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8783811330795288, + "num_tokens": 255321130.0, + "step": 6575 + }, + { + "epoch": 0.8365347920111945, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.352507591247559, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8727688193321228, + "num_tokens": 255364359.0, + "step": 6576 + }, + { + "epoch": 0.836662002289785, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.342792510986328, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.872024655342102, + "num_tokens": 255404104.0, + "step": 6577 + }, + { + "epoch": 0.8367892125683756, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.314800262451172, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8863683342933655, + "num_tokens": 255443435.0, + "step": 6578 + }, + { + "epoch": 0.836916422846966, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.244797706604004, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8971148133277893, + "num_tokens": 255487626.0, + "step": 6579 + }, + { + "epoch": 0.8370436331255565, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.29018783569336, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.880143404006958, + "num_tokens": 255528914.0, + "step": 6580 + }, + { + "epoch": 0.837170843404147, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.40636157989502, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8716059923171997, + "num_tokens": 255564729.0, + "step": 6581 + }, + { + "epoch": 0.8372980536827376, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.300586700439453, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8774340152740479, + "num_tokens": 255605196.0, + "step": 6582 + }, + { + "epoch": 0.8374252639613281, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.29599380493164, + "learning_rate": 1e-06, + "loss": 0.3935, + "mean_token_accuracy": 0.8696871399879456, + "num_tokens": 255650667.0, + "step": 6583 + }, + { + "epoch": 0.8375524742399186, + "ewc_loss": 0.01202392578125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.29813289642334, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8838019371032715, + "num_tokens": 255693307.0, + "step": 6584 + }, + { + "epoch": 0.837679684518509, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.280800819396973, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8893377780914307, + "num_tokens": 255727887.0, + "step": 6585 + }, + { + "epoch": 0.8378068947970996, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.481648445129395, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8793597221374512, + "num_tokens": 255757062.0, + "step": 6586 + }, + { + "epoch": 0.8379341050756901, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.280190467834473, + "learning_rate": 1e-06, + "loss": 0.4087, + "mean_token_accuracy": 0.8645728230476379, + "num_tokens": 255803486.0, + "step": 6587 + }, + { + "epoch": 0.8380613153542806, + "ewc_loss": 0.011962890625, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.31989574432373, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8768956661224365, + "num_tokens": 255843149.0, + "step": 6588 + }, + { + "epoch": 0.8381885256328712, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.378392219543457, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8749834299087524, + "num_tokens": 255878614.0, + "step": 6589 + }, + { + "epoch": 0.8383157359114617, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.320930480957031, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8778818845748901, + "num_tokens": 255917779.0, + "step": 6590 + }, + { + "epoch": 0.8384429461900521, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.410058975219727, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8778700828552246, + "num_tokens": 255953975.0, + "step": 6591 + }, + { + "epoch": 0.8385701564686426, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.378670692443848, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8845845460891724, + "num_tokens": 255993904.0, + "step": 6592 + }, + { + "epoch": 0.8386973667472332, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.399772644042969, + "learning_rate": 1e-06, + "loss": 0.3597, + "mean_token_accuracy": 0.8799427151679993, + "num_tokens": 256028679.0, + "step": 6593 + }, + { + "epoch": 0.8388245770258237, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.483938217163086, + "learning_rate": 1e-06, + "loss": 0.3849, + "mean_token_accuracy": 0.8739052414894104, + "num_tokens": 256058445.0, + "step": 6594 + }, + { + "epoch": 0.8389517873044142, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.289356231689453, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8851717114448547, + "num_tokens": 256093963.0, + "step": 6595 + }, + { + "epoch": 0.8390789975830047, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.361559867858887, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8729522228240967, + "num_tokens": 256129027.0, + "step": 6596 + }, + { + "epoch": 0.8392062078615952, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.42625617980957, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8778656125068665, + "num_tokens": 256164178.0, + "step": 6597 + }, + { + "epoch": 0.8393334181401857, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.358503341674805, + "learning_rate": 1e-06, + "loss": 0.4103, + "mean_token_accuracy": 0.8645761013031006, + "num_tokens": 256202867.0, + "step": 6598 + }, + { + "epoch": 0.8394606284187762, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.24275016784668, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8728039264678955, + "num_tokens": 256247383.0, + "step": 6599 + }, + { + "epoch": 0.8395878386973668, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.43921947479248, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8786307573318481, + "num_tokens": 256283226.0, + "step": 6600 + }, + { + "epoch": 0.8397150489759573, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.37420654296875, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.878295361995697, + "num_tokens": 256320759.0, + "step": 6601 + }, + { + "epoch": 0.8398422592545478, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.250649452209473, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8846613168716431, + "num_tokens": 256363473.0, + "step": 6602 + }, + { + "epoch": 0.8399694695331382, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.401223182678223, + "learning_rate": 1e-06, + "loss": 0.4006, + "mean_token_accuracy": 0.8641141057014465, + "num_tokens": 256403436.0, + "step": 6603 + }, + { + "epoch": 0.8400966798117288, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.405760765075684, + "learning_rate": 1e-06, + "loss": 0.4059, + "mean_token_accuracy": 0.8658322691917419, + "num_tokens": 256440927.0, + "step": 6604 + }, + { + "epoch": 0.8402238900903193, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.345560073852539, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8759194612503052, + "num_tokens": 256480590.0, + "step": 6605 + }, + { + "epoch": 0.8403511003689098, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.312788963317871, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8761099576950073, + "num_tokens": 256520219.0, + "step": 6606 + }, + { + "epoch": 0.8404783106475003, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.342752456665039, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8803187608718872, + "num_tokens": 256556263.0, + "step": 6607 + }, + { + "epoch": 0.8406055209260909, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.297804832458496, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8761532306671143, + "num_tokens": 256598112.0, + "step": 6608 + }, + { + "epoch": 0.8407327312046813, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.268145561218262, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8803003430366516, + "num_tokens": 256636535.0, + "step": 6609 + }, + { + "epoch": 0.8408599414832718, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.338083267211914, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8847934007644653, + "num_tokens": 256673895.0, + "step": 6610 + }, + { + "epoch": 0.8409871517618623, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.249839782714844, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8732181787490845, + "num_tokens": 256720757.0, + "step": 6611 + }, + { + "epoch": 0.8411143620404529, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.448909759521484, + "learning_rate": 1e-06, + "loss": 0.3729, + "mean_token_accuracy": 0.8759243488311768, + "num_tokens": 256755809.0, + "step": 6612 + }, + { + "epoch": 0.8412415723190434, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.471094131469727, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.870267927646637, + "num_tokens": 256793431.0, + "step": 6613 + }, + { + "epoch": 0.8413687825976339, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.329216957092285, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8776772618293762, + "num_tokens": 256831681.0, + "step": 6614 + }, + { + "epoch": 0.8414959928762245, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.320941925048828, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8913826942443848, + "num_tokens": 256874595.0, + "step": 6615 + }, + { + "epoch": 0.8416232031548149, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.330083847045898, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8813099265098572, + "num_tokens": 256911237.0, + "step": 6616 + }, + { + "epoch": 0.8417504134334054, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.421310424804688, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8736532330513, + "num_tokens": 256943587.0, + "step": 6617 + }, + { + "epoch": 0.8418776237119959, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.387415885925293, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8809423446655273, + "num_tokens": 256977647.0, + "step": 6618 + }, + { + "epoch": 0.8420048339905865, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.332414627075195, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8836067914962769, + "num_tokens": 257016374.0, + "step": 6619 + }, + { + "epoch": 0.842132044269177, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.404546737670898, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8876860737800598, + "num_tokens": 257051182.0, + "step": 6620 + }, + { + "epoch": 0.8422592545477675, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0040740966796875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.305957794189453, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8871153593063354, + "num_tokens": 257096509.0, + "step": 6621 + }, + { + "epoch": 0.8423864648263579, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0279159545898438e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.347298622131348, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8887661695480347, + "num_tokens": 257132189.0, + "step": 6622 + }, + { + "epoch": 0.8425136751049485, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0279159545898438e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.3416109085083, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8761613368988037, + "num_tokens": 257171266.0, + "step": 6623 + }, + { + "epoch": 0.842640885383539, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0279159545898438e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.347441673278809, + "learning_rate": 1e-06, + "loss": 0.3986, + "mean_token_accuracy": 0.8680166602134705, + "num_tokens": 257212326.0, + "step": 6624 + }, + { + "epoch": 0.8427680956621295, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0279159545898438e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.341750144958496, + "learning_rate": 1e-06, + "loss": 0.4077, + "mean_token_accuracy": 0.8634054064750671, + "num_tokens": 257259063.0, + "step": 6625 + }, + { + "epoch": 0.84289530594072, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0279159545898438e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.419599533081055, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8783355951309204, + "num_tokens": 257298165.0, + "step": 6626 + }, + { + "epoch": 0.8430225162193106, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0279159545898438e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.415000915527344, + "learning_rate": 1e-06, + "loss": 0.4163, + "mean_token_accuracy": 0.8652278184890747, + "num_tokens": 257335603.0, + "step": 6627 + }, + { + "epoch": 0.843149726497901, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0279159545898438e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.329511642456055, + "learning_rate": 1e-06, + "loss": 0.3925, + "mean_token_accuracy": 0.8657160997390747, + "num_tokens": 257381074.0, + "step": 6628 + }, + { + "epoch": 0.8432769367764915, + "ewc_loss": 0.0120849609375, + "ewc_loss_diag": 3.0279159545898438e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.418129920959473, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8776206970214844, + "num_tokens": 257418262.0, + "step": 6629 + }, + { + "epoch": 0.843404147055082, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0279159545898438e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.321081161499023, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8836153745651245, + "num_tokens": 257457609.0, + "step": 6630 + }, + { + "epoch": 0.8435313573336726, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0279159545898438e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.536687850952148, + "learning_rate": 1e-06, + "loss": 0.3953, + "mean_token_accuracy": 0.8651864528656006, + "num_tokens": 257492248.0, + "step": 6631 + }, + { + "epoch": 0.8436585676122631, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.3602294921875, + "learning_rate": 1e-06, + "loss": 0.3917, + "mean_token_accuracy": 0.8656735420227051, + "num_tokens": 257533726.0, + "step": 6632 + }, + { + "epoch": 0.8437857778908536, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.265826225280762, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8786550760269165, + "num_tokens": 257576019.0, + "step": 6633 + }, + { + "epoch": 0.843912988169444, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.344099044799805, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.896287202835083, + "num_tokens": 257610518.0, + "step": 6634 + }, + { + "epoch": 0.8440401984480346, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.324674606323242, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8714241981506348, + "num_tokens": 257652269.0, + "step": 6635 + }, + { + "epoch": 0.8441674087266251, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.410969734191895, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8820700645446777, + "num_tokens": 257689085.0, + "step": 6636 + }, + { + "epoch": 0.8442946190052156, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.308015823364258, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8738365769386292, + "num_tokens": 257728737.0, + "step": 6637 + }, + { + "epoch": 0.8444218292838062, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.24325180053711, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8742126226425171, + "num_tokens": 257771068.0, + "step": 6638 + }, + { + "epoch": 0.8445490395623967, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.352502822875977, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8922328948974609, + "num_tokens": 257806858.0, + "step": 6639 + }, + { + "epoch": 0.8446762498409871, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.34007453918457, + "learning_rate": 1e-06, + "loss": 0.3843, + "mean_token_accuracy": 0.8699532747268677, + "num_tokens": 257850483.0, + "step": 6640 + }, + { + "epoch": 0.8448034601195776, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.335660934448242, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8793811798095703, + "num_tokens": 257890762.0, + "step": 6641 + }, + { + "epoch": 0.8449306703981682, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.405228614807129, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8896295428276062, + "num_tokens": 257923296.0, + "step": 6642 + }, + { + "epoch": 0.8450578806767587, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.356278419494629, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8777297139167786, + "num_tokens": 257962898.0, + "step": 6643 + }, + { + "epoch": 0.8451850909553492, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.330923080444336, + "learning_rate": 1e-06, + "loss": 0.3924, + "mean_token_accuracy": 0.8695406913757324, + "num_tokens": 258006268.0, + "step": 6644 + }, + { + "epoch": 0.8453123012339397, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.391180992126465, + "learning_rate": 1e-06, + "loss": 0.3936, + "mean_token_accuracy": 0.8675936460494995, + "num_tokens": 258046682.0, + "step": 6645 + }, + { + "epoch": 0.8454395115125302, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.399540901184082, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.886754035949707, + "num_tokens": 258082216.0, + "step": 6646 + }, + { + "epoch": 0.8455667217911207, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.367789268493652, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8720674514770508, + "num_tokens": 258118625.0, + "step": 6647 + }, + { + "epoch": 0.8456939320697112, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.359752655029297, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8756459355354309, + "num_tokens": 258152884.0, + "step": 6648 + }, + { + "epoch": 0.8458211423483017, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.381978988647461, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8732569217681885, + "num_tokens": 258190900.0, + "step": 6649 + }, + { + "epoch": 0.8459483526268923, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.499317169189453, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.8700512647628784, + "num_tokens": 258223132.0, + "step": 6650 + }, + { + "epoch": 0.8460755629054828, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.340178489685059, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8715196847915649, + "num_tokens": 258262215.0, + "step": 6651 + }, + { + "epoch": 0.8462027731840732, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.239580154418945, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8854938745498657, + "num_tokens": 258309033.0, + "step": 6652 + }, + { + "epoch": 0.8463299834626637, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.415343284606934, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8849027156829834, + "num_tokens": 258341807.0, + "step": 6653 + }, + { + "epoch": 0.8464571937412543, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.327242851257324, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.882995069026947, + "num_tokens": 258380825.0, + "step": 6654 + }, + { + "epoch": 0.8465844040198448, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0517578125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.484247207641602, + "learning_rate": 1e-06, + "loss": 0.3854, + "mean_token_accuracy": 0.8731178045272827, + "num_tokens": 258414565.0, + "step": 6655 + }, + { + "epoch": 0.8467116142984353, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 12.100610733032227, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8904473781585693, + "num_tokens": 258447459.0, + "step": 6656 + }, + { + "epoch": 0.8468388245770259, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 8.801519393920898, + "learning_rate": 1e-06, + "loss": 0.3959, + "mean_token_accuracy": 0.8663362264633179, + "num_tokens": 258492715.0, + "step": 6657 + }, + { + "epoch": 0.8469660348556163, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 7.9212446212768555, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8759291768074036, + "num_tokens": 258535674.0, + "step": 6658 + }, + { + "epoch": 0.8470932451342068, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.678153038024902, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8756587505340576, + "num_tokens": 258575947.0, + "step": 6659 + }, + { + "epoch": 0.8472204554127973, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.437880516052246, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8793851137161255, + "num_tokens": 258610592.0, + "step": 6660 + }, + { + "epoch": 0.8473476656913879, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.429834365844727, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8751978874206543, + "num_tokens": 258653372.0, + "step": 6661 + }, + { + "epoch": 0.8474748759699784, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.350065231323242, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8821781873703003, + "num_tokens": 258696375.0, + "step": 6662 + }, + { + "epoch": 0.8476020862485689, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.336955070495605, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8828706741333008, + "num_tokens": 258737226.0, + "step": 6663 + }, + { + "epoch": 0.8477292965271594, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.409740447998047, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8788478374481201, + "num_tokens": 258778061.0, + "step": 6664 + }, + { + "epoch": 0.8478565068057499, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.343707084655762, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8879964351654053, + "num_tokens": 258818422.0, + "step": 6665 + }, + { + "epoch": 0.8479837170843404, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.48264217376709, + "learning_rate": 1e-06, + "loss": 0.4017, + "mean_token_accuracy": 0.8657843470573425, + "num_tokens": 258853538.0, + "step": 6666 + }, + { + "epoch": 0.8481109273629309, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.399786949157715, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8897028565406799, + "num_tokens": 258891552.0, + "step": 6667 + }, + { + "epoch": 0.8482381376415215, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.470385551452637, + "learning_rate": 1e-06, + "loss": 0.4029, + "mean_token_accuracy": 0.8658568859100342, + "num_tokens": 258927842.0, + "step": 6668 + }, + { + "epoch": 0.848365347920112, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.355583190917969, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.887779176235199, + "num_tokens": 258966606.0, + "step": 6669 + }, + { + "epoch": 0.8484925581987025, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.394994735717773, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8869640231132507, + "num_tokens": 259001075.0, + "step": 6670 + }, + { + "epoch": 0.8486197684772929, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.372213363647461, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8902145624160767, + "num_tokens": 259039478.0, + "step": 6671 + }, + { + "epoch": 0.8487469787558835, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.330695152282715, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8755051493644714, + "num_tokens": 259085807.0, + "step": 6672 + }, + { + "epoch": 0.848874189034474, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.338438987731934, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8900482654571533, + "num_tokens": 259125439.0, + "step": 6673 + }, + { + "epoch": 0.8490013993130645, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.425763130187988, + "learning_rate": 1e-06, + "loss": 0.3817, + "mean_token_accuracy": 0.8702162504196167, + "num_tokens": 259162970.0, + "step": 6674 + }, + { + "epoch": 0.849128609591655, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.54654312133789, + "learning_rate": 1e-06, + "loss": 0.4604, + "mean_token_accuracy": 0.8471395969390869, + "num_tokens": 259199052.0, + "step": 6675 + }, + { + "epoch": 0.8492558198702456, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.517550468444824, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8748886585235596, + "num_tokens": 259227866.0, + "step": 6676 + }, + { + "epoch": 0.849383030148836, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.301814079284668, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8814743757247925, + "num_tokens": 259269745.0, + "step": 6677 + }, + { + "epoch": 0.8495102404274265, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.413599014282227, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8797651529312134, + "num_tokens": 259305225.0, + "step": 6678 + }, + { + "epoch": 0.849637450706017, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.353557586669922, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8735200762748718, + "num_tokens": 259345088.0, + "step": 6679 + }, + { + "epoch": 0.8497646609846076, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.2409086227417, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8760520219802856, + "num_tokens": 259393736.0, + "step": 6680 + }, + { + "epoch": 0.8498918712631981, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.41963005065918, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.878090500831604, + "num_tokens": 259429090.0, + "step": 6681 + }, + { + "epoch": 0.8500190815417886, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.523162841796875, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8703525066375732, + "num_tokens": 259462362.0, + "step": 6682 + }, + { + "epoch": 0.850146291820379, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.261520385742188, + "learning_rate": 1e-06, + "loss": 0.4081, + "mean_token_accuracy": 0.8607885837554932, + "num_tokens": 259506477.0, + "step": 6683 + }, + { + "epoch": 0.8502735020989696, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.444391250610352, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8837200403213501, + "num_tokens": 259537896.0, + "step": 6684 + }, + { + "epoch": 0.8504007123775601, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.36205005645752, + "learning_rate": 1e-06, + "loss": 0.3816, + "mean_token_accuracy": 0.8693393468856812, + "num_tokens": 259574736.0, + "step": 6685 + }, + { + "epoch": 0.8505279226561506, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.353797912597656, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8870282769203186, + "num_tokens": 259613243.0, + "step": 6686 + }, + { + "epoch": 0.8506551329347412, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.374000549316406, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8823492527008057, + "num_tokens": 259652031.0, + "step": 6687 + }, + { + "epoch": 0.8507823432133317, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.383624076843262, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8824145793914795, + "num_tokens": 259686326.0, + "step": 6688 + }, + { + "epoch": 0.8509095534919221, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.356711387634277, + "learning_rate": 1e-06, + "loss": 0.3733, + "mean_token_accuracy": 0.8729111552238464, + "num_tokens": 259727927.0, + "step": 6689 + }, + { + "epoch": 0.8510367637705126, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.42505168914795, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8837180733680725, + "num_tokens": 259760854.0, + "step": 6690 + }, + { + "epoch": 0.8511639740491032, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.075599670410156e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.349544525146484, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8833009004592896, + "num_tokens": 259802254.0, + "step": 6691 + }, + { + "epoch": 0.8512911843276937, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.433234214782715, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8787644505500793, + "num_tokens": 259833771.0, + "step": 6692 + }, + { + "epoch": 0.8514183946062842, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.422141075134277, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8854626417160034, + "num_tokens": 259872505.0, + "step": 6693 + }, + { + "epoch": 0.8515456048848747, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.309196472167969, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8763323426246643, + "num_tokens": 259909529.0, + "step": 6694 + }, + { + "epoch": 0.8516728151634652, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.498068809509277, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8683159351348877, + "num_tokens": 259943511.0, + "step": 6695 + }, + { + "epoch": 0.8518000254420557, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.327041625976562, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.8756251931190491, + "num_tokens": 259989322.0, + "step": 6696 + }, + { + "epoch": 0.8519272357206462, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.452547073364258, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8836367130279541, + "num_tokens": 260025242.0, + "step": 6697 + }, + { + "epoch": 0.8520544459992367, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.376370429992676, + "learning_rate": 1e-06, + "loss": 0.3905, + "mean_token_accuracy": 0.8664579391479492, + "num_tokens": 260063059.0, + "step": 6698 + }, + { + "epoch": 0.8521816562778273, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.288549423217773, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8861173391342163, + "num_tokens": 260104227.0, + "step": 6699 + }, + { + "epoch": 0.8523088665564178, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.34908676147461, + "learning_rate": 1e-06, + "loss": 0.385, + "mean_token_accuracy": 0.867638885974884, + "num_tokens": 260147268.0, + "step": 6700 + }, + { + "epoch": 0.8524360768350082, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.42002010345459, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8788890838623047, + "num_tokens": 260187127.0, + "step": 6701 + }, + { + "epoch": 0.8525632871135987, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.335539817810059, + "learning_rate": 1e-06, + "loss": 0.3724, + "mean_token_accuracy": 0.874278724193573, + "num_tokens": 260230432.0, + "step": 6702 + }, + { + "epoch": 0.8526904973921893, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.377053260803223, + "learning_rate": 1e-06, + "loss": 0.3818, + "mean_token_accuracy": 0.8718152642250061, + "num_tokens": 260268295.0, + "step": 6703 + }, + { + "epoch": 0.8528177076707798, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.331415176391602, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8683228492736816, + "num_tokens": 260308979.0, + "step": 6704 + }, + { + "epoch": 0.8529449179493703, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.347602844238281, + "learning_rate": 1e-06, + "loss": 0.4185, + "mean_token_accuracy": 0.861107587814331, + "num_tokens": 260351553.0, + "step": 6705 + }, + { + "epoch": 0.8530721282279609, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.335083961486816, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8895053863525391, + "num_tokens": 260393882.0, + "step": 6706 + }, + { + "epoch": 0.8531993385065513, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.311253547668457, + "learning_rate": 1e-06, + "loss": 0.3779, + "mean_token_accuracy": 0.8753417134284973, + "num_tokens": 260434222.0, + "step": 6707 + }, + { + "epoch": 0.8533265487851418, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.290119171142578, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8859326839447021, + "num_tokens": 260473575.0, + "step": 6708 + }, + { + "epoch": 0.8534537590637323, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.358048439025879, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8867435455322266, + "num_tokens": 260511502.0, + "step": 6709 + }, + { + "epoch": 0.8535809693423229, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.327861785888672, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8791863918304443, + "num_tokens": 260550947.0, + "step": 6710 + }, + { + "epoch": 0.8537081796209134, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.328149795532227, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8825487494468689, + "num_tokens": 260591121.0, + "step": 6711 + }, + { + "epoch": 0.8538353898995039, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.397363662719727, + "learning_rate": 1e-06, + "loss": 0.3882, + "mean_token_accuracy": 0.8710322380065918, + "num_tokens": 260628831.0, + "step": 6712 + }, + { + "epoch": 0.8539626001780944, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.34990406036377, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8743373155593872, + "num_tokens": 260672565.0, + "step": 6713 + }, + { + "epoch": 0.8540898104566849, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.36860179901123, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8815127611160278, + "num_tokens": 260710882.0, + "step": 6714 + }, + { + "epoch": 0.8542170207352754, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.416595458984375, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.8788638114929199, + "num_tokens": 260743281.0, + "step": 6715 + }, + { + "epoch": 0.8543442310138659, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.283571243286133, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8777885437011719, + "num_tokens": 260784922.0, + "step": 6716 + }, + { + "epoch": 0.8544714412924564, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.392242431640625, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8738556504249573, + "num_tokens": 260821423.0, + "step": 6717 + }, + { + "epoch": 0.854598651571047, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.325875282287598, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8849013447761536, + "num_tokens": 260858573.0, + "step": 6718 + }, + { + "epoch": 0.8547258618496375, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.360236167907715, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8893747925758362, + "num_tokens": 260898854.0, + "step": 6719 + }, + { + "epoch": 0.8548530721282279, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.333150863647461, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8859129548072815, + "num_tokens": 260938759.0, + "step": 6720 + }, + { + "epoch": 0.8549802824068184, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.290712356567383, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8846700191497803, + "num_tokens": 260977080.0, + "step": 6721 + }, + { + "epoch": 0.855107492685409, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.367225646972656, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8714495897293091, + "num_tokens": 261015591.0, + "step": 6722 + }, + { + "epoch": 0.8552347029639995, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.346272468566895, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8789080381393433, + "num_tokens": 261054229.0, + "step": 6723 + }, + { + "epoch": 0.85536191324259, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.493894577026367, + "learning_rate": 1e-06, + "loss": 0.4102, + "mean_token_accuracy": 0.8621377944946289, + "num_tokens": 261092559.0, + "step": 6724 + }, + { + "epoch": 0.8554891235211806, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.360708236694336, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8812417387962341, + "num_tokens": 261130038.0, + "step": 6725 + }, + { + "epoch": 0.855616333799771, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.389850616455078, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8816711902618408, + "num_tokens": 261165528.0, + "step": 6726 + }, + { + "epoch": 0.8557435440783615, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.336894035339355, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8840373754501343, + "num_tokens": 261203138.0, + "step": 6727 + }, + { + "epoch": 0.855870754356952, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.419625282287598, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8759700059890747, + "num_tokens": 261240216.0, + "step": 6728 + }, + { + "epoch": 0.8559979646355426, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.343693733215332, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.879652202129364, + "num_tokens": 261283083.0, + "step": 6729 + }, + { + "epoch": 0.8561251749141331, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.306753158569336, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.9005906581878662, + "num_tokens": 261324629.0, + "step": 6730 + }, + { + "epoch": 0.8562523851927236, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.37806510925293, + "learning_rate": 1e-06, + "loss": 0.3849, + "mean_token_accuracy": 0.8732348680496216, + "num_tokens": 261362253.0, + "step": 6731 + }, + { + "epoch": 0.856379595471314, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.459322929382324, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8758444786071777, + "num_tokens": 261395269.0, + "step": 6732 + }, + { + "epoch": 0.8565068057499046, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.285341262817383, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.881862998008728, + "num_tokens": 261439648.0, + "step": 6733 + }, + { + "epoch": 0.8566340160284951, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.42397689819336, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8789429664611816, + "num_tokens": 261476879.0, + "step": 6734 + }, + { + "epoch": 0.8567612263070856, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.309829711914062, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8783882260322571, + "num_tokens": 261517949.0, + "step": 6735 + }, + { + "epoch": 0.8568884365856761, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.332008361816406, + "learning_rate": 1e-06, + "loss": 0.3654, + "mean_token_accuracy": 0.8750419616699219, + "num_tokens": 261557073.0, + "step": 6736 + }, + { + "epoch": 0.8570156468642667, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.365400314331055, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8828369975090027, + "num_tokens": 261594835.0, + "step": 6737 + }, + { + "epoch": 0.8571428571428571, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.329319953918457, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8792447447776794, + "num_tokens": 261634969.0, + "step": 6738 + }, + { + "epoch": 0.8572700674214476, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.403956413269043, + "learning_rate": 1e-06, + "loss": 0.3951, + "mean_token_accuracy": 0.8648598194122314, + "num_tokens": 261674535.0, + "step": 6739 + }, + { + "epoch": 0.8573972777000382, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.374344825744629, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8751575946807861, + "num_tokens": 261714152.0, + "step": 6740 + }, + { + "epoch": 0.8575244879786287, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.309697151184082, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8727514147758484, + "num_tokens": 261758662.0, + "step": 6741 + }, + { + "epoch": 0.8576516982572192, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.434885025024414, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8738398551940918, + "num_tokens": 261791742.0, + "step": 6742 + }, + { + "epoch": 0.8577789085358097, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.287039756774902, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8800190687179565, + "num_tokens": 261829993.0, + "step": 6743 + }, + { + "epoch": 0.8579061188144002, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.34029769897461, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8917704820632935, + "num_tokens": 261867310.0, + "step": 6744 + }, + { + "epoch": 0.8580333290929907, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.353251457214355, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8837700486183167, + "num_tokens": 261901683.0, + "step": 6745 + }, + { + "epoch": 0.8581605393715812, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.466914176940918, + "learning_rate": 1e-06, + "loss": 0.3834, + "mean_token_accuracy": 0.8696645498275757, + "num_tokens": 261934874.0, + "step": 6746 + }, + { + "epoch": 0.8582877496501717, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.327285766601562, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8823478817939758, + "num_tokens": 261975477.0, + "step": 6747 + }, + { + "epoch": 0.8584149599287623, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.305133819580078, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8885976076126099, + "num_tokens": 262014353.0, + "step": 6748 + }, + { + "epoch": 0.8585421702073528, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.386812210083008, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8843551874160767, + "num_tokens": 262051048.0, + "step": 6749 + }, + { + "epoch": 0.8586693804859432, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.398159980773926, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8836797475814819, + "num_tokens": 262085588.0, + "step": 6750 + }, + { + "epoch": 0.8587965907645337, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.342535018920898, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8851734399795532, + "num_tokens": 262121109.0, + "step": 6751 + }, + { + "epoch": 0.8589238010431243, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.366994857788086, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8864143490791321, + "num_tokens": 262154892.0, + "step": 6752 + }, + { + "epoch": 0.8590510113217148, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.432994842529297, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8751964569091797, + "num_tokens": 262187451.0, + "step": 6753 + }, + { + "epoch": 0.8591782216003053, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.0994415283203125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.400465965270996, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8712518215179443, + "num_tokens": 262225228.0, + "step": 6754 + }, + { + "epoch": 0.8593054318788959, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.404239654541016, + "learning_rate": 1e-06, + "loss": 0.4201, + "mean_token_accuracy": 0.8583803772926331, + "num_tokens": 262265091.0, + "step": 6755 + }, + { + "epoch": 0.8594326421574863, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.510436058044434, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8787476420402527, + "num_tokens": 262300816.0, + "step": 6756 + }, + { + "epoch": 0.8595598524360768, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.397374153137207, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.870709240436554, + "num_tokens": 262335836.0, + "step": 6757 + }, + { + "epoch": 0.8596870627146673, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.306178092956543, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8838136196136475, + "num_tokens": 262378457.0, + "step": 6758 + }, + { + "epoch": 0.8598142729932579, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.374908447265625, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8924708366394043, + "num_tokens": 262418808.0, + "step": 6759 + }, + { + "epoch": 0.8599414832718484, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.32840633392334, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8844122290611267, + "num_tokens": 262461063.0, + "step": 6760 + }, + { + "epoch": 0.8600686935504389, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.53354263305664, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8747643828392029, + "num_tokens": 262491166.0, + "step": 6761 + }, + { + "epoch": 0.8601959038290294, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.422318458557129, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8778595924377441, + "num_tokens": 262523199.0, + "step": 6762 + }, + { + "epoch": 0.8603231141076199, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.438695907592773, + "learning_rate": 1e-06, + "loss": 0.4125, + "mean_token_accuracy": 0.8644880652427673, + "num_tokens": 262562640.0, + "step": 6763 + }, + { + "epoch": 0.8604503243862104, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.37013053894043, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8849337100982666, + "num_tokens": 262596287.0, + "step": 6764 + }, + { + "epoch": 0.8605775346648009, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.380049705505371, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8808073997497559, + "num_tokens": 262633118.0, + "step": 6765 + }, + { + "epoch": 0.8607047449433914, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.461133003234863, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8865439891815186, + "num_tokens": 262665924.0, + "step": 6766 + }, + { + "epoch": 0.860831955221982, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.408108711242676, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8816119432449341, + "num_tokens": 262700842.0, + "step": 6767 + }, + { + "epoch": 0.8609591655005725, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.302720069885254, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8933838605880737, + "num_tokens": 262739882.0, + "step": 6768 + }, + { + "epoch": 0.8610863757791629, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.331432342529297, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8808388113975525, + "num_tokens": 262775343.0, + "step": 6769 + }, + { + "epoch": 0.8612135860577534, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.423906326293945, + "learning_rate": 1e-06, + "loss": 0.4273, + "mean_token_accuracy": 0.8597612380981445, + "num_tokens": 262818586.0, + "step": 6770 + }, + { + "epoch": 0.861340796336344, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.393009185791016, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8776538968086243, + "num_tokens": 262856178.0, + "step": 6771 + }, + { + "epoch": 0.8614680066149345, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.388434410095215, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.8705080151557922, + "num_tokens": 262897515.0, + "step": 6772 + }, + { + "epoch": 0.861595216893525, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.418770790100098, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8793103694915771, + "num_tokens": 262933944.0, + "step": 6773 + }, + { + "epoch": 0.8617224271721156, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.280278205871582, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8815293312072754, + "num_tokens": 262978786.0, + "step": 6774 + }, + { + "epoch": 0.861849637450706, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.423792839050293, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.8684924840927124, + "num_tokens": 263020932.0, + "step": 6775 + }, + { + "epoch": 0.8619768477292965, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.46333122253418, + "learning_rate": 1e-06, + "loss": 0.3904, + "mean_token_accuracy": 0.8677701950073242, + "num_tokens": 263055266.0, + "step": 6776 + }, + { + "epoch": 0.862104058007887, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.409879684448242, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8808572888374329, + "num_tokens": 263091826.0, + "step": 6777 + }, + { + "epoch": 0.8622312682864776, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.398195266723633, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.875502347946167, + "num_tokens": 263130377.0, + "step": 6778 + }, + { + "epoch": 0.8623584785650681, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.399896621704102, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.881803572177887, + "num_tokens": 263164956.0, + "step": 6779 + }, + { + "epoch": 0.8624856888436586, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.353616714477539, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.886426568031311, + "num_tokens": 263202695.0, + "step": 6780 + }, + { + "epoch": 0.862612899122249, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.422629356384277, + "learning_rate": 1e-06, + "loss": 0.3946, + "mean_token_accuracy": 0.8649088144302368, + "num_tokens": 263238965.0, + "step": 6781 + }, + { + "epoch": 0.8627401094008396, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.346814155578613, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8775790929794312, + "num_tokens": 263279065.0, + "step": 6782 + }, + { + "epoch": 0.8628673196794301, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.35527229309082, + "learning_rate": 1e-06, + "loss": 0.3784, + "mean_token_accuracy": 0.8714644908905029, + "num_tokens": 263318644.0, + "step": 6783 + }, + { + "epoch": 0.8629945299580206, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.41454792022705, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.875773549079895, + "num_tokens": 263358182.0, + "step": 6784 + }, + { + "epoch": 0.8631217402366111, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.37083911895752, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8809010982513428, + "num_tokens": 263394385.0, + "step": 6785 + }, + { + "epoch": 0.8632489505152017, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.367152214050293, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8769418597221375, + "num_tokens": 263431431.0, + "step": 6786 + }, + { + "epoch": 0.8633761607937921, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.377601623535156, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8833485245704651, + "num_tokens": 263467789.0, + "step": 6787 + }, + { + "epoch": 0.8635033710723826, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.333795547485352, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.8736522793769836, + "num_tokens": 263507305.0, + "step": 6788 + }, + { + "epoch": 0.8636305813509731, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.300032615661621, + "learning_rate": 1e-06, + "loss": 0.3962, + "mean_token_accuracy": 0.8671069145202637, + "num_tokens": 263554154.0, + "step": 6789 + }, + { + "epoch": 0.8637577916295637, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.382883071899414, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8729356527328491, + "num_tokens": 263593235.0, + "step": 6790 + }, + { + "epoch": 0.8638850019081542, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.350722312927246, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8903658986091614, + "num_tokens": 263634662.0, + "step": 6791 + }, + { + "epoch": 0.8640122121867447, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.267805099487305, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8918024301528931, + "num_tokens": 263678075.0, + "step": 6792 + }, + { + "epoch": 0.8641394224653351, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.376890182495117, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.881118893623352, + "num_tokens": 263715818.0, + "step": 6793 + }, + { + "epoch": 0.8642666327439257, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.366145133972168, + "learning_rate": 1e-06, + "loss": 0.3576, + "mean_token_accuracy": 0.878251314163208, + "num_tokens": 263752062.0, + "step": 6794 + }, + { + "epoch": 0.8643938430225162, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.36139965057373, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.876858651638031, + "num_tokens": 263789575.0, + "step": 6795 + }, + { + "epoch": 0.8645210533011067, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.3643217086792, + "learning_rate": 1e-06, + "loss": 0.3894, + "mean_token_accuracy": 0.8697444200515747, + "num_tokens": 263833391.0, + "step": 6796 + }, + { + "epoch": 0.8646482635796973, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.359200477600098, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8872212171554565, + "num_tokens": 263868599.0, + "step": 6797 + }, + { + "epoch": 0.8647754738582878, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.419012069702148, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8807799816131592, + "num_tokens": 263904517.0, + "step": 6798 + }, + { + "epoch": 0.8649026841368782, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.30502700805664, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8726765513420105, + "num_tokens": 263942858.0, + "step": 6799 + }, + { + "epoch": 0.8650298944154687, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.313045501708984, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8780970573425293, + "num_tokens": 263984166.0, + "step": 6800 + }, + { + "epoch": 0.8651571046940593, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 9.002606391906738, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8794964551925659, + "num_tokens": 264016405.0, + "step": 6801 + }, + { + "epoch": 0.8652843149726498, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.330469131469727, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.884453296661377, + "num_tokens": 264055863.0, + "step": 6802 + }, + { + "epoch": 0.8654115252512403, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.269648551940918, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.878411591053009, + "num_tokens": 264095369.0, + "step": 6803 + }, + { + "epoch": 0.8655387355298308, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.353450775146484, + "learning_rate": 1e-06, + "loss": 0.3834, + "mean_token_accuracy": 0.8706773519515991, + "num_tokens": 264142188.0, + "step": 6804 + }, + { + "epoch": 0.8656659458084213, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.302227973937988, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.885603129863739, + "num_tokens": 264180956.0, + "step": 6805 + }, + { + "epoch": 0.8657931560870118, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.436612129211426, + "learning_rate": 1e-06, + "loss": 0.3963, + "mean_token_accuracy": 0.8694336414337158, + "num_tokens": 264225464.0, + "step": 6806 + }, + { + "epoch": 0.8659203663656023, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.351006507873535, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.881781280040741, + "num_tokens": 264264182.0, + "step": 6807 + }, + { + "epoch": 0.8660475766441929, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.265678405761719, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8928871750831604, + "num_tokens": 264304683.0, + "step": 6808 + }, + { + "epoch": 0.8661747869227834, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.447677612304688, + "learning_rate": 1e-06, + "loss": 0.4229, + "mean_token_accuracy": 0.860074520111084, + "num_tokens": 264345792.0, + "step": 6809 + }, + { + "epoch": 0.8663019972013739, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.405375480651855, + "learning_rate": 1e-06, + "loss": 0.3888, + "mean_token_accuracy": 0.8707563281059265, + "num_tokens": 264383362.0, + "step": 6810 + }, + { + "epoch": 0.8664292074799644, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.357070922851562, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8807174563407898, + "num_tokens": 264422262.0, + "step": 6811 + }, + { + "epoch": 0.8665564177585549, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.123283386230469e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.298406600952148, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8966481685638428, + "num_tokens": 264464374.0, + "step": 6812 + }, + { + "epoch": 0.8666836280371454, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.384981155395508, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8794645071029663, + "num_tokens": 264501241.0, + "step": 6813 + }, + { + "epoch": 0.8668108383157359, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.38536548614502, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8807283639907837, + "num_tokens": 264541601.0, + "step": 6814 + }, + { + "epoch": 0.8669380485943264, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.331038475036621, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8822078108787537, + "num_tokens": 264581201.0, + "step": 6815 + }, + { + "epoch": 0.867065258872917, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.435495376586914, + "learning_rate": 1e-06, + "loss": 0.3861, + "mean_token_accuracy": 0.8709712028503418, + "num_tokens": 264622459.0, + "step": 6816 + }, + { + "epoch": 0.8671924691515075, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.416505813598633, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8773841261863708, + "num_tokens": 264659881.0, + "step": 6817 + }, + { + "epoch": 0.8673196794300979, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.3988037109375, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8935611844062805, + "num_tokens": 264691690.0, + "step": 6818 + }, + { + "epoch": 0.8674468897086884, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.557547569274902, + "learning_rate": 1e-06, + "loss": 0.4025, + "mean_token_accuracy": 0.8610612154006958, + "num_tokens": 264723200.0, + "step": 6819 + }, + { + "epoch": 0.867574099987279, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.334613800048828, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.881729245185852, + "num_tokens": 264763083.0, + "step": 6820 + }, + { + "epoch": 0.8677013102658695, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.370022773742676, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8807324767112732, + "num_tokens": 264799025.0, + "step": 6821 + }, + { + "epoch": 0.86782852054446, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.422967910766602, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8808145523071289, + "num_tokens": 264834521.0, + "step": 6822 + }, + { + "epoch": 0.8679557308230506, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.403040885925293, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8742420673370361, + "num_tokens": 264870646.0, + "step": 6823 + }, + { + "epoch": 0.868082941101641, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.349095344543457, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8902276158332825, + "num_tokens": 264913555.0, + "step": 6824 + }, + { + "epoch": 0.8682101513802315, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.314934730529785, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.878691554069519, + "num_tokens": 264954988.0, + "step": 6825 + }, + { + "epoch": 0.868337361658822, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.345252990722656, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8760453462600708, + "num_tokens": 264996866.0, + "step": 6826 + }, + { + "epoch": 0.8684645719374126, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.394518852233887, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8766399621963501, + "num_tokens": 265036234.0, + "step": 6827 + }, + { + "epoch": 0.8685917822160031, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.398598670959473, + "learning_rate": 1e-06, + "loss": 0.4092, + "mean_token_accuracy": 0.8643177151679993, + "num_tokens": 265078448.0, + "step": 6828 + }, + { + "epoch": 0.8687189924945936, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.305047988891602, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8858048319816589, + "num_tokens": 265121530.0, + "step": 6829 + }, + { + "epoch": 0.868846202773184, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.539735794067383, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8833370804786682, + "num_tokens": 265153321.0, + "step": 6830 + }, + { + "epoch": 0.8689734130517746, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.351771354675293, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8906662464141846, + "num_tokens": 265191334.0, + "step": 6831 + }, + { + "epoch": 0.8691006233303651, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.315863609313965, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8795801401138306, + "num_tokens": 265229570.0, + "step": 6832 + }, + { + "epoch": 0.8692278336089556, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.386565208435059, + "learning_rate": 1e-06, + "loss": 0.4019, + "mean_token_accuracy": 0.8654820919036865, + "num_tokens": 265272091.0, + "step": 6833 + }, + { + "epoch": 0.8693550438875461, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.414896011352539, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8859133720397949, + "num_tokens": 265309180.0, + "step": 6834 + }, + { + "epoch": 0.8694822541661367, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.41701889038086, + "learning_rate": 1e-06, + "loss": 0.422, + "mean_token_accuracy": 0.8588812351226807, + "num_tokens": 265345556.0, + "step": 6835 + }, + { + "epoch": 0.8696094644447271, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.405413627624512, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.886486291885376, + "num_tokens": 265380518.0, + "step": 6836 + }, + { + "epoch": 0.8697366747233176, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.295472145080566, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8914190530776978, + "num_tokens": 265422879.0, + "step": 6837 + }, + { + "epoch": 0.8698638850019081, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.526571273803711, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8772728443145752, + "num_tokens": 265455177.0, + "step": 6838 + }, + { + "epoch": 0.8699910952804987, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.327142715454102, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8776081204414368, + "num_tokens": 265500409.0, + "step": 6839 + }, + { + "epoch": 0.8701183055590892, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.443964004516602, + "learning_rate": 1e-06, + "loss": 0.4251, + "mean_token_accuracy": 0.857571005821228, + "num_tokens": 265539102.0, + "step": 6840 + }, + { + "epoch": 0.8702455158376797, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.55976676940918, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.879883885383606, + "num_tokens": 265568054.0, + "step": 6841 + }, + { + "epoch": 0.8703727261162701, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.285014152526855, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8893513083457947, + "num_tokens": 265603980.0, + "step": 6842 + }, + { + "epoch": 0.8704999363948607, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.511490821838379, + "learning_rate": 1e-06, + "loss": 0.3919, + "mean_token_accuracy": 0.8671236038208008, + "num_tokens": 265638339.0, + "step": 6843 + }, + { + "epoch": 0.8706271466734512, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.449991226196289, + "learning_rate": 1e-06, + "loss": 0.4038, + "mean_token_accuracy": 0.8686333298683167, + "num_tokens": 265678543.0, + "step": 6844 + }, + { + "epoch": 0.8707543569520417, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.371707916259766, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.877030611038208, + "num_tokens": 265714667.0, + "step": 6845 + }, + { + "epoch": 0.8708815672306323, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.473503112792969, + "learning_rate": 1e-06, + "loss": 0.3999, + "mean_token_accuracy": 0.8643665313720703, + "num_tokens": 265750218.0, + "step": 6846 + }, + { + "epoch": 0.8710087775092228, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.314140319824219, + "learning_rate": 1e-06, + "loss": 0.4097, + "mean_token_accuracy": 0.8639342188835144, + "num_tokens": 265792695.0, + "step": 6847 + }, + { + "epoch": 0.8711359877878132, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.433781623840332, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8710187673568726, + "num_tokens": 265830128.0, + "step": 6848 + }, + { + "epoch": 0.8712631980664037, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.444337844848633, + "learning_rate": 1e-06, + "loss": 0.4402, + "mean_token_accuracy": 0.8532041311264038, + "num_tokens": 265868635.0, + "step": 6849 + }, + { + "epoch": 0.8713904083449943, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.292181015014648, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8897395133972168, + "num_tokens": 265907818.0, + "step": 6850 + }, + { + "epoch": 0.8715176186235848, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.40972900390625, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8928499221801758, + "num_tokens": 265941258.0, + "step": 6851 + }, + { + "epoch": 0.8716448289021753, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.43896198272705, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8801974058151245, + "num_tokens": 265977405.0, + "step": 6852 + }, + { + "epoch": 0.8717720391807658, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.250914573669434, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8917841911315918, + "num_tokens": 266021782.0, + "step": 6853 + }, + { + "epoch": 0.8718992494593563, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.333230018615723, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8852781653404236, + "num_tokens": 266063753.0, + "step": 6854 + }, + { + "epoch": 0.8720264597379468, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.402115821838379, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8856146335601807, + "num_tokens": 266100625.0, + "step": 6855 + }, + { + "epoch": 0.8721536700165373, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.412745475769043, + "learning_rate": 1e-06, + "loss": 0.4057, + "mean_token_accuracy": 0.8592618107795715, + "num_tokens": 266142898.0, + "step": 6856 + }, + { + "epoch": 0.8722808802951278, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.294819831848145, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8812118768692017, + "num_tokens": 266184423.0, + "step": 6857 + }, + { + "epoch": 0.8724080905737184, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.378155708312988, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8888267874717712, + "num_tokens": 266218135.0, + "step": 6858 + }, + { + "epoch": 0.8725353008523089, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.432272911071777, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8929966688156128, + "num_tokens": 266249340.0, + "step": 6859 + }, + { + "epoch": 0.8726625111308994, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.351503372192383, + "learning_rate": 1e-06, + "loss": 0.4083, + "mean_token_accuracy": 0.8658473491668701, + "num_tokens": 266291941.0, + "step": 6860 + }, + { + "epoch": 0.8727897214094898, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.304290771484375, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.876008152961731, + "num_tokens": 266333892.0, + "step": 6861 + }, + { + "epoch": 0.8729169316880804, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.396096229553223, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8803015351295471, + "num_tokens": 266370977.0, + "step": 6862 + }, + { + "epoch": 0.8730441419666709, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.425705909729004, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8850101232528687, + "num_tokens": 266408315.0, + "step": 6863 + }, + { + "epoch": 0.8731713522452614, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.293832778930664, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8815248012542725, + "num_tokens": 266452400.0, + "step": 6864 + }, + { + "epoch": 0.873298562523852, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.30876636505127, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8850140571594238, + "num_tokens": 266497452.0, + "step": 6865 + }, + { + "epoch": 0.8734257728024425, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.422868728637695, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8868357539176941, + "num_tokens": 266532033.0, + "step": 6866 + }, + { + "epoch": 0.8735529830810329, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.40978717803955, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8756403923034668, + "num_tokens": 266571353.0, + "step": 6867 + }, + { + "epoch": 0.8736801933596234, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.382847785949707, + "learning_rate": 1e-06, + "loss": 0.4074, + "mean_token_accuracy": 0.8627145886421204, + "num_tokens": 266610153.0, + "step": 6868 + }, + { + "epoch": 0.873807403638214, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 9.076574325561523, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.8724299669265747, + "num_tokens": 266642352.0, + "step": 6869 + }, + { + "epoch": 0.8739346139168045, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.337813377380371, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8810363411903381, + "num_tokens": 266680265.0, + "step": 6870 + }, + { + "epoch": 0.874061824195395, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.059906005859375e-06, + "grad_norm": 8.314871788024902, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8719611167907715, + "num_tokens": 266724106.0, + "step": 6871 + }, + { + "epoch": 0.8741890344739855, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.422560691833496, + "learning_rate": 1e-06, + "loss": 0.4009, + "mean_token_accuracy": 0.8660444021224976, + "num_tokens": 266761846.0, + "step": 6872 + }, + { + "epoch": 0.874316244752576, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.333353996276855, + "learning_rate": 1e-06, + "loss": 0.402, + "mean_token_accuracy": 0.8654255867004395, + "num_tokens": 266802909.0, + "step": 6873 + }, + { + "epoch": 0.8744434550311665, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.354146003723145, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8810421824455261, + "num_tokens": 266840989.0, + "step": 6874 + }, + { + "epoch": 0.874570665309757, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.341663360595703, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8795174956321716, + "num_tokens": 266883670.0, + "step": 6875 + }, + { + "epoch": 0.8746978755883476, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.37717056274414, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8728427886962891, + "num_tokens": 266922375.0, + "step": 6876 + }, + { + "epoch": 0.8748250858669381, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.379319190979004, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8895537257194519, + "num_tokens": 266957912.0, + "step": 6877 + }, + { + "epoch": 0.8749522961455286, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.374503135681152, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8795772194862366, + "num_tokens": 266999514.0, + "step": 6878 + }, + { + "epoch": 0.875079506424119, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.408473014831543, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8907054662704468, + "num_tokens": 267038580.0, + "step": 6879 + }, + { + "epoch": 0.8752067167027096, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.370162010192871, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8793618679046631, + "num_tokens": 267076096.0, + "step": 6880 + }, + { + "epoch": 0.8753339269813001, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.262297630310059, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8868107795715332, + "num_tokens": 267122951.0, + "step": 6881 + }, + { + "epoch": 0.8754611372598906, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.360963821411133, + "learning_rate": 1e-06, + "loss": 0.3647, + "mean_token_accuracy": 0.87720787525177, + "num_tokens": 267161307.0, + "step": 6882 + }, + { + "epoch": 0.8755883475384811, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.358810424804688, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8968782424926758, + "num_tokens": 267202821.0, + "step": 6883 + }, + { + "epoch": 0.8757155578170717, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.359506607055664, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8725147247314453, + "num_tokens": 267245310.0, + "step": 6884 + }, + { + "epoch": 0.8758427680956621, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.410118103027344, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8752106428146362, + "num_tokens": 267282432.0, + "step": 6885 + }, + { + "epoch": 0.8759699783742526, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.345532417297363, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8820443153381348, + "num_tokens": 267319835.0, + "step": 6886 + }, + { + "epoch": 0.8760971886528431, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.363582611083984, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.872891902923584, + "num_tokens": 267362305.0, + "step": 6887 + }, + { + "epoch": 0.8762243989314337, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.498790740966797, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8712509870529175, + "num_tokens": 267396659.0, + "step": 6888 + }, + { + "epoch": 0.8763516092100242, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.3768892288208, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8774152994155884, + "num_tokens": 267434362.0, + "step": 6889 + }, + { + "epoch": 0.8764788194886147, + "ewc_loss": 0.01220703125, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.32091236114502, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.874856173992157, + "num_tokens": 267475974.0, + "step": 6890 + }, + { + "epoch": 0.8766060297672051, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.432892799377441, + "learning_rate": 1e-06, + "loss": 0.3861, + "mean_token_accuracy": 0.869243860244751, + "num_tokens": 267516984.0, + "step": 6891 + }, + { + "epoch": 0.8767332400457957, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.438026428222656, + "learning_rate": 1e-06, + "loss": 0.4009, + "mean_token_accuracy": 0.8660221695899963, + "num_tokens": 267556495.0, + "step": 6892 + }, + { + "epoch": 0.8768604503243862, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.437867164611816, + "learning_rate": 1e-06, + "loss": 0.4078, + "mean_token_accuracy": 0.8622240424156189, + "num_tokens": 267592603.0, + "step": 6893 + }, + { + "epoch": 0.8769876606029767, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.281209945678711, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8830132484436035, + "num_tokens": 267632833.0, + "step": 6894 + }, + { + "epoch": 0.8771148708815673, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.33591079711914, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8876803517341614, + "num_tokens": 267675663.0, + "step": 6895 + }, + { + "epoch": 0.8772420811601578, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.434159278869629, + "learning_rate": 1e-06, + "loss": 0.4206, + "mean_token_accuracy": 0.8633488416671753, + "num_tokens": 267715629.0, + "step": 6896 + }, + { + "epoch": 0.8773692914387482, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.419147491455078, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8832627534866333, + "num_tokens": 267754545.0, + "step": 6897 + }, + { + "epoch": 0.8774965017173387, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.400858879089355, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8796976804733276, + "num_tokens": 267788862.0, + "step": 6898 + }, + { + "epoch": 0.8776237119959293, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.559523582458496, + "learning_rate": 1e-06, + "loss": 0.3809, + "mean_token_accuracy": 0.87031489610672, + "num_tokens": 267818665.0, + "step": 6899 + }, + { + "epoch": 0.8777509222745198, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.147125244140625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.309816360473633, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8786136507987976, + "num_tokens": 267861380.0, + "step": 6900 + }, + { + "epoch": 0.8778781325531103, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.295212745666504, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8958461880683899, + "num_tokens": 267900311.0, + "step": 6901 + }, + { + "epoch": 0.8780053428317008, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.4725923538208, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8825896382331848, + "num_tokens": 267932501.0, + "step": 6902 + }, + { + "epoch": 0.8781325531102913, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.328678131103516, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8742645382881165, + "num_tokens": 267979122.0, + "step": 6903 + }, + { + "epoch": 0.8782597633888818, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.339375495910645, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8853450417518616, + "num_tokens": 268017841.0, + "step": 6904 + }, + { + "epoch": 0.8783869736674723, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.321829795837402, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8809083700180054, + "num_tokens": 268062392.0, + "step": 6905 + }, + { + "epoch": 0.8785141839460628, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.41181755065918, + "learning_rate": 1e-06, + "loss": 0.4093, + "mean_token_accuracy": 0.8639476895332336, + "num_tokens": 268102336.0, + "step": 6906 + }, + { + "epoch": 0.8786413942246534, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.449017524719238, + "learning_rate": 1e-06, + "loss": 0.4151, + "mean_token_accuracy": 0.8635599613189697, + "num_tokens": 268142046.0, + "step": 6907 + }, + { + "epoch": 0.8787686045032439, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.36489486694336, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.8705900311470032, + "num_tokens": 268186768.0, + "step": 6908 + }, + { + "epoch": 0.8788958147818343, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.354409217834473, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8799062967300415, + "num_tokens": 268224406.0, + "step": 6909 + }, + { + "epoch": 0.8790230250604248, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.414809226989746, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8845205307006836, + "num_tokens": 268258775.0, + "step": 6910 + }, + { + "epoch": 0.8791502353390154, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.380899429321289, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8761147260665894, + "num_tokens": 268297530.0, + "step": 6911 + }, + { + "epoch": 0.8792774456176059, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.463761329650879, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.881796658039093, + "num_tokens": 268333893.0, + "step": 6912 + }, + { + "epoch": 0.8794046558961964, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.326355934143066, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8820552229881287, + "num_tokens": 268372633.0, + "step": 6913 + }, + { + "epoch": 0.879531866174787, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.38890266418457, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8800253868103027, + "num_tokens": 268413223.0, + "step": 6914 + }, + { + "epoch": 0.8796590764533775, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.342287063598633, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8884389400482178, + "num_tokens": 268449859.0, + "step": 6915 + }, + { + "epoch": 0.8797862867319679, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.434764862060547, + "learning_rate": 1e-06, + "loss": 0.3988, + "mean_token_accuracy": 0.8649032115936279, + "num_tokens": 268487692.0, + "step": 6916 + }, + { + "epoch": 0.8799134970105584, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.341906547546387, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8817604780197144, + "num_tokens": 268529697.0, + "step": 6917 + }, + { + "epoch": 0.880040707289149, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.378456115722656, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8809701204299927, + "num_tokens": 268568917.0, + "step": 6918 + }, + { + "epoch": 0.8801679175677395, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.470256805419922, + "learning_rate": 1e-06, + "loss": 0.4007, + "mean_token_accuracy": 0.8670368790626526, + "num_tokens": 268607511.0, + "step": 6919 + }, + { + "epoch": 0.88029512784633, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.498333930969238, + "learning_rate": 1e-06, + "loss": 0.426, + "mean_token_accuracy": 0.85921311378479, + "num_tokens": 268640246.0, + "step": 6920 + }, + { + "epoch": 0.8804223381249205, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.361008644104004, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8880500793457031, + "num_tokens": 268678024.0, + "step": 6921 + }, + { + "epoch": 0.880549548403511, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.332791328430176, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8738064765930176, + "num_tokens": 268720567.0, + "step": 6922 + }, + { + "epoch": 0.8806767586821015, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.405862808227539, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8859913349151611, + "num_tokens": 268754977.0, + "step": 6923 + }, + { + "epoch": 0.880803968960692, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.486739158630371, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8782389760017395, + "num_tokens": 268788933.0, + "step": 6924 + }, + { + "epoch": 0.8809311792392825, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.375990867614746, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8816462159156799, + "num_tokens": 268823693.0, + "step": 6925 + }, + { + "epoch": 0.8810583895178731, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.39869213104248, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8777515888214111, + "num_tokens": 268863134.0, + "step": 6926 + }, + { + "epoch": 0.8811855997964636, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.417532920837402, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.870858907699585, + "num_tokens": 268902002.0, + "step": 6927 + }, + { + "epoch": 0.881312810075054, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.401116371154785, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8846210241317749, + "num_tokens": 268940870.0, + "step": 6928 + }, + { + "epoch": 0.8814400203536445, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.351815223693848, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8801027536392212, + "num_tokens": 268981221.0, + "step": 6929 + }, + { + "epoch": 0.8815672306322351, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.40628433227539, + "learning_rate": 1e-06, + "loss": 0.4291, + "mean_token_accuracy": 0.8591396808624268, + "num_tokens": 269024656.0, + "step": 6930 + }, + { + "epoch": 0.8816944409108256, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.351751327514648, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8807640671730042, + "num_tokens": 269063488.0, + "step": 6931 + }, + { + "epoch": 0.8818216511894161, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.424848556518555, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8882818222045898, + "num_tokens": 269103272.0, + "step": 6932 + }, + { + "epoch": 0.8819488614680067, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.373563766479492, + "learning_rate": 1e-06, + "loss": 0.3916, + "mean_token_accuracy": 0.8687868118286133, + "num_tokens": 269144883.0, + "step": 6933 + }, + { + "epoch": 0.8820760717465971, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.514115333557129, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8712762594223022, + "num_tokens": 269177166.0, + "step": 6934 + }, + { + "epoch": 0.8822032820251876, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.378937721252441, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.8736512660980225, + "num_tokens": 269219803.0, + "step": 6935 + }, + { + "epoch": 0.8823304923037781, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.33491039276123, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8841601014137268, + "num_tokens": 269257010.0, + "step": 6936 + }, + { + "epoch": 0.8824577025823687, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.355897903442383, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8745582103729248, + "num_tokens": 269303499.0, + "step": 6937 + }, + { + "epoch": 0.8825849128609592, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.402709007263184, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8844934701919556, + "num_tokens": 269337739.0, + "step": 6938 + }, + { + "epoch": 0.8827121231395497, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.318450927734375, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.891353189945221, + "num_tokens": 269381126.0, + "step": 6939 + }, + { + "epoch": 0.8828393334181401, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.362499237060547, + "learning_rate": 1e-06, + "loss": 0.3806, + "mean_token_accuracy": 0.8717373609542847, + "num_tokens": 269423745.0, + "step": 6940 + }, + { + "epoch": 0.8829665436967307, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.568794250488281, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8734585046768188, + "num_tokens": 269456054.0, + "step": 6941 + }, + { + "epoch": 0.8830937539753212, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.457983016967773, + "learning_rate": 1e-06, + "loss": 0.3986, + "mean_token_accuracy": 0.8643857836723328, + "num_tokens": 269490427.0, + "step": 6942 + }, + { + "epoch": 0.8832209642539117, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.360895156860352, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8727552890777588, + "num_tokens": 269529787.0, + "step": 6943 + }, + { + "epoch": 0.8833481745325023, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.303385734558105, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8837769627571106, + "num_tokens": 269572995.0, + "step": 6944 + }, + { + "epoch": 0.8834753848110928, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.379598617553711, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8776815533638, + "num_tokens": 269612699.0, + "step": 6945 + }, + { + "epoch": 0.8836025950896832, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.366944313049316, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8857789635658264, + "num_tokens": 269644368.0, + "step": 6946 + }, + { + "epoch": 0.8837298053682737, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.39396858215332, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8761065006256104, + "num_tokens": 269683900.0, + "step": 6947 + }, + { + "epoch": 0.8838570156468643, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.35156536102295, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8939931392669678, + "num_tokens": 269723916.0, + "step": 6948 + }, + { + "epoch": 0.8839842259254548, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.463685035705566, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8746628165245056, + "num_tokens": 269757974.0, + "step": 6949 + }, + { + "epoch": 0.8841114362040453, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.518620491027832, + "learning_rate": 1e-06, + "loss": 0.4169, + "mean_token_accuracy": 0.86087965965271, + "num_tokens": 269794538.0, + "step": 6950 + }, + { + "epoch": 0.8842386464826358, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.364608764648438, + "learning_rate": 1e-06, + "loss": 0.4332, + "mean_token_accuracy": 0.8552036285400391, + "num_tokens": 269839569.0, + "step": 6951 + }, + { + "epoch": 0.8843658567612263, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.4065523147583, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.870464563369751, + "num_tokens": 269883219.0, + "step": 6952 + }, + { + "epoch": 0.8844930670398168, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.404535293579102, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.879085123538971, + "num_tokens": 269923677.0, + "step": 6953 + }, + { + "epoch": 0.8846202773184073, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.406819343566895, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8898026347160339, + "num_tokens": 269958980.0, + "step": 6954 + }, + { + "epoch": 0.8847474875969978, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.342915534973145, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8859966993331909, + "num_tokens": 270000413.0, + "step": 6955 + }, + { + "epoch": 0.8848746978755884, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.380903244018555, + "learning_rate": 1e-06, + "loss": 0.3793, + "mean_token_accuracy": 0.8699893951416016, + "num_tokens": 270044801.0, + "step": 6956 + }, + { + "epoch": 0.8850019081541789, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.40978717803955, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8726605176925659, + "num_tokens": 270081617.0, + "step": 6957 + }, + { + "epoch": 0.8851291184327693, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.434639930725098, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8735601902008057, + "num_tokens": 270118701.0, + "step": 6958 + }, + { + "epoch": 0.8852563287113598, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.38427448272705, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8846076726913452, + "num_tokens": 270155271.0, + "step": 6959 + }, + { + "epoch": 0.8853835389899504, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.366683959960938, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8829655647277832, + "num_tokens": 270192683.0, + "step": 6960 + }, + { + "epoch": 0.8855107492685409, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.405549049377441, + "learning_rate": 1e-06, + "loss": 0.4079, + "mean_token_accuracy": 0.8633871078491211, + "num_tokens": 270232413.0, + "step": 6961 + }, + { + "epoch": 0.8856379595471314, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.414933204650879, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8806151151657104, + "num_tokens": 270270395.0, + "step": 6962 + }, + { + "epoch": 0.885765169825722, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.334959983825684, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8822356462478638, + "num_tokens": 270313561.0, + "step": 6963 + }, + { + "epoch": 0.8858923801043125, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.350431442260742, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8794246315956116, + "num_tokens": 270356341.0, + "step": 6964 + }, + { + "epoch": 0.8860195903829029, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.354649543762207, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8796066641807556, + "num_tokens": 270395238.0, + "step": 6965 + }, + { + "epoch": 0.8861468006614934, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.404045104980469, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.8734697699546814, + "num_tokens": 270433845.0, + "step": 6966 + }, + { + "epoch": 0.886274010940084, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.438352584838867, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8832428455352783, + "num_tokens": 270472324.0, + "step": 6967 + }, + { + "epoch": 0.8864012212186745, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.397353172302246, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8745843172073364, + "num_tokens": 270510709.0, + "step": 6968 + }, + { + "epoch": 0.886528431497265, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.427291870117188, + "learning_rate": 1e-06, + "loss": 0.4362, + "mean_token_accuracy": 0.852234423160553, + "num_tokens": 270553423.0, + "step": 6969 + }, + { + "epoch": 0.8866556417758555, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.408478736877441, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8886469602584839, + "num_tokens": 270589069.0, + "step": 6970 + }, + { + "epoch": 0.886782852054446, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.385834693908691, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8805102705955505, + "num_tokens": 270628012.0, + "step": 6971 + }, + { + "epoch": 0.8869100623330365, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.423752784729004, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8869057297706604, + "num_tokens": 270660194.0, + "step": 6972 + }, + { + "epoch": 0.887037272611627, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.561484336853027, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8742592334747314, + "num_tokens": 270694078.0, + "step": 6973 + }, + { + "epoch": 0.8871644828902175, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.418058395385742, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.877346932888031, + "num_tokens": 270726895.0, + "step": 6974 + }, + { + "epoch": 0.8872916931688081, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.33184814453125, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8933613896369934, + "num_tokens": 270768059.0, + "step": 6975 + }, + { + "epoch": 0.8874189034473986, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.336479187011719, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8838422298431396, + "num_tokens": 270806487.0, + "step": 6976 + }, + { + "epoch": 0.887546113725989, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.413728713989258, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8784198760986328, + "num_tokens": 270842675.0, + "step": 6977 + }, + { + "epoch": 0.8876733240045795, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.453852653503418, + "learning_rate": 1e-06, + "loss": 0.4411, + "mean_token_accuracy": 0.8502172231674194, + "num_tokens": 270884132.0, + "step": 6978 + }, + { + "epoch": 0.8878005342831701, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.347082138061523, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8867020010948181, + "num_tokens": 270924412.0, + "step": 6979 + }, + { + "epoch": 0.8879277445617606, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.417484283447266, + "learning_rate": 1e-06, + "loss": 0.388, + "mean_token_accuracy": 0.8663869500160217, + "num_tokens": 270960160.0, + "step": 6980 + }, + { + "epoch": 0.8880549548403511, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.462611198425293, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8825324773788452, + "num_tokens": 270992783.0, + "step": 6981 + }, + { + "epoch": 0.8881821651189417, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.306015014648438, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8787431120872498, + "num_tokens": 271034531.0, + "step": 6982 + }, + { + "epoch": 0.8883093753975321, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.366958618164062, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.889882504940033, + "num_tokens": 271071314.0, + "step": 6983 + }, + { + "epoch": 0.8884365856761226, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.324151039123535, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8873276114463806, + "num_tokens": 271114452.0, + "step": 6984 + }, + { + "epoch": 0.8885637959547131, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.393040657043457, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8876379728317261, + "num_tokens": 271151042.0, + "step": 6985 + }, + { + "epoch": 0.8886910062333037, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.384243965148926, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8867567777633667, + "num_tokens": 271191826.0, + "step": 6986 + }, + { + "epoch": 0.8888182165118942, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.426071166992188, + "learning_rate": 1e-06, + "loss": 0.389, + "mean_token_accuracy": 0.8674682378768921, + "num_tokens": 271225506.0, + "step": 6987 + }, + { + "epoch": 0.8889454267904847, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.346586227416992, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.8698960542678833, + "num_tokens": 271268187.0, + "step": 6988 + }, + { + "epoch": 0.8890726370690751, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.365456581115723, + "learning_rate": 1e-06, + "loss": 0.3754, + "mean_token_accuracy": 0.8725219964981079, + "num_tokens": 271309818.0, + "step": 6989 + }, + { + "epoch": 0.8891998473476657, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.350873947143555, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8838344216346741, + "num_tokens": 271347999.0, + "step": 6990 + }, + { + "epoch": 0.8893270576262562, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.510457038879395, + "learning_rate": 1e-06, + "loss": 0.4086, + "mean_token_accuracy": 0.859673261642456, + "num_tokens": 271387847.0, + "step": 6991 + }, + { + "epoch": 0.8894542679048467, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.389983177185059, + "learning_rate": 1e-06, + "loss": 0.3941, + "mean_token_accuracy": 0.8647144436836243, + "num_tokens": 271428082.0, + "step": 6992 + }, + { + "epoch": 0.8895814781834372, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.372495651245117, + "learning_rate": 1e-06, + "loss": 0.3774, + "mean_token_accuracy": 0.8700200319290161, + "num_tokens": 271472956.0, + "step": 6993 + }, + { + "epoch": 0.8897086884620278, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.463685035705566, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8801604509353638, + "num_tokens": 271506080.0, + "step": 6994 + }, + { + "epoch": 0.8898358987406182, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.309149742126465, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8778036236763, + "num_tokens": 271552645.0, + "step": 6995 + }, + { + "epoch": 0.8899631090192087, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.381486892700195, + "learning_rate": 1e-06, + "loss": 0.4012, + "mean_token_accuracy": 0.8629868626594543, + "num_tokens": 271594779.0, + "step": 6996 + }, + { + "epoch": 0.8900903192977992, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.372254371643066, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8749195337295532, + "num_tokens": 271637853.0, + "step": 6997 + }, + { + "epoch": 0.8902175295763898, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.372699737548828, + "learning_rate": 1e-06, + "loss": 0.3597, + "mean_token_accuracy": 0.8775204420089722, + "num_tokens": 271675826.0, + "step": 6998 + }, + { + "epoch": 0.8903447398549803, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.433627128601074, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8777506351470947, + "num_tokens": 271709721.0, + "step": 6999 + }, + { + "epoch": 0.8904719501335708, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.478038787841797, + "learning_rate": 1e-06, + "loss": 0.4035, + "mean_token_accuracy": 0.8683724999427795, + "num_tokens": 271747960.0, + "step": 7000 + }, + { + "epoch": 0.8905991604121613, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.49251937866211, + "learning_rate": 1e-06, + "loss": 0.4092, + "mean_token_accuracy": 0.8670294284820557, + "num_tokens": 271780540.0, + "step": 7001 + }, + { + "epoch": 0.8907263706907518, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.370412826538086, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8699520230293274, + "num_tokens": 271818635.0, + "step": 7002 + }, + { + "epoch": 0.8908535809693423, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 9.448745727539062, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8811904788017273, + "num_tokens": 271861282.0, + "step": 7003 + }, + { + "epoch": 0.8909807912479328, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.342852592468262, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8827848434448242, + "num_tokens": 271902408.0, + "step": 7004 + }, + { + "epoch": 0.8911080015265234, + "ewc_loss": 0.01214599609375, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.000301361083984e-06, + "grad_norm": 8.380882263183594, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8804428577423096, + "num_tokens": 271938765.0, + "step": 7005 + }, + { + "epoch": 0.8912352118051139, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.420105934143066, + "learning_rate": 1e-06, + "loss": 0.4244, + "mean_token_accuracy": 0.859235405921936, + "num_tokens": 271979959.0, + "step": 7006 + }, + { + "epoch": 0.8913624220837043, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.32082748413086, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8740831613540649, + "num_tokens": 272016947.0, + "step": 7007 + }, + { + "epoch": 0.8914896323622948, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.376245498657227, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8855681419372559, + "num_tokens": 272058006.0, + "step": 7008 + }, + { + "epoch": 0.8916168426408854, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.388205528259277, + "learning_rate": 1e-06, + "loss": 0.3952, + "mean_token_accuracy": 0.8650925159454346, + "num_tokens": 272099384.0, + "step": 7009 + }, + { + "epoch": 0.8917440529194759, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.438971519470215, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8821751475334167, + "num_tokens": 272140501.0, + "step": 7010 + }, + { + "epoch": 0.8918712631980664, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.404590606689453, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8736436367034912, + "num_tokens": 272179931.0, + "step": 7011 + }, + { + "epoch": 0.891998473476657, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.46384048461914, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.874180793762207, + "num_tokens": 272213891.0, + "step": 7012 + }, + { + "epoch": 0.8921256837552475, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.3800630569458, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8769592046737671, + "num_tokens": 272252968.0, + "step": 7013 + }, + { + "epoch": 0.8922528940338379, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.406249046325684, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8781290650367737, + "num_tokens": 272294136.0, + "step": 7014 + }, + { + "epoch": 0.8923801043124284, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.490199089050293, + "learning_rate": 1e-06, + "loss": 0.4161, + "mean_token_accuracy": 0.857811689376831, + "num_tokens": 272332840.0, + "step": 7015 + }, + { + "epoch": 0.892507314591019, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.41284465789795, + "learning_rate": 1e-06, + "loss": 0.3676, + "mean_token_accuracy": 0.8785051107406616, + "num_tokens": 272369913.0, + "step": 7016 + }, + { + "epoch": 0.8926345248696095, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.431075096130371, + "learning_rate": 1e-06, + "loss": 0.3724, + "mean_token_accuracy": 0.87652188539505, + "num_tokens": 272407354.0, + "step": 7017 + }, + { + "epoch": 0.8927617351482, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.444742202758789, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8765602707862854, + "num_tokens": 272442978.0, + "step": 7018 + }, + { + "epoch": 0.8928889454267905, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.39594554901123, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.882083535194397, + "num_tokens": 272477183.0, + "step": 7019 + }, + { + "epoch": 0.893016155705381, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.316120147705078, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.871416449546814, + "num_tokens": 272517213.0, + "step": 7020 + }, + { + "epoch": 0.8931433659839715, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.435388565063477, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8856359124183655, + "num_tokens": 272557475.0, + "step": 7021 + }, + { + "epoch": 0.893270576262562, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.343317031860352, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8863720297813416, + "num_tokens": 272600368.0, + "step": 7022 + }, + { + "epoch": 0.8933977865411525, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.298052787780762, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8932902812957764, + "num_tokens": 272643702.0, + "step": 7023 + }, + { + "epoch": 0.8935249968197431, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.413552284240723, + "learning_rate": 1e-06, + "loss": 0.381, + "mean_token_accuracy": 0.8770068883895874, + "num_tokens": 272680703.0, + "step": 7024 + }, + { + "epoch": 0.8936522070983336, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.408315658569336, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8906197547912598, + "num_tokens": 272720829.0, + "step": 7025 + }, + { + "epoch": 0.893779417376924, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.529889106750488, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.89232337474823, + "num_tokens": 272750978.0, + "step": 7026 + }, + { + "epoch": 0.8939066276555145, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.48568058013916, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.876455545425415, + "num_tokens": 272786787.0, + "step": 7027 + }, + { + "epoch": 0.8940338379341051, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.462956428527832, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8778133392333984, + "num_tokens": 272820817.0, + "step": 7028 + }, + { + "epoch": 0.8941610482126956, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.371978759765625, + "learning_rate": 1e-06, + "loss": 0.3647, + "mean_token_accuracy": 0.8777705430984497, + "num_tokens": 272862918.0, + "step": 7029 + }, + { + "epoch": 0.8942882584912861, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.447155952453613, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.8735809922218323, + "num_tokens": 272907887.0, + "step": 7030 + }, + { + "epoch": 0.8944154687698767, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.370593070983887, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8862119317054749, + "num_tokens": 272946516.0, + "step": 7031 + }, + { + "epoch": 0.8945426790484671, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.424095153808594, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8754517436027527, + "num_tokens": 272983015.0, + "step": 7032 + }, + { + "epoch": 0.8946698893270576, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.531556129455566, + "learning_rate": 1e-06, + "loss": 0.3938, + "mean_token_accuracy": 0.8695537447929382, + "num_tokens": 273016601.0, + "step": 7033 + }, + { + "epoch": 0.8947970996056481, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.42704963684082, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8914774656295776, + "num_tokens": 273053042.0, + "step": 7034 + }, + { + "epoch": 0.8949243098842387, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.3854398727417, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8928031921386719, + "num_tokens": 273088668.0, + "step": 7035 + }, + { + "epoch": 0.8950515201628292, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.446983337402344, + "learning_rate": 1e-06, + "loss": 0.3851, + "mean_token_accuracy": 0.8712755441665649, + "num_tokens": 273126895.0, + "step": 7036 + }, + { + "epoch": 0.8951787304414197, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.377782821655273, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.876279354095459, + "num_tokens": 273166428.0, + "step": 7037 + }, + { + "epoch": 0.8953059407200101, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.390419006347656, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8857616186141968, + "num_tokens": 273200877.0, + "step": 7038 + }, + { + "epoch": 0.8954331509986007, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.440900802612305, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8739102482795715, + "num_tokens": 273234960.0, + "step": 7039 + }, + { + "epoch": 0.8955603612771912, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.41640853881836, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8720442056655884, + "num_tokens": 273270311.0, + "step": 7040 + }, + { + "epoch": 0.8956875715557817, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.34270191192627, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8719799518585205, + "num_tokens": 273312677.0, + "step": 7041 + }, + { + "epoch": 0.8958147818343722, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.387345314025879, + "learning_rate": 1e-06, + "loss": 0.3965, + "mean_token_accuracy": 0.8676162958145142, + "num_tokens": 273359596.0, + "step": 7042 + }, + { + "epoch": 0.8959419921129628, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.348856925964355, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8819493055343628, + "num_tokens": 273401872.0, + "step": 7043 + }, + { + "epoch": 0.8960692023915532, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.344111442565918, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8870049118995667, + "num_tokens": 273440069.0, + "step": 7044 + }, + { + "epoch": 0.8961964126701437, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.432032585144043, + "learning_rate": 1e-06, + "loss": 0.4065, + "mean_token_accuracy": 0.8604913949966431, + "num_tokens": 273480155.0, + "step": 7045 + }, + { + "epoch": 0.8963236229487342, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.418353080749512, + "learning_rate": 1e-06, + "loss": 0.3917, + "mean_token_accuracy": 0.8695454597473145, + "num_tokens": 273516674.0, + "step": 7046 + }, + { + "epoch": 0.8964508332273248, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.403775215148926, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8856090307235718, + "num_tokens": 273550958.0, + "step": 7047 + }, + { + "epoch": 0.8965780435059153, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.439713478088379, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8812297582626343, + "num_tokens": 273582736.0, + "step": 7048 + }, + { + "epoch": 0.8967052537845058, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.391444206237793, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8756135702133179, + "num_tokens": 273622318.0, + "step": 7049 + }, + { + "epoch": 0.8968324640630962, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.339142799377441, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8793874979019165, + "num_tokens": 273664653.0, + "step": 7050 + }, + { + "epoch": 0.8969596743416868, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.369161605834961, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8868566751480103, + "num_tokens": 273702189.0, + "step": 7051 + }, + { + "epoch": 0.8970868846202773, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.37691593170166, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.887932300567627, + "num_tokens": 273741113.0, + "step": 7052 + }, + { + "epoch": 0.8972140948988678, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.461847305297852, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8758423328399658, + "num_tokens": 273779403.0, + "step": 7053 + }, + { + "epoch": 0.8973413051774584, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.383309364318848, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8811637163162231, + "num_tokens": 273816149.0, + "step": 7054 + }, + { + "epoch": 0.8974685154560489, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.37059497833252, + "learning_rate": 1e-06, + "loss": 0.361, + "mean_token_accuracy": 0.8793900012969971, + "num_tokens": 273855910.0, + "step": 7055 + }, + { + "epoch": 0.8975957257346393, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.403762817382812, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8813604116439819, + "num_tokens": 273891994.0, + "step": 7056 + }, + { + "epoch": 0.8977229360132298, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.373941421508789, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.886059045791626, + "num_tokens": 273932453.0, + "step": 7057 + }, + { + "epoch": 0.8978501462918204, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.170967102050781e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.362279891967773, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8849883079528809, + "num_tokens": 273972324.0, + "step": 7058 + }, + { + "epoch": 0.8979773565704109, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.37623119354248, + "learning_rate": 1e-06, + "loss": 0.3788, + "mean_token_accuracy": 0.8742274641990662, + "num_tokens": 274018147.0, + "step": 7059 + }, + { + "epoch": 0.8981045668490014, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.436455726623535, + "learning_rate": 1e-06, + "loss": 0.3868, + "mean_token_accuracy": 0.8732713460922241, + "num_tokens": 274059332.0, + "step": 7060 + }, + { + "epoch": 0.898231777127592, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.355203628540039, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8773951530456543, + "num_tokens": 274101043.0, + "step": 7061 + }, + { + "epoch": 0.8983589874061825, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.43061351776123, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8791475296020508, + "num_tokens": 274140218.0, + "step": 7062 + }, + { + "epoch": 0.8984861976847729, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.519619941711426, + "learning_rate": 1e-06, + "loss": 0.4229, + "mean_token_accuracy": 0.8561930656433105, + "num_tokens": 274176890.0, + "step": 7063 + }, + { + "epoch": 0.8986134079633634, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.39116096496582, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8992244601249695, + "num_tokens": 274213840.0, + "step": 7064 + }, + { + "epoch": 0.898740618241954, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.44274616241455, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.879044771194458, + "num_tokens": 274249319.0, + "step": 7065 + }, + { + "epoch": 0.8988678285205445, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.349241256713867, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8819116353988647, + "num_tokens": 274291814.0, + "step": 7066 + }, + { + "epoch": 0.898995038799135, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.507954597473145, + "learning_rate": 1e-06, + "loss": 0.3818, + "mean_token_accuracy": 0.8715397715568542, + "num_tokens": 274326054.0, + "step": 7067 + }, + { + "epoch": 0.8991222490777255, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.35969352722168, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8888764381408691, + "num_tokens": 274364150.0, + "step": 7068 + }, + { + "epoch": 0.899249459356316, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.427755355834961, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8671254515647888, + "num_tokens": 274404486.0, + "step": 7069 + }, + { + "epoch": 0.8993766696349065, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.41135025024414, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8836380839347839, + "num_tokens": 274440194.0, + "step": 7070 + }, + { + "epoch": 0.899503879913497, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.338709831237793, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8895509839057922, + "num_tokens": 274479582.0, + "step": 7071 + }, + { + "epoch": 0.8996310901920875, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.45434284210205, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8788710832595825, + "num_tokens": 274513455.0, + "step": 7072 + }, + { + "epoch": 0.8997583004706781, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.402177810668945, + "learning_rate": 1e-06, + "loss": 0.3681, + "mean_token_accuracy": 0.8738257884979248, + "num_tokens": 274552024.0, + "step": 7073 + }, + { + "epoch": 0.8998855107492686, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.377945899963379, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8812188506126404, + "num_tokens": 274592435.0, + "step": 7074 + }, + { + "epoch": 0.900012721027859, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.358912467956543, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8861753940582275, + "num_tokens": 274632676.0, + "step": 7075 + }, + { + "epoch": 0.9001399313064495, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.400896072387695, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8822388052940369, + "num_tokens": 274672787.0, + "step": 7076 + }, + { + "epoch": 0.9002671415850401, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.429502487182617, + "learning_rate": 1e-06, + "loss": 0.4154, + "mean_token_accuracy": 0.8558673858642578, + "num_tokens": 274711999.0, + "step": 7077 + }, + { + "epoch": 0.9003943518636306, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.360180854797363, + "learning_rate": 1e-06, + "loss": 0.403, + "mean_token_accuracy": 0.8637199401855469, + "num_tokens": 274755611.0, + "step": 7078 + }, + { + "epoch": 0.9005215621422211, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.37779712677002, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8729298114776611, + "num_tokens": 274795218.0, + "step": 7079 + }, + { + "epoch": 0.9006487724208116, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.32640266418457, + "learning_rate": 1e-06, + "loss": 0.3923, + "mean_token_accuracy": 0.8665037751197815, + "num_tokens": 274841235.0, + "step": 7080 + }, + { + "epoch": 0.9007759826994021, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.345603942871094, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8856959939002991, + "num_tokens": 274886488.0, + "step": 7081 + }, + { + "epoch": 0.9009031929779926, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.31906509399414, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8788895606994629, + "num_tokens": 274930197.0, + "step": 7082 + }, + { + "epoch": 0.9010304032565831, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.50560188293457, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8779057264328003, + "num_tokens": 274970780.0, + "step": 7083 + }, + { + "epoch": 0.9011576135351737, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.380396842956543, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.882429301738739, + "num_tokens": 275013578.0, + "step": 7084 + }, + { + "epoch": 0.9012848238137642, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.443673133850098, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.887195348739624, + "num_tokens": 275046230.0, + "step": 7085 + }, + { + "epoch": 0.9014120340923547, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.37807846069336, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8754977583885193, + "num_tokens": 275092174.0, + "step": 7086 + }, + { + "epoch": 0.9015392443709451, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.42355728149414, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8713322281837463, + "num_tokens": 275135743.0, + "step": 7087 + }, + { + "epoch": 0.9016664546495357, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.441794395446777, + "learning_rate": 1e-06, + "loss": 0.385, + "mean_token_accuracy": 0.8715255856513977, + "num_tokens": 275173721.0, + "step": 7088 + }, + { + "epoch": 0.9017936649281262, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.382126808166504, + "learning_rate": 1e-06, + "loss": 0.4099, + "mean_token_accuracy": 0.8643303513526917, + "num_tokens": 275218628.0, + "step": 7089 + }, + { + "epoch": 0.9019208752067167, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.346060752868652, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8962469696998596, + "num_tokens": 275255349.0, + "step": 7090 + }, + { + "epoch": 0.9020480854853072, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.397192001342773, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8864119052886963, + "num_tokens": 275298909.0, + "step": 7091 + }, + { + "epoch": 0.9021752957638978, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.382231712341309, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8885157704353333, + "num_tokens": 275336333.0, + "step": 7092 + }, + { + "epoch": 0.9023025060424882, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.427752494812012, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8910101652145386, + "num_tokens": 275372925.0, + "step": 7093 + }, + { + "epoch": 0.9024297163210787, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.393303871154785, + "learning_rate": 1e-06, + "loss": 0.4158, + "mean_token_accuracy": 0.863822877407074, + "num_tokens": 275411458.0, + "step": 7094 + }, + { + "epoch": 0.9025569265996692, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.461233139038086, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8824524283409119, + "num_tokens": 275446480.0, + "step": 7095 + }, + { + "epoch": 0.9026841368782598, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.498946189880371, + "learning_rate": 1e-06, + "loss": 0.4118, + "mean_token_accuracy": 0.862515926361084, + "num_tokens": 275480158.0, + "step": 7096 + }, + { + "epoch": 0.9028113471568503, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.479384422302246, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8852580785751343, + "num_tokens": 275522177.0, + "step": 7097 + }, + { + "epoch": 0.9029385574354408, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.305392265319824, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.892551839351654, + "num_tokens": 275559988.0, + "step": 7098 + }, + { + "epoch": 0.9030657677140312, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.398530006408691, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8784517645835876, + "num_tokens": 275601200.0, + "step": 7099 + }, + { + "epoch": 0.9031929779926218, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.433991432189941, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8975056409835815, + "num_tokens": 275634051.0, + "step": 7100 + }, + { + "epoch": 0.9033201882712123, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.414831161499023, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8834409713745117, + "num_tokens": 275670661.0, + "step": 7101 + }, + { + "epoch": 0.9034473985498028, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.381922721862793, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8745400309562683, + "num_tokens": 275711937.0, + "step": 7102 + }, + { + "epoch": 0.9035746088283934, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.453761100769043, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8656983971595764, + "num_tokens": 275753125.0, + "step": 7103 + }, + { + "epoch": 0.9037018191069839, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.406448364257812, + "learning_rate": 1e-06, + "loss": 0.3866, + "mean_token_accuracy": 0.8710423707962036, + "num_tokens": 275794637.0, + "step": 7104 + }, + { + "epoch": 0.9038290293855743, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.441065788269043, + "learning_rate": 1e-06, + "loss": 0.4187, + "mean_token_accuracy": 0.8618312478065491, + "num_tokens": 275833259.0, + "step": 7105 + }, + { + "epoch": 0.9039562396641648, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.428290367126465, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8731785416603088, + "num_tokens": 275876607.0, + "step": 7106 + }, + { + "epoch": 0.9040834499427554, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.361096382141113, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8985203504562378, + "num_tokens": 275911617.0, + "step": 7107 + }, + { + "epoch": 0.9042106602213459, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.4262056350708, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8819262981414795, + "num_tokens": 275949000.0, + "step": 7108 + }, + { + "epoch": 0.9043378704999364, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.42529582977295, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.879028856754303, + "num_tokens": 275984934.0, + "step": 7109 + }, + { + "epoch": 0.9044650807785269, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.34831714630127, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.884904682636261, + "num_tokens": 276019185.0, + "step": 7110 + }, + { + "epoch": 0.9045922910571175, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.36899471282959, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8890508413314819, + "num_tokens": 276060429.0, + "step": 7111 + }, + { + "epoch": 0.9047195013357079, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.432531356811523, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8825961351394653, + "num_tokens": 276096544.0, + "step": 7112 + }, + { + "epoch": 0.9048467116142984, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.500473976135254, + "learning_rate": 1e-06, + "loss": 0.4172, + "mean_token_accuracy": 0.8616306781768799, + "num_tokens": 276132663.0, + "step": 7113 + }, + { + "epoch": 0.9049739218928889, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.377321243286133, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8863744735717773, + "num_tokens": 276169677.0, + "step": 7114 + }, + { + "epoch": 0.9051011321714795, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.361038208007812, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8721333742141724, + "num_tokens": 276216196.0, + "step": 7115 + }, + { + "epoch": 0.90522834245007, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.450828552246094, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8848457336425781, + "num_tokens": 276250760.0, + "step": 7116 + }, + { + "epoch": 0.9053555527286605, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.402979850769043, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8814696669578552, + "num_tokens": 276284238.0, + "step": 7117 + }, + { + "epoch": 0.905482763007251, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.509007453918457, + "learning_rate": 1e-06, + "loss": 0.4034, + "mean_token_accuracy": 0.8644708395004272, + "num_tokens": 276324845.0, + "step": 7118 + }, + { + "epoch": 0.9056099732858415, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.298203468322754, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8761253356933594, + "num_tokens": 276368651.0, + "step": 7119 + }, + { + "epoch": 0.905737183564432, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.515015602111816, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.870880663394928, + "num_tokens": 276401260.0, + "step": 7120 + }, + { + "epoch": 0.9058643938430225, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.34604263305664, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8839582204818726, + "num_tokens": 276440558.0, + "step": 7121 + }, + { + "epoch": 0.9059916041216131, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.369013786315918, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8836082816123962, + "num_tokens": 276477897.0, + "step": 7122 + }, + { + "epoch": 0.9061188144002036, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.454453468322754, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8716306686401367, + "num_tokens": 276519301.0, + "step": 7123 + }, + { + "epoch": 0.906246024678794, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.399069786071777, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8834271430969238, + "num_tokens": 276555053.0, + "step": 7124 + }, + { + "epoch": 0.9063732349573845, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.42123794555664, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.8710180521011353, + "num_tokens": 276595161.0, + "step": 7125 + }, + { + "epoch": 0.9065004452359751, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.400775909423828, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8828684091567993, + "num_tokens": 276634814.0, + "step": 7126 + }, + { + "epoch": 0.9066276555145656, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.436530113220215, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8871033787727356, + "num_tokens": 276670245.0, + "step": 7127 + }, + { + "epoch": 0.9067548657931561, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.482842445373535, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8835352659225464, + "num_tokens": 276704646.0, + "step": 7128 + }, + { + "epoch": 0.9068820760717466, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.342670440673828, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8792782425880432, + "num_tokens": 276743406.0, + "step": 7129 + }, + { + "epoch": 0.9070092863503371, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.462347030639648, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8792903423309326, + "num_tokens": 276779855.0, + "step": 7130 + }, + { + "epoch": 0.9071364966289276, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.356374740600586, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8873715996742249, + "num_tokens": 276818346.0, + "step": 7131 + }, + { + "epoch": 0.9072637069075181, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.356364250183105, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8870936036109924, + "num_tokens": 276858519.0, + "step": 7132 + }, + { + "epoch": 0.9073909171861086, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.367286682128906, + "learning_rate": 1e-06, + "loss": 0.3681, + "mean_token_accuracy": 0.8718267679214478, + "num_tokens": 276898522.0, + "step": 7133 + }, + { + "epoch": 0.9075181274646992, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.415947914123535, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8775111436843872, + "num_tokens": 276934184.0, + "step": 7134 + }, + { + "epoch": 0.9076453377432897, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.452223777770996, + "learning_rate": 1e-06, + "loss": 0.4018, + "mean_token_accuracy": 0.8668547868728638, + "num_tokens": 276974519.0, + "step": 7135 + }, + { + "epoch": 0.9077725480218801, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.396515846252441, + "learning_rate": 1e-06, + "loss": 0.3779, + "mean_token_accuracy": 0.870137095451355, + "num_tokens": 277014659.0, + "step": 7136 + }, + { + "epoch": 0.9078997583004706, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.419926643371582, + "learning_rate": 1e-06, + "loss": 0.3774, + "mean_token_accuracy": 0.873950719833374, + "num_tokens": 277053416.0, + "step": 7137 + }, + { + "epoch": 0.9080269685790612, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.264838218688965, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.882868230342865, + "num_tokens": 277098720.0, + "step": 7138 + }, + { + "epoch": 0.9081541788576517, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.448554992675781, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8785821795463562, + "num_tokens": 277135461.0, + "step": 7139 + }, + { + "epoch": 0.9082813891362422, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.437162399291992, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8797886967658997, + "num_tokens": 277167759.0, + "step": 7140 + }, + { + "epoch": 0.9084085994148328, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.437639236450195, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8899543285369873, + "num_tokens": 277211113.0, + "step": 7141 + }, + { + "epoch": 0.9085358096934232, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.373125076293945, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8754369020462036, + "num_tokens": 277250327.0, + "step": 7142 + }, + { + "epoch": 0.9086630199720137, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.36803913116455, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8941296339035034, + "num_tokens": 277291381.0, + "step": 7143 + }, + { + "epoch": 0.9087902302506042, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.450417518615723, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8833374977111816, + "num_tokens": 277326807.0, + "step": 7144 + }, + { + "epoch": 0.9089174405291948, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.432833671569824, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8790451288223267, + "num_tokens": 277366992.0, + "step": 7145 + }, + { + "epoch": 0.9090446508077853, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.422030448913574, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8808845281600952, + "num_tokens": 277404446.0, + "step": 7146 + }, + { + "epoch": 0.9091718610863758, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.31159496307373, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8804289698600769, + "num_tokens": 277444641.0, + "step": 7147 + }, + { + "epoch": 0.9092990713649662, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.371345520019531, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8805802464485168, + "num_tokens": 277484899.0, + "step": 7148 + }, + { + "epoch": 0.9094262816435568, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.398197174072266, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.8691789507865906, + "num_tokens": 277527705.0, + "step": 7149 + }, + { + "epoch": 0.9095534919221473, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.429277420043945, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8761540651321411, + "num_tokens": 277565698.0, + "step": 7150 + }, + { + "epoch": 0.9096807022007378, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.367576599121094, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8803797960281372, + "num_tokens": 277605544.0, + "step": 7151 + }, + { + "epoch": 0.9098079124793284, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.428983688354492, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8726191520690918, + "num_tokens": 277641156.0, + "step": 7152 + }, + { + "epoch": 0.9099351227579189, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.411761283874512, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8839694261550903, + "num_tokens": 277679891.0, + "step": 7153 + }, + { + "epoch": 0.9100623330365093, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.403528213500977, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8825992345809937, + "num_tokens": 277715276.0, + "step": 7154 + }, + { + "epoch": 0.9101895433150998, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.344675064086914, + "learning_rate": 1e-06, + "loss": 0.3974, + "mean_token_accuracy": 0.8684627413749695, + "num_tokens": 277758504.0, + "step": 7155 + }, + { + "epoch": 0.9103167535936904, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.41533088684082, + "learning_rate": 1e-06, + "loss": 0.4, + "mean_token_accuracy": 0.864648699760437, + "num_tokens": 277797544.0, + "step": 7156 + }, + { + "epoch": 0.9104439638722809, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.426411628723145, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8929362297058105, + "num_tokens": 277834109.0, + "step": 7157 + }, + { + "epoch": 0.9105711741508714, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.36314582824707, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8895262479782104, + "num_tokens": 277871609.0, + "step": 7158 + }, + { + "epoch": 0.9106983844294619, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.589232444763184, + "learning_rate": 1e-06, + "loss": 0.438, + "mean_token_accuracy": 0.854392409324646, + "num_tokens": 277906779.0, + "step": 7159 + }, + { + "epoch": 0.9108255947080524, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.461177825927734, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.8732278347015381, + "num_tokens": 277943930.0, + "step": 7160 + }, + { + "epoch": 0.9109528049866429, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.370030403137207, + "learning_rate": 1e-06, + "loss": 0.3865, + "mean_token_accuracy": 0.865770697593689, + "num_tokens": 277985058.0, + "step": 7161 + }, + { + "epoch": 0.9110800152652334, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.37996768951416, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8834689855575562, + "num_tokens": 278028465.0, + "step": 7162 + }, + { + "epoch": 0.9112072255438239, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.33182144165039, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8861221671104431, + "num_tokens": 278072534.0, + "step": 7163 + }, + { + "epoch": 0.9113344358224145, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.400247573852539, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8710856437683105, + "num_tokens": 278112428.0, + "step": 7164 + }, + { + "epoch": 0.911461646101005, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.393998146057129, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8836390972137451, + "num_tokens": 278150793.0, + "step": 7165 + }, + { + "epoch": 0.9115888563795955, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.404027938842773, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8754074573516846, + "num_tokens": 278190772.0, + "step": 7166 + }, + { + "epoch": 0.9117160666581859, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.345605850219727, + "learning_rate": 1e-06, + "loss": 0.3631, + "mean_token_accuracy": 0.8787890672683716, + "num_tokens": 278236192.0, + "step": 7167 + }, + { + "epoch": 0.9118432769367765, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.391895294189453, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8765560984611511, + "num_tokens": 278282689.0, + "step": 7168 + }, + { + "epoch": 0.911970487215367, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.41154670715332, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.88599693775177, + "num_tokens": 278319600.0, + "step": 7169 + }, + { + "epoch": 0.9120976974939575, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.491984367370605, + "learning_rate": 1e-06, + "loss": 0.3838, + "mean_token_accuracy": 0.8728134632110596, + "num_tokens": 278359142.0, + "step": 7170 + }, + { + "epoch": 0.912224907772548, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.498903274536133, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8829257488250732, + "num_tokens": 278391241.0, + "step": 7171 + }, + { + "epoch": 0.9123521180511386, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.525531768798828, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8713437914848328, + "num_tokens": 278427489.0, + "step": 7172 + }, + { + "epoch": 0.912479328329729, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.494429588317871, + "learning_rate": 1e-06, + "loss": 0.4061, + "mean_token_accuracy": 0.8645877838134766, + "num_tokens": 278459912.0, + "step": 7173 + }, + { + "epoch": 0.9126065386083195, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.544002532958984, + "learning_rate": 1e-06, + "loss": 0.3969, + "mean_token_accuracy": 0.866415798664093, + "num_tokens": 278492311.0, + "step": 7174 + }, + { + "epoch": 0.9127337488869101, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.382563591003418, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.893002450466156, + "num_tokens": 278528548.0, + "step": 7175 + }, + { + "epoch": 0.9128609591655006, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.450006484985352, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.881493866443634, + "num_tokens": 278570224.0, + "step": 7176 + }, + { + "epoch": 0.9129881694440911, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.404666900634766, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8767349720001221, + "num_tokens": 278606215.0, + "step": 7177 + }, + { + "epoch": 0.9131153797226816, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.568799018859863, + "learning_rate": 1e-06, + "loss": 0.395, + "mean_token_accuracy": 0.8676088452339172, + "num_tokens": 278637217.0, + "step": 7178 + }, + { + "epoch": 0.9132425900012721, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.388203620910645, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8872394561767578, + "num_tokens": 278676010.0, + "step": 7179 + }, + { + "epoch": 0.9133698002798626, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.45805835723877, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.882157027721405, + "num_tokens": 278710606.0, + "step": 7180 + }, + { + "epoch": 0.9134970105584531, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.452055931091309, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8873085975646973, + "num_tokens": 278750289.0, + "step": 7181 + }, + { + "epoch": 0.9136242208370436, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.457454681396484, + "learning_rate": 1e-06, + "loss": 0.3671, + "mean_token_accuracy": 0.8766477108001709, + "num_tokens": 278788473.0, + "step": 7182 + }, + { + "epoch": 0.9137514311156342, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.509784698486328, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8741562366485596, + "num_tokens": 278819379.0, + "step": 7183 + }, + { + "epoch": 0.9138786413942247, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.464492797851562, + "learning_rate": 1e-06, + "loss": 0.3955, + "mean_token_accuracy": 0.8675094842910767, + "num_tokens": 278861718.0, + "step": 7184 + }, + { + "epoch": 0.9140058516728151, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.434732437133789, + "learning_rate": 1e-06, + "loss": 0.3892, + "mean_token_accuracy": 0.8662135004997253, + "num_tokens": 278903955.0, + "step": 7185 + }, + { + "epoch": 0.9141330619514056, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.404542922973633, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8821709752082825, + "num_tokens": 278947723.0, + "step": 7186 + }, + { + "epoch": 0.9142602722299962, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.465794563293457, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8792805671691895, + "num_tokens": 278984469.0, + "step": 7187 + }, + { + "epoch": 0.9143874825085867, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.498035430908203, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8797917366027832, + "num_tokens": 279019025.0, + "step": 7188 + }, + { + "epoch": 0.9145146927871772, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.48498821258545, + "learning_rate": 1e-06, + "loss": 0.3988, + "mean_token_accuracy": 0.8650835156440735, + "num_tokens": 279054513.0, + "step": 7189 + }, + { + "epoch": 0.9146419030657678, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.474987030029297, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8827980756759644, + "num_tokens": 279088913.0, + "step": 7190 + }, + { + "epoch": 0.9147691133443582, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.468057632446289, + "learning_rate": 1e-06, + "loss": 0.3771, + "mean_token_accuracy": 0.8726059198379517, + "num_tokens": 279126489.0, + "step": 7191 + }, + { + "epoch": 0.9148963236229487, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.433894157409668, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8838821649551392, + "num_tokens": 279167485.0, + "step": 7192 + }, + { + "epoch": 0.9150235339015392, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.470833778381348, + "learning_rate": 1e-06, + "loss": 0.4161, + "mean_token_accuracy": 0.8588817119598389, + "num_tokens": 279204525.0, + "step": 7193 + }, + { + "epoch": 0.9151507441801298, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.395082473754883, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8885133266448975, + "num_tokens": 279246372.0, + "step": 7194 + }, + { + "epoch": 0.9152779544587203, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.452984809875488, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8692994117736816, + "num_tokens": 279286543.0, + "step": 7195 + }, + { + "epoch": 0.9154051647373108, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.49701976776123, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8837202787399292, + "num_tokens": 279322115.0, + "step": 7196 + }, + { + "epoch": 0.9155323750159012, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.3721342086792, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8873282670974731, + "num_tokens": 279360161.0, + "step": 7197 + }, + { + "epoch": 0.9156595852944918, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.507073402404785, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.8702424764633179, + "num_tokens": 279392136.0, + "step": 7198 + }, + { + "epoch": 0.9157867955730823, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.400259017944336, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8840339183807373, + "num_tokens": 279433046.0, + "step": 7199 + }, + { + "epoch": 0.9159140058516728, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.683403968811035, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8780546188354492, + "num_tokens": 279474601.0, + "step": 7200 + }, + { + "epoch": 0.9160412161302633, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.41650390625, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8860390782356262, + "num_tokens": 279511641.0, + "step": 7201 + }, + { + "epoch": 0.9161684264088539, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.39267635345459, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8828732967376709, + "num_tokens": 279547718.0, + "step": 7202 + }, + { + "epoch": 0.9162956366874443, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.499127388000488, + "learning_rate": 1e-06, + "loss": 0.368, + "mean_token_accuracy": 0.8701107501983643, + "num_tokens": 279584069.0, + "step": 7203 + }, + { + "epoch": 0.9164228469660348, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.34894847869873, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8874374032020569, + "num_tokens": 279623084.0, + "step": 7204 + }, + { + "epoch": 0.9165500572446253, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.384442329406738, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8773993253707886, + "num_tokens": 279664965.0, + "step": 7205 + }, + { + "epoch": 0.9166772675232159, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.434496879577637, + "learning_rate": 1e-06, + "loss": 0.402, + "mean_token_accuracy": 0.8656418323516846, + "num_tokens": 279704351.0, + "step": 7206 + }, + { + "epoch": 0.9168044778018064, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.540365219116211, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8691555261611938, + "num_tokens": 279738831.0, + "step": 7207 + }, + { + "epoch": 0.9169316880803969, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.39232349395752, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.8771867752075195, + "num_tokens": 279778883.0, + "step": 7208 + }, + { + "epoch": 0.9170588983589874, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.312728881835938, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8861948251724243, + "num_tokens": 279823930.0, + "step": 7209 + }, + { + "epoch": 0.9171861086375779, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.432456016540527, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.881472110748291, + "num_tokens": 279859800.0, + "step": 7210 + }, + { + "epoch": 0.9173133189161684, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.392610549926758, + "learning_rate": 1e-06, + "loss": 0.3797, + "mean_token_accuracy": 0.8718394041061401, + "num_tokens": 279900830.0, + "step": 7211 + }, + { + "epoch": 0.9174405291947589, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.355663299560547, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.874846339225769, + "num_tokens": 279942594.0, + "step": 7212 + }, + { + "epoch": 0.9175677394733495, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.478053092956543, + "learning_rate": 1e-06, + "loss": 0.3953, + "mean_token_accuracy": 0.8647889494895935, + "num_tokens": 279979645.0, + "step": 7213 + }, + { + "epoch": 0.91769494975194, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.392742156982422, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8809177279472351, + "num_tokens": 280022261.0, + "step": 7214 + }, + { + "epoch": 0.9178221600305305, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.475911140441895, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8775511980056763, + "num_tokens": 280054496.0, + "step": 7215 + }, + { + "epoch": 0.9179493703091209, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.5105619430542, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.8699358105659485, + "num_tokens": 280090658.0, + "step": 7216 + }, + { + "epoch": 0.9180765805877115, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.361135482788086, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8816697001457214, + "num_tokens": 280130380.0, + "step": 7217 + }, + { + "epoch": 0.918203790866302, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.435275077819824, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8827075958251953, + "num_tokens": 280165872.0, + "step": 7218 + }, + { + "epoch": 0.9183310011448925, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.436319351196289, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8917661905288696, + "num_tokens": 280201234.0, + "step": 7219 + }, + { + "epoch": 0.918458211423483, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.425535202026367, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8885861039161682, + "num_tokens": 280233031.0, + "step": 7220 + }, + { + "epoch": 0.9185854217020736, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.436659812927246, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.8847975134849548, + "num_tokens": 280272189.0, + "step": 7221 + }, + { + "epoch": 0.918712631980664, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.333710670471191, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8795079588890076, + "num_tokens": 280317450.0, + "step": 7222 + }, + { + "epoch": 0.9188398422592545, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.5343656539917, + "learning_rate": 1e-06, + "loss": 0.391, + "mean_token_accuracy": 0.868503212928772, + "num_tokens": 280352372.0, + "step": 7223 + }, + { + "epoch": 0.918967052537845, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.386462211608887, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8815287947654724, + "num_tokens": 280393733.0, + "step": 7224 + }, + { + "epoch": 0.9190942628164356, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.410181999206543, + "learning_rate": 1e-06, + "loss": 0.3804, + "mean_token_accuracy": 0.8725995421409607, + "num_tokens": 280438669.0, + "step": 7225 + }, + { + "epoch": 0.9192214730950261, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.434940338134766, + "learning_rate": 1e-06, + "loss": 0.4319, + "mean_token_accuracy": 0.8552270531654358, + "num_tokens": 280479823.0, + "step": 7226 + }, + { + "epoch": 0.9193486833736166, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.445743560791016, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.870075523853302, + "num_tokens": 280518786.0, + "step": 7227 + }, + { + "epoch": 0.919475893652207, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.382936477661133, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8777154088020325, + "num_tokens": 280558576.0, + "step": 7228 + }, + { + "epoch": 0.9196031039307976, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.53020191192627, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8793612718582153, + "num_tokens": 280590432.0, + "step": 7229 + }, + { + "epoch": 0.9197303142093881, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.357620239257812, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8953449726104736, + "num_tokens": 280631968.0, + "step": 7230 + }, + { + "epoch": 0.9198575244879786, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.416471481323242, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8715792894363403, + "num_tokens": 280669074.0, + "step": 7231 + }, + { + "epoch": 0.9199847347665692, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.356999397277832, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.877318263053894, + "num_tokens": 280708797.0, + "step": 7232 + }, + { + "epoch": 0.9201119450451597, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.422567367553711, + "learning_rate": 1e-06, + "loss": 0.3869, + "mean_token_accuracy": 0.8717081546783447, + "num_tokens": 280750205.0, + "step": 7233 + }, + { + "epoch": 0.9202391553237501, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.523213386535645, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8820338845252991, + "num_tokens": 280782362.0, + "step": 7234 + }, + { + "epoch": 0.9203663656023406, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.395535469055176, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8850327134132385, + "num_tokens": 280820036.0, + "step": 7235 + }, + { + "epoch": 0.9204935758809312, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.342613220214844, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8948806524276733, + "num_tokens": 280855516.0, + "step": 7236 + }, + { + "epoch": 0.9206207861595217, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.429957389831543, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8793128132820129, + "num_tokens": 280892133.0, + "step": 7237 + }, + { + "epoch": 0.9207479964381122, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.450437545776367, + "learning_rate": 1e-06, + "loss": 0.3847, + "mean_token_accuracy": 0.8708864450454712, + "num_tokens": 280931553.0, + "step": 7238 + }, + { + "epoch": 0.9208752067167028, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.336430549621582, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.871317982673645, + "num_tokens": 280978328.0, + "step": 7239 + }, + { + "epoch": 0.9210024169952932, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.432156562805176, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.867276668548584, + "num_tokens": 281013162.0, + "step": 7240 + }, + { + "epoch": 0.9211296272738837, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.347769737243652, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9003047347068787, + "num_tokens": 281052845.0, + "step": 7241 + }, + { + "epoch": 0.9212568375524742, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.413980484008789, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8794310092926025, + "num_tokens": 281091228.0, + "step": 7242 + }, + { + "epoch": 0.9213840478310648, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.394354820251465, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8795080184936523, + "num_tokens": 281132801.0, + "step": 7243 + }, + { + "epoch": 0.9215112581096553, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.40025806427002, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.87948077917099, + "num_tokens": 281169965.0, + "step": 7244 + }, + { + "epoch": 0.9216384683882458, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.381290435791016, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8716904520988464, + "num_tokens": 281210341.0, + "step": 7245 + }, + { + "epoch": 0.9217656786668362, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.404740333557129, + "learning_rate": 1e-06, + "loss": 0.3804, + "mean_token_accuracy": 0.8720757961273193, + "num_tokens": 281254586.0, + "step": 7246 + }, + { + "epoch": 0.9218928889454268, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.434271812438965, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8760441541671753, + "num_tokens": 281291584.0, + "step": 7247 + }, + { + "epoch": 0.9220200992240173, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.369782447814941, + "learning_rate": 1e-06, + "loss": 0.3676, + "mean_token_accuracy": 0.8754380345344543, + "num_tokens": 281333510.0, + "step": 7248 + }, + { + "epoch": 0.9221473095026078, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.401198387145996, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8839854001998901, + "num_tokens": 281372933.0, + "step": 7249 + }, + { + "epoch": 0.9222745197811983, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.439594268798828, + "learning_rate": 1e-06, + "loss": 0.3825, + "mean_token_accuracy": 0.8703927397727966, + "num_tokens": 281410935.0, + "step": 7250 + }, + { + "epoch": 0.9224017300597889, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.459835052490234, + "learning_rate": 1e-06, + "loss": 0.4005, + "mean_token_accuracy": 0.8658896684646606, + "num_tokens": 281451043.0, + "step": 7251 + }, + { + "epoch": 0.9225289403383793, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.44814682006836, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8789417743682861, + "num_tokens": 281489327.0, + "step": 7252 + }, + { + "epoch": 0.9226561506169698, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.377799034118652, + "learning_rate": 1e-06, + "loss": 0.3804, + "mean_token_accuracy": 0.8739824891090393, + "num_tokens": 281534896.0, + "step": 7253 + }, + { + "epoch": 0.9227833608955603, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.411160469055176, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.87566739320755, + "num_tokens": 281577310.0, + "step": 7254 + }, + { + "epoch": 0.9229105711741509, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.448391914367676, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8713923692703247, + "num_tokens": 281613643.0, + "step": 7255 + }, + { + "epoch": 0.9230377814527414, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.369750022888184, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8784486651420593, + "num_tokens": 281652162.0, + "step": 7256 + }, + { + "epoch": 0.9231649917313319, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.366654396057129, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8801673054695129, + "num_tokens": 281697148.0, + "step": 7257 + }, + { + "epoch": 0.9232922020099223, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.399073600769043, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8841363191604614, + "num_tokens": 281739738.0, + "step": 7258 + }, + { + "epoch": 0.9234194122885129, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.438408851623535, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8718729019165039, + "num_tokens": 281777282.0, + "step": 7259 + }, + { + "epoch": 0.9235466225671034, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.39578628540039, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8833515644073486, + "num_tokens": 281813179.0, + "step": 7260 + }, + { + "epoch": 0.9236738328456939, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.430559158325195, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8824623227119446, + "num_tokens": 281853386.0, + "step": 7261 + }, + { + "epoch": 0.9238010431242845, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.404741287231445, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8874425888061523, + "num_tokens": 281890018.0, + "step": 7262 + }, + { + "epoch": 0.923928253402875, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.47512435913086, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8746505975723267, + "num_tokens": 281931167.0, + "step": 7263 + }, + { + "epoch": 0.9240554636814655, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.414859771728516, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8731566667556763, + "num_tokens": 281975072.0, + "step": 7264 + }, + { + "epoch": 0.9241826739600559, + "ewc_loss": 0.01226806640625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.119510650634766e-06, + "grad_norm": 8.349122047424316, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8894912600517273, + "num_tokens": 282015791.0, + "step": 7265 + }, + { + "epoch": 0.9243098842386465, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.56234073638916, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8804409503936768, + "num_tokens": 282049082.0, + "step": 7266 + }, + { + "epoch": 0.924437094517237, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.408353805541992, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8749033212661743, + "num_tokens": 282088826.0, + "step": 7267 + }, + { + "epoch": 0.9245643047958275, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.38204288482666, + "learning_rate": 1e-06, + "loss": 0.368, + "mean_token_accuracy": 0.8731274008750916, + "num_tokens": 282131078.0, + "step": 7268 + }, + { + "epoch": 0.924691515074418, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.46687126159668, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8708150386810303, + "num_tokens": 282167183.0, + "step": 7269 + }, + { + "epoch": 0.9248187253530086, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.384689331054688, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8755073547363281, + "num_tokens": 282208507.0, + "step": 7270 + }, + { + "epoch": 0.924945935631599, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.443236351013184, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8845934867858887, + "num_tokens": 282248883.0, + "step": 7271 + }, + { + "epoch": 0.9250731459101895, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.41886043548584, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8821693062782288, + "num_tokens": 282286129.0, + "step": 7272 + }, + { + "epoch": 0.92520035618878, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.426152229309082, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8740674257278442, + "num_tokens": 282326795.0, + "step": 7273 + }, + { + "epoch": 0.9253275664673706, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.375382423400879, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8859080672264099, + "num_tokens": 282370449.0, + "step": 7274 + }, + { + "epoch": 0.9254547767459611, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.375375747680664, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8786709308624268, + "num_tokens": 282414101.0, + "step": 7275 + }, + { + "epoch": 0.9255819870245516, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.433392524719238, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8918548822402954, + "num_tokens": 282449237.0, + "step": 7276 + }, + { + "epoch": 0.925709197303142, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.405214309692383, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8873941898345947, + "num_tokens": 282491526.0, + "step": 7277 + }, + { + "epoch": 0.9258364075817326, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.459661483764648, + "learning_rate": 1e-06, + "loss": 0.3736, + "mean_token_accuracy": 0.8732467889785767, + "num_tokens": 282531585.0, + "step": 7278 + }, + { + "epoch": 0.9259636178603231, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.519009590148926, + "learning_rate": 1e-06, + "loss": 0.3729, + "mean_token_accuracy": 0.8755660057067871, + "num_tokens": 282568883.0, + "step": 7279 + }, + { + "epoch": 0.9260908281389136, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.489657402038574, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8744828701019287, + "num_tokens": 282603584.0, + "step": 7280 + }, + { + "epoch": 0.9262180384175042, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.539400100708008, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8786125183105469, + "num_tokens": 282634175.0, + "step": 7281 + }, + { + "epoch": 0.9263452486960947, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.405055046081543, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8748337030410767, + "num_tokens": 282671571.0, + "step": 7282 + }, + { + "epoch": 0.9264724589746851, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.3863525390625, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8788747787475586, + "num_tokens": 282708683.0, + "step": 7283 + }, + { + "epoch": 0.9265996692532756, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.39682674407959, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8832732439041138, + "num_tokens": 282743883.0, + "step": 7284 + }, + { + "epoch": 0.9267268795318662, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.31033992767334, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.891310453414917, + "num_tokens": 282784641.0, + "step": 7285 + }, + { + "epoch": 0.9268540898104567, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.41275405883789, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8689398765563965, + "num_tokens": 282826486.0, + "step": 7286 + }, + { + "epoch": 0.9269813000890472, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.31663703918457, + "learning_rate": 1e-06, + "loss": 0.3847, + "mean_token_accuracy": 0.8714313507080078, + "num_tokens": 282873275.0, + "step": 7287 + }, + { + "epoch": 0.9271085103676378, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.425686836242676, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8923161029815674, + "num_tokens": 282910333.0, + "step": 7288 + }, + { + "epoch": 0.9272357206462282, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.468971252441406, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8804506063461304, + "num_tokens": 282949229.0, + "step": 7289 + }, + { + "epoch": 0.9273629309248187, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.450081825256348, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8790552020072937, + "num_tokens": 282984929.0, + "step": 7290 + }, + { + "epoch": 0.9274901412034092, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.41688060760498, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.877657413482666, + "num_tokens": 283021872.0, + "step": 7291 + }, + { + "epoch": 0.9276173514819998, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.413406372070312, + "learning_rate": 1e-06, + "loss": 0.4219, + "mean_token_accuracy": 0.8580774068832397, + "num_tokens": 283067540.0, + "step": 7292 + }, + { + "epoch": 0.9277445617605903, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.419242858886719, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8833597302436829, + "num_tokens": 283105246.0, + "step": 7293 + }, + { + "epoch": 0.9278717720391808, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.397527694702148, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8973737955093384, + "num_tokens": 283138869.0, + "step": 7294 + }, + { + "epoch": 0.9279989823177712, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.45407485961914, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.8690098524093628, + "num_tokens": 283176988.0, + "step": 7295 + }, + { + "epoch": 0.9281261925963618, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.372645378112793, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8855956196784973, + "num_tokens": 283220638.0, + "step": 7296 + }, + { + "epoch": 0.9282534028749523, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.417393684387207, + "learning_rate": 1e-06, + "loss": 0.3962, + "mean_token_accuracy": 0.8738676309585571, + "num_tokens": 283263665.0, + "step": 7297 + }, + { + "epoch": 0.9283806131535428, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.524238586425781, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.880432665348053, + "num_tokens": 283295757.0, + "step": 7298 + }, + { + "epoch": 0.9285078234321333, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.403046607971191, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8827508091926575, + "num_tokens": 283334365.0, + "step": 7299 + }, + { + "epoch": 0.9286350337107239, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.4885835647583, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8771395087242126, + "num_tokens": 283372663.0, + "step": 7300 + }, + { + "epoch": 0.9287622439893143, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.366491317749023, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8952445983886719, + "num_tokens": 283411315.0, + "step": 7301 + }, + { + "epoch": 0.9288894542679048, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.448410987854004, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8857713341712952, + "num_tokens": 283451489.0, + "step": 7302 + }, + { + "epoch": 0.9290166645464953, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.327077865600586, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.886372447013855, + "num_tokens": 283494286.0, + "step": 7303 + }, + { + "epoch": 0.9291438748250859, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.561240196228027, + "learning_rate": 1e-06, + "loss": 0.3949, + "mean_token_accuracy": 0.8679924011230469, + "num_tokens": 283532892.0, + "step": 7304 + }, + { + "epoch": 0.9292710851036764, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.419499397277832, + "learning_rate": 1e-06, + "loss": 0.3946, + "mean_token_accuracy": 0.8669012784957886, + "num_tokens": 283578707.0, + "step": 7305 + }, + { + "epoch": 0.9293982953822669, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.432961463928223, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8813440799713135, + "num_tokens": 283613219.0, + "step": 7306 + }, + { + "epoch": 0.9295255056608573, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.589290618896484, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8808293342590332, + "num_tokens": 283641646.0, + "step": 7307 + }, + { + "epoch": 0.9296527159394479, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.297365188598633, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8884537816047668, + "num_tokens": 283686546.0, + "step": 7308 + }, + { + "epoch": 0.9297799262180384, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.514543533325195, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8796499371528625, + "num_tokens": 283724446.0, + "step": 7309 + }, + { + "epoch": 0.9299071364966289, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.424498558044434, + "learning_rate": 1e-06, + "loss": 0.3832, + "mean_token_accuracy": 0.8690972328186035, + "num_tokens": 283770718.0, + "step": 7310 + }, + { + "epoch": 0.9300343467752195, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.457673072814941, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8920891880989075, + "num_tokens": 283801207.0, + "step": 7311 + }, + { + "epoch": 0.93016155705381, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.560468673706055, + "learning_rate": 1e-06, + "loss": 0.3997, + "mean_token_accuracy": 0.8677917122840881, + "num_tokens": 283835799.0, + "step": 7312 + }, + { + "epoch": 0.9302887673324005, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.433558464050293, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8862118721008301, + "num_tokens": 283875707.0, + "step": 7313 + }, + { + "epoch": 0.9304159776109909, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.543156623840332, + "learning_rate": 1e-06, + "loss": 0.3868, + "mean_token_accuracy": 0.8680370450019836, + "num_tokens": 283910424.0, + "step": 7314 + }, + { + "epoch": 0.9305431878895815, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.67495059967041, + "learning_rate": 1e-06, + "loss": 0.4052, + "mean_token_accuracy": 0.8637117147445679, + "num_tokens": 283943628.0, + "step": 7315 + }, + { + "epoch": 0.930670398168172, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.40324878692627, + "learning_rate": 1e-06, + "loss": 0.3992, + "mean_token_accuracy": 0.8660987615585327, + "num_tokens": 283992048.0, + "step": 7316 + }, + { + "epoch": 0.9307976084467625, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.40418529510498, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8798873424530029, + "num_tokens": 284032584.0, + "step": 7317 + }, + { + "epoch": 0.930924818725353, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.42257308959961, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8898029327392578, + "num_tokens": 284077301.0, + "step": 7318 + }, + { + "epoch": 0.9310520290039436, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.35726547241211, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8826934099197388, + "num_tokens": 284122342.0, + "step": 7319 + }, + { + "epoch": 0.931179239282534, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.5059175491333, + "learning_rate": 1e-06, + "loss": 0.3884, + "mean_token_accuracy": 0.8678133487701416, + "num_tokens": 284164798.0, + "step": 7320 + }, + { + "epoch": 0.9313064495611245, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.361638069152832, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8855528235435486, + "num_tokens": 284207375.0, + "step": 7321 + }, + { + "epoch": 0.931433659839715, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.560553550720215, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8826863765716553, + "num_tokens": 284244379.0, + "step": 7322 + }, + { + "epoch": 0.9315608701183056, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.384129524230957, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8921582102775574, + "num_tokens": 284285875.0, + "step": 7323 + }, + { + "epoch": 0.9316880803968961, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.435338973999023, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8845049738883972, + "num_tokens": 284322877.0, + "step": 7324 + }, + { + "epoch": 0.9318152906754866, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.49386215209961, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.876125693321228, + "num_tokens": 284360002.0, + "step": 7325 + }, + { + "epoch": 0.931942500954077, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.452845573425293, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8800369501113892, + "num_tokens": 284394902.0, + "step": 7326 + }, + { + "epoch": 0.9320697112326676, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.416463851928711, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8824100494384766, + "num_tokens": 284434050.0, + "step": 7327 + }, + { + "epoch": 0.9321969215112581, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.454286575317383, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8695660829544067, + "num_tokens": 284475330.0, + "step": 7328 + }, + { + "epoch": 0.9323241317898486, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.387479782104492, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8883290886878967, + "num_tokens": 284511305.0, + "step": 7329 + }, + { + "epoch": 0.9324513420684392, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.407957077026367, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8722280263900757, + "num_tokens": 284553444.0, + "step": 7330 + }, + { + "epoch": 0.9325785523470297, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.449238777160645, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8807377815246582, + "num_tokens": 284588354.0, + "step": 7331 + }, + { + "epoch": 0.9327057626256201, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.43934440612793, + "learning_rate": 1e-06, + "loss": 0.3809, + "mean_token_accuracy": 0.8705205321311951, + "num_tokens": 284630432.0, + "step": 7332 + }, + { + "epoch": 0.9328329729042106, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.430338859558105, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8840530514717102, + "num_tokens": 284664933.0, + "step": 7333 + }, + { + "epoch": 0.9329601831828012, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.50959300994873, + "learning_rate": 1e-06, + "loss": 0.4051, + "mean_token_accuracy": 0.862971305847168, + "num_tokens": 284699584.0, + "step": 7334 + }, + { + "epoch": 0.9330873934613917, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.45002269744873, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8901447653770447, + "num_tokens": 284737881.0, + "step": 7335 + }, + { + "epoch": 0.9332146037399822, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.453387260437012, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.8741494417190552, + "num_tokens": 284776568.0, + "step": 7336 + }, + { + "epoch": 0.9333418140185727, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.537786483764648, + "learning_rate": 1e-06, + "loss": 0.3906, + "mean_token_accuracy": 0.8659347891807556, + "num_tokens": 284808689.0, + "step": 7337 + }, + { + "epoch": 0.9334690242971632, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.548821449279785, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.8712028861045837, + "num_tokens": 284841516.0, + "step": 7338 + }, + { + "epoch": 0.9335962345757537, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.403046607971191, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8824518322944641, + "num_tokens": 284878357.0, + "step": 7339 + }, + { + "epoch": 0.9337234448543442, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.592291831970215, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.880160391330719, + "num_tokens": 284908201.0, + "step": 7340 + }, + { + "epoch": 0.9338506551329347, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.424149513244629, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8786064386367798, + "num_tokens": 284949436.0, + "step": 7341 + }, + { + "epoch": 0.9339778654115253, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.430930137634277, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8820664882659912, + "num_tokens": 284989770.0, + "step": 7342 + }, + { + "epoch": 0.9341050756901158, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.487887382507324, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8895277976989746, + "num_tokens": 285024085.0, + "step": 7343 + }, + { + "epoch": 0.9342322859687062, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.495296478271484, + "learning_rate": 1e-06, + "loss": 0.3879, + "mean_token_accuracy": 0.8658565282821655, + "num_tokens": 285061196.0, + "step": 7344 + }, + { + "epoch": 0.9343594962472968, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.459405899047852, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8858668208122253, + "num_tokens": 285097553.0, + "step": 7345 + }, + { + "epoch": 0.9344867065258873, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.406401634216309, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8738992214202881, + "num_tokens": 285139080.0, + "step": 7346 + }, + { + "epoch": 0.9346139168044778, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.374719619750977, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8961671590805054, + "num_tokens": 285178343.0, + "step": 7347 + }, + { + "epoch": 0.9347411270830683, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.523170471191406, + "learning_rate": 1e-06, + "loss": 0.4314, + "mean_token_accuracy": 0.8636287450790405, + "num_tokens": 285217402.0, + "step": 7348 + }, + { + "epoch": 0.9348683373616589, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.593626976013184, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.8651413917541504, + "num_tokens": 285253001.0, + "step": 7349 + }, + { + "epoch": 0.9349955476402493, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.421793937683105, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8795403242111206, + "num_tokens": 285292709.0, + "step": 7350 + }, + { + "epoch": 0.9351227579188398, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.391678810119629, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.8732084035873413, + "num_tokens": 285336702.0, + "step": 7351 + }, + { + "epoch": 0.9352499681974303, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.484565734863281, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.880631685256958, + "num_tokens": 285377154.0, + "step": 7352 + }, + { + "epoch": 0.9353771784760209, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.40053653717041, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.882251501083374, + "num_tokens": 285417921.0, + "step": 7353 + }, + { + "epoch": 0.9355043887546114, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.617229461669922, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8819479942321777, + "num_tokens": 285447691.0, + "step": 7354 + }, + { + "epoch": 0.9356315990332019, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.399221420288086, + "learning_rate": 1e-06, + "loss": 0.3847, + "mean_token_accuracy": 0.8710675239562988, + "num_tokens": 285490458.0, + "step": 7355 + }, + { + "epoch": 0.9357588093117923, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.399559020996094, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8943181037902832, + "num_tokens": 285531104.0, + "step": 7356 + }, + { + "epoch": 0.9358860195903829, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.43869686126709, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8847956657409668, + "num_tokens": 285569938.0, + "step": 7357 + }, + { + "epoch": 0.9360132298689734, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.524093627929688, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.8743658065795898, + "num_tokens": 285603578.0, + "step": 7358 + }, + { + "epoch": 0.9361404401475639, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.409368515014648, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.876907229423523, + "num_tokens": 285645470.0, + "step": 7359 + }, + { + "epoch": 0.9362676504261545, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.40684986114502, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8862905502319336, + "num_tokens": 285684202.0, + "step": 7360 + }, + { + "epoch": 0.936394860704745, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.499929428100586, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8908895254135132, + "num_tokens": 285717540.0, + "step": 7361 + }, + { + "epoch": 0.9365220709833355, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.530860900878906, + "learning_rate": 1e-06, + "loss": 0.4284, + "mean_token_accuracy": 0.8567031025886536, + "num_tokens": 285762285.0, + "step": 7362 + }, + { + "epoch": 0.9366492812619259, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.421764373779297, + "learning_rate": 1e-06, + "loss": 0.3987, + "mean_token_accuracy": 0.8661282658576965, + "num_tokens": 285805634.0, + "step": 7363 + }, + { + "epoch": 0.9367764915405165, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.597289085388184, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8748844861984253, + "num_tokens": 285838368.0, + "step": 7364 + }, + { + "epoch": 0.936903701819107, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.453914642333984, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8850493431091309, + "num_tokens": 285876795.0, + "step": 7365 + }, + { + "epoch": 0.9370309120976975, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.421808242797852, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8815755844116211, + "num_tokens": 285917699.0, + "step": 7366 + }, + { + "epoch": 0.937158122376288, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.458227157592773, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8866856098175049, + "num_tokens": 285956908.0, + "step": 7367 + }, + { + "epoch": 0.9372853326548786, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.449965476989746, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8724898099899292, + "num_tokens": 285995068.0, + "step": 7368 + }, + { + "epoch": 0.937412542933469, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.374751091003418, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.89308762550354, + "num_tokens": 286035295.0, + "step": 7369 + }, + { + "epoch": 0.9375397532120595, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.461774826049805, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8832146525382996, + "num_tokens": 286079482.0, + "step": 7370 + }, + { + "epoch": 0.93766696349065, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.441692352294922, + "learning_rate": 1e-06, + "loss": 0.4193, + "mean_token_accuracy": 0.8627067804336548, + "num_tokens": 286115694.0, + "step": 7371 + }, + { + "epoch": 0.9377941737692406, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.410173416137695, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8839248418807983, + "num_tokens": 286156033.0, + "step": 7372 + }, + { + "epoch": 0.9379213840478311, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.430373191833496, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8987070322036743, + "num_tokens": 286191409.0, + "step": 7373 + }, + { + "epoch": 0.9380485943264216, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.498048782348633, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8858860731124878, + "num_tokens": 286228291.0, + "step": 7374 + }, + { + "epoch": 0.938175804605012, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.403794288635254, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8952425718307495, + "num_tokens": 286261595.0, + "step": 7375 + }, + { + "epoch": 0.9383030148836026, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.42072582244873, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8851531744003296, + "num_tokens": 286296999.0, + "step": 7376 + }, + { + "epoch": 0.9384302251621931, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.527454376220703, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8740924596786499, + "num_tokens": 286335023.0, + "step": 7377 + }, + { + "epoch": 0.9385574354407836, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.381335258483887, + "learning_rate": 1e-06, + "loss": 0.4082, + "mean_token_accuracy": 0.8641301393508911, + "num_tokens": 286381204.0, + "step": 7378 + }, + { + "epoch": 0.9386846457193742, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.4109468460083, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8943297863006592, + "num_tokens": 286418448.0, + "step": 7379 + }, + { + "epoch": 0.9388118559979647, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.444209098815918, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8879103660583496, + "num_tokens": 286455761.0, + "step": 7380 + }, + { + "epoch": 0.9389390662765551, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.404656410217285, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8915773630142212, + "num_tokens": 286495436.0, + "step": 7381 + }, + { + "epoch": 0.9390662765551456, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.422836303710938, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8712787628173828, + "num_tokens": 286537187.0, + "step": 7382 + }, + { + "epoch": 0.9391934868337362, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.370901107788086, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8963903784751892, + "num_tokens": 286580715.0, + "step": 7383 + }, + { + "epoch": 0.9393206971123267, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.444812774658203, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.881365180015564, + "num_tokens": 286618712.0, + "step": 7384 + }, + { + "epoch": 0.9394479073909172, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.435230255126953, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8813021183013916, + "num_tokens": 286656603.0, + "step": 7385 + }, + { + "epoch": 0.9395751176695077, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.415221214294434, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8899486660957336, + "num_tokens": 286695194.0, + "step": 7386 + }, + { + "epoch": 0.9397023279480982, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.478508949279785, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8803818225860596, + "num_tokens": 286733843.0, + "step": 7387 + }, + { + "epoch": 0.9398295382266887, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.404789924621582, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8805445432662964, + "num_tokens": 286777227.0, + "step": 7388 + }, + { + "epoch": 0.9399567485052792, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.398035049438477, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8770272731781006, + "num_tokens": 286820518.0, + "step": 7389 + }, + { + "epoch": 0.9400839587838697, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.46973991394043, + "learning_rate": 1e-06, + "loss": 0.3809, + "mean_token_accuracy": 0.8692147135734558, + "num_tokens": 286857253.0, + "step": 7390 + }, + { + "epoch": 0.9402111690624603, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.493253707885742, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8864646553993225, + "num_tokens": 286892155.0, + "step": 7391 + }, + { + "epoch": 0.9403383793410508, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.35964298248291, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9044618606567383, + "num_tokens": 286931189.0, + "step": 7392 + }, + { + "epoch": 0.9404655896196412, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.533408164978027, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.875957727432251, + "num_tokens": 286965511.0, + "step": 7393 + }, + { + "epoch": 0.9405927998982317, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.422342300415039, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8698239922523499, + "num_tokens": 287009962.0, + "step": 7394 + }, + { + "epoch": 0.9407200101768223, + "ewc_loss": 0.0123291015625, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.438173294067383, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.8701281547546387, + "num_tokens": 287056501.0, + "step": 7395 + }, + { + "epoch": 0.9408472204554128, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.519777297973633, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8836899995803833, + "num_tokens": 287092852.0, + "step": 7396 + }, + { + "epoch": 0.9409744307340033, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.423270225524902, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8773815631866455, + "num_tokens": 287131598.0, + "step": 7397 + }, + { + "epoch": 0.9411016410125939, + "ewc_loss": 0.01239013671875, + "ewc_loss_diag": 3.1948089599609375e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.408954620361328, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8973026275634766, + "num_tokens": 287166726.0, + "step": 7398 + }, + { + "epoch": 0.9412288512911843, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.54030704498291, + "learning_rate": 1e-06, + "loss": 0.4191, + "mean_token_accuracy": 0.8598094582557678, + "num_tokens": 287203495.0, + "step": 7399 + }, + { + "epoch": 0.9413560615697748, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.397442817687988, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.880548894405365, + "num_tokens": 287241917.0, + "step": 7400 + }, + { + "epoch": 0.9414832718483653, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.397183418273926, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8890089392662048, + "num_tokens": 287283591.0, + "step": 7401 + }, + { + "epoch": 0.9416104821269559, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.517780303955078, + "learning_rate": 1e-06, + "loss": 0.3985, + "mean_token_accuracy": 0.8641113042831421, + "num_tokens": 287322305.0, + "step": 7402 + }, + { + "epoch": 0.9417376924055464, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.641301155090332, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8786109685897827, + "num_tokens": 287352991.0, + "step": 7403 + }, + { + "epoch": 0.9418649026841369, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.425792694091797, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8906535506248474, + "num_tokens": 287393439.0, + "step": 7404 + }, + { + "epoch": 0.9419921129627273, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.540567398071289, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8861592411994934, + "num_tokens": 287431941.0, + "step": 7405 + }, + { + "epoch": 0.9421193232413179, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.442800521850586, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8775856494903564, + "num_tokens": 287482267.0, + "step": 7406 + }, + { + "epoch": 0.9422465335199084, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.468820571899414, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8741120100021362, + "num_tokens": 287523718.0, + "step": 7407 + }, + { + "epoch": 0.9423737437984989, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.523504257202148, + "learning_rate": 1e-06, + "loss": 0.4465, + "mean_token_accuracy": 0.8534706830978394, + "num_tokens": 287568834.0, + "step": 7408 + }, + { + "epoch": 0.9425009540770894, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.671205520629883, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.877484917640686, + "num_tokens": 287595034.0, + "step": 7409 + }, + { + "epoch": 0.94262816435568, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.426336288452148, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8956670761108398, + "num_tokens": 287631264.0, + "step": 7410 + }, + { + "epoch": 0.9427553746342705, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.419363021850586, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8828483819961548, + "num_tokens": 287671158.0, + "step": 7411 + }, + { + "epoch": 0.9428825849128609, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.584890365600586, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8829916715621948, + "num_tokens": 287707521.0, + "step": 7412 + }, + { + "epoch": 0.9430097951914514, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.456692695617676, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8733838200569153, + "num_tokens": 287750725.0, + "step": 7413 + }, + { + "epoch": 0.943137005470042, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.491268157958984, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8746474981307983, + "num_tokens": 287790593.0, + "step": 7414 + }, + { + "epoch": 0.9432642157486325, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.451220512390137, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8852449655532837, + "num_tokens": 287828368.0, + "step": 7415 + }, + { + "epoch": 0.943391426027223, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.369239807128906, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.9015288352966309, + "num_tokens": 287865135.0, + "step": 7416 + }, + { + "epoch": 0.9435186363058136, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.218650817871094e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.526671409606934, + "learning_rate": 1e-06, + "loss": 0.3867, + "mean_token_accuracy": 0.8684423565864563, + "num_tokens": 287902344.0, + "step": 7417 + }, + { + "epoch": 0.943645846584404, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.446348190307617, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.877918004989624, + "num_tokens": 287942575.0, + "step": 7418 + }, + { + "epoch": 0.9437730568629945, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.443160057067871, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8710240125656128, + "num_tokens": 287982720.0, + "step": 7419 + }, + { + "epoch": 0.943900267141585, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.494433403015137, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8815349340438843, + "num_tokens": 288020839.0, + "step": 7420 + }, + { + "epoch": 0.9440274774201756, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.433483123779297, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8833933472633362, + "num_tokens": 288057920.0, + "step": 7421 + }, + { + "epoch": 0.9441546876987661, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.535303115844727, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8736134767532349, + "num_tokens": 288096010.0, + "step": 7422 + }, + { + "epoch": 0.9442818979773566, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.428350448608398, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8923133015632629, + "num_tokens": 288132882.0, + "step": 7423 + }, + { + "epoch": 0.944409108255947, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.446199417114258, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8838538527488708, + "num_tokens": 288169944.0, + "step": 7424 + }, + { + "epoch": 0.9445363185345376, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.499289512634277, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8759173154830933, + "num_tokens": 288208417.0, + "step": 7425 + }, + { + "epoch": 0.9446635288131281, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.577858924865723, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8775330781936646, + "num_tokens": 288238916.0, + "step": 7426 + }, + { + "epoch": 0.9447907390917186, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.432421684265137, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8954978585243225, + "num_tokens": 288277798.0, + "step": 7427 + }, + { + "epoch": 0.9449179493703092, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.38098430633545, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8833000659942627, + "num_tokens": 288319251.0, + "step": 7428 + }, + { + "epoch": 0.9450451596488997, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.567146301269531, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8851655721664429, + "num_tokens": 288351923.0, + "step": 7429 + }, + { + "epoch": 0.9451723699274901, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.50582218170166, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8790116310119629, + "num_tokens": 288388912.0, + "step": 7430 + }, + { + "epoch": 0.9452995802060806, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.399727821350098, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8781291842460632, + "num_tokens": 288425558.0, + "step": 7431 + }, + { + "epoch": 0.9454267904846712, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.443936347961426, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8719068765640259, + "num_tokens": 288464780.0, + "step": 7432 + }, + { + "epoch": 0.9455540007632617, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.58273983001709, + "learning_rate": 1e-06, + "loss": 0.4326, + "mean_token_accuracy": 0.8508263826370239, + "num_tokens": 288503194.0, + "step": 7433 + }, + { + "epoch": 0.9456812110418522, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.55027961730957, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8875868916511536, + "num_tokens": 288533119.0, + "step": 7434 + }, + { + "epoch": 0.9458084213204427, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.382246971130371, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8875051736831665, + "num_tokens": 288571140.0, + "step": 7435 + }, + { + "epoch": 0.9459356315990332, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.53649616241455, + "learning_rate": 1e-06, + "loss": 0.3889, + "mean_token_accuracy": 0.86655193567276, + "num_tokens": 288605908.0, + "step": 7436 + }, + { + "epoch": 0.9460628418776237, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.456832885742188, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.878063440322876, + "num_tokens": 288643776.0, + "step": 7437 + }, + { + "epoch": 0.9461900521562142, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.575234413146973, + "learning_rate": 1e-06, + "loss": 0.3893, + "mean_token_accuracy": 0.8717520236968994, + "num_tokens": 288676926.0, + "step": 7438 + }, + { + "epoch": 0.9463172624348047, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.459000587463379, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8917614221572876, + "num_tokens": 288710088.0, + "step": 7439 + }, + { + "epoch": 0.9464444727133953, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.399054527282715, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8796296715736389, + "num_tokens": 288749170.0, + "step": 7440 + }, + { + "epoch": 0.9465716829919858, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.575605392456055, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8763895034790039, + "num_tokens": 288780257.0, + "step": 7441 + }, + { + "epoch": 0.9466988932705762, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.444231986999512, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8807293176651001, + "num_tokens": 288821869.0, + "step": 7442 + }, + { + "epoch": 0.9468261035491667, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.556746482849121, + "learning_rate": 1e-06, + "loss": 0.3889, + "mean_token_accuracy": 0.866188645362854, + "num_tokens": 288856928.0, + "step": 7443 + }, + { + "epoch": 0.9469533138277573, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.24249267578125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.433453559875488, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8876850008964539, + "num_tokens": 288894764.0, + "step": 7444 + }, + { + "epoch": 0.9470805241063478, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.377138137817383, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8783964514732361, + "num_tokens": 288935492.0, + "step": 7445 + }, + { + "epoch": 0.9472077343849383, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.439827919006348, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8792084455490112, + "num_tokens": 288975917.0, + "step": 7446 + }, + { + "epoch": 0.9473349446635289, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.474000930786133, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8844797611236572, + "num_tokens": 289010338.0, + "step": 7447 + }, + { + "epoch": 0.9474621549421193, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.393227577209473, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8894158601760864, + "num_tokens": 289049039.0, + "step": 7448 + }, + { + "epoch": 0.9475893652207098, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.475862503051758, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8800382018089294, + "num_tokens": 289088389.0, + "step": 7449 + }, + { + "epoch": 0.9477165754993003, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.549982070922852, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8832040429115295, + "num_tokens": 289120600.0, + "step": 7450 + }, + { + "epoch": 0.9478437857778909, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.398938179016113, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8882838487625122, + "num_tokens": 289162992.0, + "step": 7451 + }, + { + "epoch": 0.9479709960564814, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.43873405456543, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8856185674667358, + "num_tokens": 289195831.0, + "step": 7452 + }, + { + "epoch": 0.9480982063350719, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.549402236938477, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8839234113693237, + "num_tokens": 289238069.0, + "step": 7453 + }, + { + "epoch": 0.9482254166136623, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.510647773742676, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8702805638313293, + "num_tokens": 289274232.0, + "step": 7454 + }, + { + "epoch": 0.9483526268922529, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.412138938903809, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.8726884722709656, + "num_tokens": 289316764.0, + "step": 7455 + }, + { + "epoch": 0.9484798371708434, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.509799003601074, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8800563812255859, + "num_tokens": 289350393.0, + "step": 7456 + }, + { + "epoch": 0.9486070474494339, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.460020065307617, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8749469518661499, + "num_tokens": 289392219.0, + "step": 7457 + }, + { + "epoch": 0.9487342577280244, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.526185989379883, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8698612451553345, + "num_tokens": 289427419.0, + "step": 7458 + }, + { + "epoch": 0.948861468006615, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.444100379943848, + "learning_rate": 1e-06, + "loss": 0.3664, + "mean_token_accuracy": 0.8772525191307068, + "num_tokens": 289467902.0, + "step": 7459 + }, + { + "epoch": 0.9489886782852054, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.479583740234375, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.8767511248588562, + "num_tokens": 289505891.0, + "step": 7460 + }, + { + "epoch": 0.9491158885637959, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.458581924438477, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.897646427154541, + "num_tokens": 289545070.0, + "step": 7461 + }, + { + "epoch": 0.9492430988423864, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.472460746765137, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8856054544448853, + "num_tokens": 289579366.0, + "step": 7462 + }, + { + "epoch": 0.949370309120977, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.421662330627441, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.8748409748077393, + "num_tokens": 289621844.0, + "step": 7463 + }, + { + "epoch": 0.9494975193995675, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.498855590820312, + "learning_rate": 1e-06, + "loss": 0.3671, + "mean_token_accuracy": 0.8761577606201172, + "num_tokens": 289663740.0, + "step": 7464 + }, + { + "epoch": 0.949624729678158, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.490876197814941, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8831371665000916, + "num_tokens": 289697251.0, + "step": 7465 + }, + { + "epoch": 0.9497519399567486, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.469882011413574, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8777984380722046, + "num_tokens": 289738651.0, + "step": 7466 + }, + { + "epoch": 0.949879150235339, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.449405670166016, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8840879201889038, + "num_tokens": 289775721.0, + "step": 7467 + }, + { + "epoch": 0.9500063605139295, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.40530014038086, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8770517110824585, + "num_tokens": 289821140.0, + "step": 7468 + }, + { + "epoch": 0.95013357079252, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.502866744995117, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8810468316078186, + "num_tokens": 289856063.0, + "step": 7469 + }, + { + "epoch": 0.9502607810711106, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.493022918701172, + "learning_rate": 1e-06, + "loss": 0.418, + "mean_token_accuracy": 0.862058162689209, + "num_tokens": 289899408.0, + "step": 7470 + }, + { + "epoch": 0.9503879913497011, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.457690238952637, + "learning_rate": 1e-06, + "loss": 0.3826, + "mean_token_accuracy": 0.8699191808700562, + "num_tokens": 289938106.0, + "step": 7471 + }, + { + "epoch": 0.9505152016282916, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.669747352600098, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.872955322265625, + "num_tokens": 289971890.0, + "step": 7472 + }, + { + "epoch": 0.950642411906882, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.411786079406738, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.886685848236084, + "num_tokens": 290008575.0, + "step": 7473 + }, + { + "epoch": 0.9507696221854726, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.444584846496582, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8804967403411865, + "num_tokens": 290050459.0, + "step": 7474 + }, + { + "epoch": 0.9508968324640631, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.604180335998535, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8710199594497681, + "num_tokens": 290082510.0, + "step": 7475 + }, + { + "epoch": 0.9510240427426536, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.38586139678955, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8825165033340454, + "num_tokens": 290124418.0, + "step": 7476 + }, + { + "epoch": 0.9511512530212441, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.587872505187988, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8837844133377075, + "num_tokens": 290154312.0, + "step": 7477 + }, + { + "epoch": 0.9512784632998347, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.397162437438965, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8792881965637207, + "num_tokens": 290193703.0, + "step": 7478 + }, + { + "epoch": 0.9514056735784251, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.4839506149292, + "learning_rate": 1e-06, + "loss": 0.4262, + "mean_token_accuracy": 0.859983503818512, + "num_tokens": 290235119.0, + "step": 7479 + }, + { + "epoch": 0.9515328838570156, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.457087516784668, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8710439205169678, + "num_tokens": 290275747.0, + "step": 7480 + }, + { + "epoch": 0.9516600941356061, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.435623168945312, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8762515187263489, + "num_tokens": 290310941.0, + "step": 7481 + }, + { + "epoch": 0.9517873044141967, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.437592506408691, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8844376802444458, + "num_tokens": 290352067.0, + "step": 7482 + }, + { + "epoch": 0.9519145146927872, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.492105484008789, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8838233947753906, + "num_tokens": 290383602.0, + "step": 7483 + }, + { + "epoch": 0.9520417249713777, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.380961418151855, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8799508810043335, + "num_tokens": 290422213.0, + "step": 7484 + }, + { + "epoch": 0.9521689352499682, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.459437370300293, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8951207399368286, + "num_tokens": 290463393.0, + "step": 7485 + }, + { + "epoch": 0.9522961455285587, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.340444564819336, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8849104642868042, + "num_tokens": 290508121.0, + "step": 7486 + }, + { + "epoch": 0.9524233558071492, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.521854400634766, + "learning_rate": 1e-06, + "loss": 0.427, + "mean_token_accuracy": 0.8599351048469543, + "num_tokens": 290549580.0, + "step": 7487 + }, + { + "epoch": 0.9525505660857397, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.460582733154297, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8797427415847778, + "num_tokens": 290588274.0, + "step": 7488 + }, + { + "epoch": 0.9526777763643303, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.463569641113281, + "learning_rate": 1e-06, + "loss": 0.408, + "mean_token_accuracy": 0.8634436130523682, + "num_tokens": 290629759.0, + "step": 7489 + }, + { + "epoch": 0.9528049866429208, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.399675369262695, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8751910328865051, + "num_tokens": 290673279.0, + "step": 7490 + }, + { + "epoch": 0.9529321969215112, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.486773490905762, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.888361930847168, + "num_tokens": 290707588.0, + "step": 7491 + }, + { + "epoch": 0.9530594072001017, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.543874740600586, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8822330236434937, + "num_tokens": 290740048.0, + "step": 7492 + }, + { + "epoch": 0.9531866174786923, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.502556800842285, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8733765482902527, + "num_tokens": 290781225.0, + "step": 7493 + }, + { + "epoch": 0.9533138277572828, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.349433898925781, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8991944789886475, + "num_tokens": 290820852.0, + "step": 7494 + }, + { + "epoch": 0.9534410380358733, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.40343952178955, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8861972689628601, + "num_tokens": 290864217.0, + "step": 7495 + }, + { + "epoch": 0.9535682483144639, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.52463150024414, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8725289106369019, + "num_tokens": 290903313.0, + "step": 7496 + }, + { + "epoch": 0.9536954585930543, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.42335319519043, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8863360285758972, + "num_tokens": 290944516.0, + "step": 7497 + }, + { + "epoch": 0.9538226688716448, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.622722625732422, + "learning_rate": 1e-06, + "loss": 0.3907, + "mean_token_accuracy": 0.8668832778930664, + "num_tokens": 290977348.0, + "step": 7498 + }, + { + "epoch": 0.9539498791502353, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.438538551330566, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8784092664718628, + "num_tokens": 291017267.0, + "step": 7499 + }, + { + "epoch": 0.9540770894288259, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.48730754852295, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8752882480621338, + "num_tokens": 291057255.0, + "step": 7500 + }, + { + "epoch": 0.9542042997074164, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.477569580078125, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8888492584228516, + "num_tokens": 291089619.0, + "step": 7501 + }, + { + "epoch": 0.9543315099860069, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.495709419250488, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8839765191078186, + "num_tokens": 291126538.0, + "step": 7502 + }, + { + "epoch": 0.9544587202645973, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.481270790100098, + "learning_rate": 1e-06, + "loss": 0.411, + "mean_token_accuracy": 0.8631987571716309, + "num_tokens": 291167054.0, + "step": 7503 + }, + { + "epoch": 0.9545859305431879, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.45776081085205, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8737296462059021, + "num_tokens": 291206610.0, + "step": 7504 + }, + { + "epoch": 0.9547131408217784, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.3848237991333, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8850608468055725, + "num_tokens": 291249848.0, + "step": 7505 + }, + { + "epoch": 0.9548403511003689, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.497474670410156, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8824089765548706, + "num_tokens": 291288655.0, + "step": 7506 + }, + { + "epoch": 0.9549675613789594, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.416491508483887, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8835658431053162, + "num_tokens": 291327786.0, + "step": 7507 + }, + { + "epoch": 0.95509477165755, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.266334533691406e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.401228904724121, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8843305110931396, + "num_tokens": 291371217.0, + "step": 7508 + }, + { + "epoch": 0.9552219819361404, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.466894149780273, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8789938688278198, + "num_tokens": 291410792.0, + "step": 7509 + }, + { + "epoch": 0.9553491922147309, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.429730415344238, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8925608396530151, + "num_tokens": 291450210.0, + "step": 7510 + }, + { + "epoch": 0.9554764024933214, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.42111587524414, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8690953850746155, + "num_tokens": 291491607.0, + "step": 7511 + }, + { + "epoch": 0.955603612771912, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.481498718261719, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8821632266044617, + "num_tokens": 291528630.0, + "step": 7512 + }, + { + "epoch": 0.9557308230505025, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.438375473022461, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.879810631275177, + "num_tokens": 291568216.0, + "step": 7513 + }, + { + "epoch": 0.955858033329093, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.425442695617676, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8772722482681274, + "num_tokens": 291613189.0, + "step": 7514 + }, + { + "epoch": 0.9559852436076836, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.358919143676758, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.881375789642334, + "num_tokens": 291657710.0, + "step": 7515 + }, + { + "epoch": 0.956112453886274, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.549911499023438, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8778476715087891, + "num_tokens": 291690290.0, + "step": 7516 + }, + { + "epoch": 0.9562396641648645, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.43952465057373, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8772763013839722, + "num_tokens": 291729229.0, + "step": 7517 + }, + { + "epoch": 0.956366874443455, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.371212005615234, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8853585720062256, + "num_tokens": 291771378.0, + "step": 7518 + }, + { + "epoch": 0.9564940847220456, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.439688682556152, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8869578242301941, + "num_tokens": 291816704.0, + "step": 7519 + }, + { + "epoch": 0.9566212950006361, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.407515525817871, + "learning_rate": 1e-06, + "loss": 0.4094, + "mean_token_accuracy": 0.8602579236030579, + "num_tokens": 291859750.0, + "step": 7520 + }, + { + "epoch": 0.9567485052792266, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.423830032348633, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8837025165557861, + "num_tokens": 291906309.0, + "step": 7521 + }, + { + "epoch": 0.956875715557817, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.474371910095215, + "learning_rate": 1e-06, + "loss": 0.3714, + "mean_token_accuracy": 0.8712946176528931, + "num_tokens": 291950904.0, + "step": 7522 + }, + { + "epoch": 0.9570029258364076, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.463074684143066, + "learning_rate": 1e-06, + "loss": 0.3992, + "mean_token_accuracy": 0.8654106259346008, + "num_tokens": 291994720.0, + "step": 7523 + }, + { + "epoch": 0.9571301361149981, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.496973991394043, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.882785439491272, + "num_tokens": 292029004.0, + "step": 7524 + }, + { + "epoch": 0.9572573463935886, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.478179931640625, + "learning_rate": 1e-06, + "loss": 0.3875, + "mean_token_accuracy": 0.8690853118896484, + "num_tokens": 292067349.0, + "step": 7525 + }, + { + "epoch": 0.9573845566721791, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.450928688049316, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8847599029541016, + "num_tokens": 292111007.0, + "step": 7526 + }, + { + "epoch": 0.9575117669507697, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.480542182922363, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8838079571723938, + "num_tokens": 292148497.0, + "step": 7527 + }, + { + "epoch": 0.9576389772293601, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.419719696044922, + "learning_rate": 1e-06, + "loss": 0.3851, + "mean_token_accuracy": 0.8731081485748291, + "num_tokens": 292188634.0, + "step": 7528 + }, + { + "epoch": 0.9577661875079506, + "ewc_loss": 0.012451171875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.179115295410156e-06, + "grad_norm": 8.479641914367676, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.891070544719696, + "num_tokens": 292224838.0, + "step": 7529 + }, + { + "epoch": 0.9578933977865411, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.554813385009766, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.8748719692230225, + "num_tokens": 292262675.0, + "step": 7530 + }, + { + "epoch": 0.9580206080651317, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.488512992858887, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8680251240730286, + "num_tokens": 292299377.0, + "step": 7531 + }, + { + "epoch": 0.9581478183437222, + "ewc_loss": 0.01251220703125, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.462491035461426, + "learning_rate": 1e-06, + "loss": 0.3732, + "mean_token_accuracy": 0.8766582608222961, + "num_tokens": 292338049.0, + "step": 7532 + }, + { + "epoch": 0.9582750286223127, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.510822296142578, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8728020787239075, + "num_tokens": 292373121.0, + "step": 7533 + }, + { + "epoch": 0.9584022389009031, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.472749710083008, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8901426792144775, + "num_tokens": 292403951.0, + "step": 7534 + }, + { + "epoch": 0.9585294491794937, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.529786109924316, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.8669476509094238, + "num_tokens": 292443933.0, + "step": 7535 + }, + { + "epoch": 0.9586566594580842, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.43343448638916, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8802601099014282, + "num_tokens": 292480001.0, + "step": 7536 + }, + { + "epoch": 0.9587838697366747, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.474894523620605, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8788685202598572, + "num_tokens": 292514300.0, + "step": 7537 + }, + { + "epoch": 0.9589110800152653, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.2901763916015625e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.428603172302246, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.8664446473121643, + "num_tokens": 292554742.0, + "step": 7538 + }, + { + "epoch": 0.9590382902938558, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.409258842468262, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.884343147277832, + "num_tokens": 292591997.0, + "step": 7539 + }, + { + "epoch": 0.9591655005724462, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.440651893615723, + "learning_rate": 1e-06, + "loss": 0.3923, + "mean_token_accuracy": 0.8716084957122803, + "num_tokens": 292629987.0, + "step": 7540 + }, + { + "epoch": 0.9592927108510367, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.556231498718262, + "learning_rate": 1e-06, + "loss": 0.3825, + "mean_token_accuracy": 0.8748654127120972, + "num_tokens": 292663507.0, + "step": 7541 + }, + { + "epoch": 0.9594199211296273, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.469169616699219, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.8753828406333923, + "num_tokens": 292701291.0, + "step": 7542 + }, + { + "epoch": 0.9595471314082178, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.391138076782227, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8790159821510315, + "num_tokens": 292743797.0, + "step": 7543 + }, + { + "epoch": 0.9596743416868083, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.384160995483398, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8847694396972656, + "num_tokens": 292785178.0, + "step": 7544 + }, + { + "epoch": 0.9598015519653988, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.371319770812988, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8820299506187439, + "num_tokens": 292826475.0, + "step": 7545 + }, + { + "epoch": 0.9599287622439893, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.501660346984863, + "learning_rate": 1e-06, + "loss": 0.3981, + "mean_token_accuracy": 0.8686766624450684, + "num_tokens": 292867720.0, + "step": 7546 + }, + { + "epoch": 0.9600559725225798, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.564990043640137, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8771735429763794, + "num_tokens": 292903383.0, + "step": 7547 + }, + { + "epoch": 0.9601831828011703, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.54561996459961, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8803451061248779, + "num_tokens": 292941673.0, + "step": 7548 + }, + { + "epoch": 0.9603103930797608, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.41630744934082, + "learning_rate": 1e-06, + "loss": 0.3853, + "mean_token_accuracy": 0.8698848485946655, + "num_tokens": 292984393.0, + "step": 7549 + }, + { + "epoch": 0.9604376033583514, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.463639259338379, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8739925026893616, + "num_tokens": 293025327.0, + "step": 7550 + }, + { + "epoch": 0.9605648136369419, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.366354942321777, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.887937605381012, + "num_tokens": 293066162.0, + "step": 7551 + }, + { + "epoch": 0.9606920239155323, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.490702629089355, + "learning_rate": 1e-06, + "loss": 0.3903, + "mean_token_accuracy": 0.8709690570831299, + "num_tokens": 293103564.0, + "step": 7552 + }, + { + "epoch": 0.9608192341941229, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.5391206741333, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8859463334083557, + "num_tokens": 293138586.0, + "step": 7553 + }, + { + "epoch": 0.9609464444727134, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.45926284790039, + "learning_rate": 1e-06, + "loss": 0.3911, + "mean_token_accuracy": 0.8665181398391724, + "num_tokens": 293174017.0, + "step": 7554 + }, + { + "epoch": 0.9610736547513039, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.472212791442871, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8900685906410217, + "num_tokens": 293207481.0, + "step": 7555 + }, + { + "epoch": 0.9612008650298944, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.422334671020508, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8957570195198059, + "num_tokens": 293245476.0, + "step": 7556 + }, + { + "epoch": 0.961328075308485, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.422513008117676, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8860739469528198, + "num_tokens": 293287401.0, + "step": 7557 + }, + { + "epoch": 0.9614552855870754, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.520065307617188, + "learning_rate": 1e-06, + "loss": 0.407, + "mean_token_accuracy": 0.8668569326400757, + "num_tokens": 293324172.0, + "step": 7558 + }, + { + "epoch": 0.9615824958656659, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.381940841674805, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.879064679145813, + "num_tokens": 293366035.0, + "step": 7559 + }, + { + "epoch": 0.9617097061442564, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.393043518066406, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8852529525756836, + "num_tokens": 293412315.0, + "step": 7560 + }, + { + "epoch": 0.961836916422847, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.468940734863281, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8816330432891846, + "num_tokens": 293449231.0, + "step": 7561 + }, + { + "epoch": 0.9619641267014375, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.510433197021484, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.896537721157074, + "num_tokens": 293481772.0, + "step": 7562 + }, + { + "epoch": 0.962091336980028, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.416604995727539, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8874327540397644, + "num_tokens": 293520871.0, + "step": 7563 + }, + { + "epoch": 0.9622185472586186, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.484391212463379, + "learning_rate": 1e-06, + "loss": 0.4143, + "mean_token_accuracy": 0.8615331649780273, + "num_tokens": 293565515.0, + "step": 7564 + }, + { + "epoch": 0.962345757537209, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.401195526123047, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8726660013198853, + "num_tokens": 293607526.0, + "step": 7565 + }, + { + "epoch": 0.9624729678157995, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.487903594970703, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8742485046386719, + "num_tokens": 293646699.0, + "step": 7566 + }, + { + "epoch": 0.96260017809439, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.43823528289795, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8820309638977051, + "num_tokens": 293681387.0, + "step": 7567 + }, + { + "epoch": 0.9627273883729806, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.514595985412598, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8826081156730652, + "num_tokens": 293717038.0, + "step": 7568 + }, + { + "epoch": 0.9628545986515711, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.505183219909668, + "learning_rate": 1e-06, + "loss": 0.3681, + "mean_token_accuracy": 0.8776938915252686, + "num_tokens": 293755449.0, + "step": 7569 + }, + { + "epoch": 0.9629818089301616, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.43614673614502, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.890326201915741, + "num_tokens": 293795513.0, + "step": 7570 + }, + { + "epoch": 0.963109019208752, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.514558792114258, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.8714251518249512, + "num_tokens": 293834994.0, + "step": 7571 + }, + { + "epoch": 0.9632362294873426, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.430521965026855, + "learning_rate": 1e-06, + "loss": 0.3907, + "mean_token_accuracy": 0.8670922517776489, + "num_tokens": 293875468.0, + "step": 7572 + }, + { + "epoch": 0.9633634397659331, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.46216106414795, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.89185631275177, + "num_tokens": 293910003.0, + "step": 7573 + }, + { + "epoch": 0.9634906500445236, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.455890655517578, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8918566703796387, + "num_tokens": 293948593.0, + "step": 7574 + }, + { + "epoch": 0.9636178603231141, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.487565040588379, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.8717643022537231, + "num_tokens": 293987092.0, + "step": 7575 + }, + { + "epoch": 0.9637450706017047, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.44032096862793, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8704550266265869, + "num_tokens": 294027340.0, + "step": 7576 + }, + { + "epoch": 0.9638722808802951, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.451369285583496, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.8676770329475403, + "num_tokens": 294075565.0, + "step": 7577 + }, + { + "epoch": 0.9639994911588856, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.437723159790039, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8928082585334778, + "num_tokens": 294112246.0, + "step": 7578 + }, + { + "epoch": 0.9641267014374761, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.489089012145996, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8787037134170532, + "num_tokens": 294151603.0, + "step": 7579 + }, + { + "epoch": 0.9642539117160667, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.509498596191406, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.8699438571929932, + "num_tokens": 294187135.0, + "step": 7580 + }, + { + "epoch": 0.9643811219946572, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.47662353515625, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8836250901222229, + "num_tokens": 294227312.0, + "step": 7581 + }, + { + "epoch": 0.9645083322732477, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.520888328552246, + "learning_rate": 1e-06, + "loss": 0.3846, + "mean_token_accuracy": 0.872516393661499, + "num_tokens": 294267806.0, + "step": 7582 + }, + { + "epoch": 0.9646355425518381, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.537772178649902, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8727607727050781, + "num_tokens": 294303623.0, + "step": 7583 + }, + { + "epoch": 0.9647627528304287, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.65271282196045, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.880893349647522, + "num_tokens": 294330779.0, + "step": 7584 + }, + { + "epoch": 0.9648899631090192, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.544787406921387, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8795502781867981, + "num_tokens": 294364171.0, + "step": 7585 + }, + { + "epoch": 0.9650171733876097, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.452804565429688, + "learning_rate": 1e-06, + "loss": 0.4068, + "mean_token_accuracy": 0.8644444942474365, + "num_tokens": 294405962.0, + "step": 7586 + }, + { + "epoch": 0.9651443836662003, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.402488708496094, + "learning_rate": 1e-06, + "loss": 0.3717, + "mean_token_accuracy": 0.8761419653892517, + "num_tokens": 294450250.0, + "step": 7587 + }, + { + "epoch": 0.9652715939447908, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.63198184967041, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8726608753204346, + "num_tokens": 294484452.0, + "step": 7588 + }, + { + "epoch": 0.9653988042233812, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.508077621459961, + "learning_rate": 1e-06, + "loss": 0.4198, + "mean_token_accuracy": 0.858355700969696, + "num_tokens": 294526523.0, + "step": 7589 + }, + { + "epoch": 0.9655260145019717, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.494446754455566, + "learning_rate": 1e-06, + "loss": 0.4059, + "mean_token_accuracy": 0.8641621470451355, + "num_tokens": 294568335.0, + "step": 7590 + }, + { + "epoch": 0.9656532247805623, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.494978904724121, + "learning_rate": 1e-06, + "loss": 0.3729, + "mean_token_accuracy": 0.8750016689300537, + "num_tokens": 294608923.0, + "step": 7591 + }, + { + "epoch": 0.9657804350591528, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.413782119750977, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8840158581733704, + "num_tokens": 294651759.0, + "step": 7592 + }, + { + "epoch": 0.9659076453377433, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.48344612121582, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8725179433822632, + "num_tokens": 294692518.0, + "step": 7593 + }, + { + "epoch": 0.9660348556163338, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.577308654785156, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8688936233520508, + "num_tokens": 294727529.0, + "step": 7594 + }, + { + "epoch": 0.9661620658949243, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.44172191619873, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8779082298278809, + "num_tokens": 294769326.0, + "step": 7595 + }, + { + "epoch": 0.9662892761735148, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.494745254516602, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.881799578666687, + "num_tokens": 294805283.0, + "step": 7596 + }, + { + "epoch": 0.9664164864521053, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.488288879394531, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8843333125114441, + "num_tokens": 294841466.0, + "step": 7597 + }, + { + "epoch": 0.9665436967306958, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.412269592285156, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8843511939048767, + "num_tokens": 294883902.0, + "step": 7598 + }, + { + "epoch": 0.9666709070092864, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.611287117004395, + "learning_rate": 1e-06, + "loss": 0.4052, + "mean_token_accuracy": 0.8637058138847351, + "num_tokens": 294920180.0, + "step": 7599 + }, + { + "epoch": 0.9667981172878769, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.516936302185059, + "learning_rate": 1e-06, + "loss": 0.4157, + "mean_token_accuracy": 0.8625721335411072, + "num_tokens": 294957375.0, + "step": 7600 + }, + { + "epoch": 0.9669253275664673, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.509035110473633, + "learning_rate": 1e-06, + "loss": 0.3903, + "mean_token_accuracy": 0.8666030168533325, + "num_tokens": 294999291.0, + "step": 7601 + }, + { + "epoch": 0.9670525378450578, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.504755973815918, + "learning_rate": 1e-06, + "loss": 0.4145, + "mean_token_accuracy": 0.8621897101402283, + "num_tokens": 295036355.0, + "step": 7602 + }, + { + "epoch": 0.9671797481236484, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.403732299804688, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8792184591293335, + "num_tokens": 295079983.0, + "step": 7603 + }, + { + "epoch": 0.9673069584022389, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.540122985839844, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.8708688616752625, + "num_tokens": 295114209.0, + "step": 7604 + }, + { + "epoch": 0.9674341686808294, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.482208251953125, + "learning_rate": 1e-06, + "loss": 0.4139, + "mean_token_accuracy": 0.8641040921211243, + "num_tokens": 295153434.0, + "step": 7605 + }, + { + "epoch": 0.96756137895942, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.429895401000977, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8887767791748047, + "num_tokens": 295197036.0, + "step": 7606 + }, + { + "epoch": 0.9676885892380104, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.454258918762207, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8900749087333679, + "num_tokens": 295237714.0, + "step": 7607 + }, + { + "epoch": 0.9678157995166009, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.468738555908203, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8895106315612793, + "num_tokens": 295272647.0, + "step": 7608 + }, + { + "epoch": 0.9679430097951914, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.460183143615723, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8851897716522217, + "num_tokens": 295315503.0, + "step": 7609 + }, + { + "epoch": 0.968070220073782, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.4442720413208, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8837987184524536, + "num_tokens": 295351330.0, + "step": 7610 + }, + { + "epoch": 0.9681974303523725, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.44680404663086, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.880715548992157, + "num_tokens": 295394204.0, + "step": 7611 + }, + { + "epoch": 0.968324640630963, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.566539764404297, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8802759647369385, + "num_tokens": 295432766.0, + "step": 7612 + }, + { + "epoch": 0.9684518509095535, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.476305961608887, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8797086477279663, + "num_tokens": 295476859.0, + "step": 7613 + }, + { + "epoch": 0.968579061188144, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.470115661621094, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8865355253219604, + "num_tokens": 295516185.0, + "step": 7614 + }, + { + "epoch": 0.9687062714667345, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.446045875549316, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8887643218040466, + "num_tokens": 295557816.0, + "step": 7615 + }, + { + "epoch": 0.968833481745325, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.591773986816406, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8783419132232666, + "num_tokens": 295591312.0, + "step": 7616 + }, + { + "epoch": 0.9689606920239155, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.49551773071289, + "learning_rate": 1e-06, + "loss": 0.4076, + "mean_token_accuracy": 0.8658086061477661, + "num_tokens": 295634523.0, + "step": 7617 + }, + { + "epoch": 0.9690879023025061, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.517934799194336, + "learning_rate": 1e-06, + "loss": 0.3717, + "mean_token_accuracy": 0.8740119934082031, + "num_tokens": 295672902.0, + "step": 7618 + }, + { + "epoch": 0.9692151125810966, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.553699493408203, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8794125914573669, + "num_tokens": 295710233.0, + "step": 7619 + }, + { + "epoch": 0.969342322859687, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.43315315246582, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8815608024597168, + "num_tokens": 295752613.0, + "step": 7620 + }, + { + "epoch": 0.9694695331382776, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.47069263458252, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8854258060455322, + "num_tokens": 295789124.0, + "step": 7621 + }, + { + "epoch": 0.9695967434168681, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.470161437988281, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8709075450897217, + "num_tokens": 295830764.0, + "step": 7622 + }, + { + "epoch": 0.9697239536954586, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.481742858886719, + "learning_rate": 1e-06, + "loss": 0.3879, + "mean_token_accuracy": 0.8734283447265625, + "num_tokens": 295870377.0, + "step": 7623 + }, + { + "epoch": 0.9698511639740491, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.57162857055664, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8839384317398071, + "num_tokens": 295902833.0, + "step": 7624 + }, + { + "epoch": 0.9699783742526397, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.483595848083496, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.875426173210144, + "num_tokens": 295940520.0, + "step": 7625 + }, + { + "epoch": 0.9701055845312301, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.412848472595215, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8724536895751953, + "num_tokens": 295981499.0, + "step": 7626 + }, + { + "epoch": 0.9702327948098206, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.444665908813477, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8842674493789673, + "num_tokens": 296020971.0, + "step": 7627 + }, + { + "epoch": 0.9703600050884111, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.473031044006348, + "learning_rate": 1e-06, + "loss": 0.3738, + "mean_token_accuracy": 0.8756837844848633, + "num_tokens": 296059201.0, + "step": 7628 + }, + { + "epoch": 0.9704872153670017, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.548047065734863, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8777629733085632, + "num_tokens": 296094524.0, + "step": 7629 + }, + { + "epoch": 0.9706144256455922, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.408159255981445, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8772908449172974, + "num_tokens": 296139407.0, + "step": 7630 + }, + { + "epoch": 0.9707416359241827, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.515043258666992, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8735074996948242, + "num_tokens": 296173926.0, + "step": 7631 + }, + { + "epoch": 0.9708688462027731, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.519237518310547, + "learning_rate": 1e-06, + "loss": 0.4162, + "mean_token_accuracy": 0.8591830730438232, + "num_tokens": 296215748.0, + "step": 7632 + }, + { + "epoch": 0.9709960564813637, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.514730453491211, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8786812424659729, + "num_tokens": 296249788.0, + "step": 7633 + }, + { + "epoch": 0.9711232667599542, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.461347579956055, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8816304802894592, + "num_tokens": 296288335.0, + "step": 7634 + }, + { + "epoch": 0.9712504770385447, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.458617210388184, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8796457052230835, + "num_tokens": 296334199.0, + "step": 7635 + }, + { + "epoch": 0.9713776873171353, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.512155532836914, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8670716285705566, + "num_tokens": 296372770.0, + "step": 7636 + }, + { + "epoch": 0.9715048975957258, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.462258338928223, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8872079849243164, + "num_tokens": 296411524.0, + "step": 7637 + }, + { + "epoch": 0.9716321078743162, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.530806541442871, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8702421188354492, + "num_tokens": 296452675.0, + "step": 7638 + }, + { + "epoch": 0.9717593181529067, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.314018249511719e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.363110542297363, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8753304481506348, + "num_tokens": 296496686.0, + "step": 7639 + }, + { + "epoch": 0.9718865284314973, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.484418869018555, + "learning_rate": 1e-06, + "loss": 0.3576, + "mean_token_accuracy": 0.8818109035491943, + "num_tokens": 296539464.0, + "step": 7640 + }, + { + "epoch": 0.9720137387100878, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.410661697387695, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8808897733688354, + "num_tokens": 296578834.0, + "step": 7641 + }, + { + "epoch": 0.9721409489886783, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.434786796569824, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8845679759979248, + "num_tokens": 296618734.0, + "step": 7642 + }, + { + "epoch": 0.9722681592672688, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.443437576293945, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8948230743408203, + "num_tokens": 296654192.0, + "step": 7643 + }, + { + "epoch": 0.9723953695458593, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.414937019348145, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8855116367340088, + "num_tokens": 296694795.0, + "step": 7644 + }, + { + "epoch": 0.9725225798244498, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.503155708312988, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8823809027671814, + "num_tokens": 296732714.0, + "step": 7645 + }, + { + "epoch": 0.9726497901030403, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.409536361694336, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8803321123123169, + "num_tokens": 296777372.0, + "step": 7646 + }, + { + "epoch": 0.9727770003816308, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.641839981079102, + "learning_rate": 1e-06, + "loss": 0.3893, + "mean_token_accuracy": 0.8719437122344971, + "num_tokens": 296810802.0, + "step": 7647 + }, + { + "epoch": 0.9729042106602214, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.404695510864258, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8858305215835571, + "num_tokens": 296851868.0, + "step": 7648 + }, + { + "epoch": 0.9730314209388119, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.44202995300293, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8827053308486938, + "num_tokens": 296890607.0, + "step": 7649 + }, + { + "epoch": 0.9731586312174023, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.442207336425781, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8985905051231384, + "num_tokens": 296926036.0, + "step": 7650 + }, + { + "epoch": 0.9732858414959928, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.449804306030273, + "learning_rate": 1e-06, + "loss": 0.3854, + "mean_token_accuracy": 0.8732162117958069, + "num_tokens": 296966616.0, + "step": 7651 + }, + { + "epoch": 0.9734130517745834, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.447400093078613, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8879457116127014, + "num_tokens": 297005855.0, + "step": 7652 + }, + { + "epoch": 0.9735402620531739, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.605422973632812, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8753359317779541, + "num_tokens": 297037139.0, + "step": 7653 + }, + { + "epoch": 0.9736674723317644, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.395733833312988, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8861798644065857, + "num_tokens": 297076146.0, + "step": 7654 + }, + { + "epoch": 0.973794682610355, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.456585884094238, + "learning_rate": 1e-06, + "loss": 0.4002, + "mean_token_accuracy": 0.863244891166687, + "num_tokens": 297117959.0, + "step": 7655 + }, + { + "epoch": 0.9739218928889454, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.448230743408203, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8800497055053711, + "num_tokens": 297158472.0, + "step": 7656 + }, + { + "epoch": 0.9740491031675359, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.411581993103027, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.8670544624328613, + "num_tokens": 297201686.0, + "step": 7657 + }, + { + "epoch": 0.9741763134461264, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.498106002807617, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8870844841003418, + "num_tokens": 297241415.0, + "step": 7658 + }, + { + "epoch": 0.974303523724717, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.543302536010742, + "learning_rate": 1e-06, + "loss": 0.3797, + "mean_token_accuracy": 0.8701361417770386, + "num_tokens": 297276138.0, + "step": 7659 + }, + { + "epoch": 0.9744307340033075, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.471379280090332, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8775661587715149, + "num_tokens": 297319153.0, + "step": 7660 + }, + { + "epoch": 0.974557944281898, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.504528999328613, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8818216919898987, + "num_tokens": 297355840.0, + "step": 7661 + }, + { + "epoch": 0.9746851545604885, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.407767295837402, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8898301124572754, + "num_tokens": 297400025.0, + "step": 7662 + }, + { + "epoch": 0.974812364839079, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.405912399291992, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8812300562858582, + "num_tokens": 297440539.0, + "step": 7663 + }, + { + "epoch": 0.9749395751176695, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.602387428283691, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.871603786945343, + "num_tokens": 297475424.0, + "step": 7664 + }, + { + "epoch": 0.97506678539626, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.375565528869629, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8868918418884277, + "num_tokens": 297518450.0, + "step": 7665 + }, + { + "epoch": 0.9751939956748505, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.492156028747559, + "learning_rate": 1e-06, + "loss": 0.3788, + "mean_token_accuracy": 0.8713718056678772, + "num_tokens": 297560267.0, + "step": 7666 + }, + { + "epoch": 0.9753212059534411, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.479227066040039, + "learning_rate": 1e-06, + "loss": 0.3895, + "mean_token_accuracy": 0.8663471937179565, + "num_tokens": 297598800.0, + "step": 7667 + }, + { + "epoch": 0.9754484162320316, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.510873794555664, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8867155313491821, + "num_tokens": 297629651.0, + "step": 7668 + }, + { + "epoch": 0.975575626510622, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.505715370178223, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8848382830619812, + "num_tokens": 297666481.0, + "step": 7669 + }, + { + "epoch": 0.9757028367892125, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.467595100402832, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8772578835487366, + "num_tokens": 297704297.0, + "step": 7670 + }, + { + "epoch": 0.9758300470678031, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.529692649841309, + "learning_rate": 1e-06, + "loss": 0.4043, + "mean_token_accuracy": 0.8641414642333984, + "num_tokens": 297743439.0, + "step": 7671 + }, + { + "epoch": 0.9759572573463936, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.539241790771484, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8787057399749756, + "num_tokens": 297776070.0, + "step": 7672 + }, + { + "epoch": 0.9760844676249841, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.468205451965332, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8751530647277832, + "num_tokens": 297819417.0, + "step": 7673 + }, + { + "epoch": 0.9762116779035747, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.511724472045898, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8901641368865967, + "num_tokens": 297854248.0, + "step": 7674 + }, + { + "epoch": 0.9763388881821651, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.556303024291992, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8774445056915283, + "num_tokens": 297887630.0, + "step": 7675 + }, + { + "epoch": 0.9764660984607556, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.393336296081543, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8887864351272583, + "num_tokens": 297930455.0, + "step": 7676 + }, + { + "epoch": 0.9765933087393461, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.442085266113281, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8758096694946289, + "num_tokens": 297975628.0, + "step": 7677 + }, + { + "epoch": 0.9767205190179367, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.52114486694336, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8814177513122559, + "num_tokens": 298015614.0, + "step": 7678 + }, + { + "epoch": 0.9768477292965272, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.485260963439941, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8783048391342163, + "num_tokens": 298052661.0, + "step": 7679 + }, + { + "epoch": 0.9769749395751177, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.444533348083496, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8912264704704285, + "num_tokens": 298094652.0, + "step": 7680 + }, + { + "epoch": 0.9771021498537081, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.556291580200195, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8834558129310608, + "num_tokens": 298129421.0, + "step": 7681 + }, + { + "epoch": 0.9772293601322987, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.433354377746582, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8882583975791931, + "num_tokens": 298169628.0, + "step": 7682 + }, + { + "epoch": 0.9773565704108892, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.546142578125, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.877483069896698, + "num_tokens": 298206094.0, + "step": 7683 + }, + { + "epoch": 0.9774837806894797, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.423625946044922, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8837813138961792, + "num_tokens": 298246865.0, + "step": 7684 + }, + { + "epoch": 0.9776109909680702, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.557599067687988, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8734939694404602, + "num_tokens": 298285717.0, + "step": 7685 + }, + { + "epoch": 0.9777382012466608, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.460099220275879, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8782232403755188, + "num_tokens": 298332713.0, + "step": 7686 + }, + { + "epoch": 0.9778654115252512, + "ewc_loss": 0.0125732421875, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.238719940185547e-06, + "grad_norm": 8.540581703186035, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8789531588554382, + "num_tokens": 298368283.0, + "step": 7687 + }, + { + "epoch": 0.9779926218038417, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.49169635772705, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8870174288749695, + "num_tokens": 298411254.0, + "step": 7688 + }, + { + "epoch": 0.9781198320824323, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.626241683959961, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.8719276785850525, + "num_tokens": 298443587.0, + "step": 7689 + }, + { + "epoch": 0.9782470423610228, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.471946716308594, + "learning_rate": 1e-06, + "loss": 0.4101, + "mean_token_accuracy": 0.8682817220687866, + "num_tokens": 298488923.0, + "step": 7690 + }, + { + "epoch": 0.9783742526396133, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.429471015930176, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8933289051055908, + "num_tokens": 298527866.0, + "step": 7691 + }, + { + "epoch": 0.9785014629182038, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.337860107421875e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.57149600982666, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8825957179069519, + "num_tokens": 298565350.0, + "step": 7692 + }, + { + "epoch": 0.9786286731967943, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.494711875915527, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8897855281829834, + "num_tokens": 298602784.0, + "step": 7693 + }, + { + "epoch": 0.9787558834753848, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.48587417602539, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8824071884155273, + "num_tokens": 298644458.0, + "step": 7694 + }, + { + "epoch": 0.9788830937539753, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.580928802490234, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8827286958694458, + "num_tokens": 298679752.0, + "step": 7695 + }, + { + "epoch": 0.9790103040325658, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.512476921081543, + "learning_rate": 1e-06, + "loss": 0.3901, + "mean_token_accuracy": 0.8687840104103088, + "num_tokens": 298718651.0, + "step": 7696 + }, + { + "epoch": 0.9791375143111564, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.575348854064941, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8797028660774231, + "num_tokens": 298756135.0, + "step": 7697 + }, + { + "epoch": 0.9792647245897469, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.55579662322998, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8786876201629639, + "num_tokens": 298788993.0, + "step": 7698 + }, + { + "epoch": 0.9793919348683373, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.484193801879883, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8748236894607544, + "num_tokens": 298831685.0, + "step": 7699 + }, + { + "epoch": 0.9795191451469278, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.532438278198242, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8863786458969116, + "num_tokens": 298867601.0, + "step": 7700 + }, + { + "epoch": 0.9796463554255184, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.54686450958252, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8768686056137085, + "num_tokens": 298905482.0, + "step": 7701 + }, + { + "epoch": 0.9797735657041089, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.502409934997559, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8792622685432434, + "num_tokens": 298946865.0, + "step": 7702 + }, + { + "epoch": 0.9799007759826994, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.531562805175781, + "learning_rate": 1e-06, + "loss": 0.4076, + "mean_token_accuracy": 0.8629423379898071, + "num_tokens": 298988063.0, + "step": 7703 + }, + { + "epoch": 0.98002798626129, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.59678840637207, + "learning_rate": 1e-06, + "loss": 0.3903, + "mean_token_accuracy": 0.8705216646194458, + "num_tokens": 299021353.0, + "step": 7704 + }, + { + "epoch": 0.9801551965398804, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.485756874084473, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8920069932937622, + "num_tokens": 299059625.0, + "step": 7705 + }, + { + "epoch": 0.9802824068184709, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.37667179107666, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8932669758796692, + "num_tokens": 299101781.0, + "step": 7706 + }, + { + "epoch": 0.9804096170970614, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.545145034790039, + "learning_rate": 1e-06, + "loss": 0.3924, + "mean_token_accuracy": 0.8677740097045898, + "num_tokens": 299142012.0, + "step": 7707 + }, + { + "epoch": 0.980536827375652, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.54002571105957, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8891663551330566, + "num_tokens": 299174557.0, + "step": 7708 + }, + { + "epoch": 0.9806640376542425, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.426209449768066, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8871028423309326, + "num_tokens": 299217711.0, + "step": 7709 + }, + { + "epoch": 0.980791247932833, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.62161922454834, + "learning_rate": 1e-06, + "loss": 0.391, + "mean_token_accuracy": 0.8652235269546509, + "num_tokens": 299251116.0, + "step": 7710 + }, + { + "epoch": 0.9809184582114235, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.438640594482422, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8923987150192261, + "num_tokens": 299291432.0, + "step": 7711 + }, + { + "epoch": 0.981045668490014, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.39582347869873, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.876408576965332, + "num_tokens": 299335890.0, + "step": 7712 + }, + { + "epoch": 0.9811728787686045, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.563558578491211, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8812482953071594, + "num_tokens": 299369075.0, + "step": 7713 + }, + { + "epoch": 0.981300089047195, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.395793914794922, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8884082436561584, + "num_tokens": 299410725.0, + "step": 7714 + }, + { + "epoch": 0.9814272993257855, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.457796096801758, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8862109184265137, + "num_tokens": 299448899.0, + "step": 7715 + }, + { + "epoch": 0.9815545096043761, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.470290184020996, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8821006417274475, + "num_tokens": 299492534.0, + "step": 7716 + }, + { + "epoch": 0.9816817198829666, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.591038703918457, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8755785226821899, + "num_tokens": 299529682.0, + "step": 7717 + }, + { + "epoch": 0.981808930161557, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.577306747436523, + "learning_rate": 1e-06, + "loss": 0.3768, + "mean_token_accuracy": 0.8725699186325073, + "num_tokens": 299564406.0, + "step": 7718 + }, + { + "epoch": 0.9819361404401475, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.563340187072754, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8837368488311768, + "num_tokens": 299596885.0, + "step": 7719 + }, + { + "epoch": 0.9820633507187381, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.513855934143066, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8815014362335205, + "num_tokens": 299633648.0, + "step": 7720 + }, + { + "epoch": 0.9821905609973286, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.63167667388916, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.877452552318573, + "num_tokens": 299666211.0, + "step": 7721 + }, + { + "epoch": 0.9823177712759191, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.467242240905762, + "learning_rate": 1e-06, + "loss": 0.3877, + "mean_token_accuracy": 0.8704841136932373, + "num_tokens": 299712359.0, + "step": 7722 + }, + { + "epoch": 0.9824449815545097, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.542396545410156, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.875761866569519, + "num_tokens": 299748399.0, + "step": 7723 + }, + { + "epoch": 0.9825721918331001, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.54166030883789, + "learning_rate": 1e-06, + "loss": 0.4116, + "mean_token_accuracy": 0.8577901124954224, + "num_tokens": 299786326.0, + "step": 7724 + }, + { + "epoch": 0.9826994021116906, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.475545883178711, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8785699009895325, + "num_tokens": 299824774.0, + "step": 7725 + }, + { + "epoch": 0.9828266123902811, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.554147720336914, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8830327391624451, + "num_tokens": 299865875.0, + "step": 7726 + }, + { + "epoch": 0.9829538226688717, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.427902221679688, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8912311792373657, + "num_tokens": 299908108.0, + "step": 7727 + }, + { + "epoch": 0.9830810329474622, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.523117065429688, + "learning_rate": 1e-06, + "loss": 0.3999, + "mean_token_accuracy": 0.8682169914245605, + "num_tokens": 299949796.0, + "step": 7728 + }, + { + "epoch": 0.9832082432260527, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.499618530273438, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8800942897796631, + "num_tokens": 299990473.0, + "step": 7729 + }, + { + "epoch": 0.9833354535046431, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.456877708435059, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8788368701934814, + "num_tokens": 300030951.0, + "step": 7730 + }, + { + "epoch": 0.9834626637832337, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.52255630493164, + "learning_rate": 1e-06, + "loss": 0.3895, + "mean_token_accuracy": 0.8698520660400391, + "num_tokens": 300076015.0, + "step": 7731 + }, + { + "epoch": 0.9835898740618242, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.461281776428223, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8849062919616699, + "num_tokens": 300116944.0, + "step": 7732 + }, + { + "epoch": 0.9837170843404147, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.618788719177246, + "learning_rate": 1e-06, + "loss": 0.4242, + "mean_token_accuracy": 0.8567993640899658, + "num_tokens": 300153762.0, + "step": 7733 + }, + { + "epoch": 0.9838442946190052, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.437606811523438, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8795685768127441, + "num_tokens": 300197931.0, + "step": 7734 + }, + { + "epoch": 0.9839715048975958, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.490470886230469, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8832294940948486, + "num_tokens": 300233164.0, + "step": 7735 + }, + { + "epoch": 0.9840987151761862, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.506424903869629, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8763752579689026, + "num_tokens": 300269954.0, + "step": 7736 + }, + { + "epoch": 0.9842259254547767, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.538999557495117, + "learning_rate": 1e-06, + "loss": 0.4338, + "mean_token_accuracy": 0.8559384346008301, + "num_tokens": 300304546.0, + "step": 7737 + }, + { + "epoch": 0.9843531357333672, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.4276762008667, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8806312084197998, + "num_tokens": 300346483.0, + "step": 7738 + }, + { + "epoch": 0.9844803460119578, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.468276977539062, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.895447850227356, + "num_tokens": 300384687.0, + "step": 7739 + }, + { + "epoch": 0.9846075562905483, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.51025676727295, + "learning_rate": 1e-06, + "loss": 0.3987, + "mean_token_accuracy": 0.8654730319976807, + "num_tokens": 300424771.0, + "step": 7740 + }, + { + "epoch": 0.9847347665691388, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.451068878173828, + "learning_rate": 1e-06, + "loss": 0.3631, + "mean_token_accuracy": 0.8760718703269958, + "num_tokens": 300466074.0, + "step": 7741 + }, + { + "epoch": 0.9848619768477292, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.483471870422363, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8751064538955688, + "num_tokens": 300506358.0, + "step": 7742 + }, + { + "epoch": 0.9849891871263198, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.454663276672363, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.882646918296814, + "num_tokens": 300543962.0, + "step": 7743 + }, + { + "epoch": 0.9851163974049103, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.480587005615234, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8781814575195312, + "num_tokens": 300585226.0, + "step": 7744 + }, + { + "epoch": 0.9852436076835008, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.427865028381348, + "learning_rate": 1e-06, + "loss": 0.3724, + "mean_token_accuracy": 0.873849630355835, + "num_tokens": 300631327.0, + "step": 7745 + }, + { + "epoch": 0.9853708179620914, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.54863166809082, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8775013089179993, + "num_tokens": 300665689.0, + "step": 7746 + }, + { + "epoch": 0.9854980282406819, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.42574691772461, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8915053606033325, + "num_tokens": 300709083.0, + "step": 7747 + }, + { + "epoch": 0.9856252385192723, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.447059631347656, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8970208764076233, + "num_tokens": 300747683.0, + "step": 7748 + }, + { + "epoch": 0.9857524487978628, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.554795265197754, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8785354495048523, + "num_tokens": 300783229.0, + "step": 7749 + }, + { + "epoch": 0.9858796590764534, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.443041801452637, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8812748193740845, + "num_tokens": 300825593.0, + "step": 7750 + }, + { + "epoch": 0.9860068693550439, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.467842102050781, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8790431022644043, + "num_tokens": 300865988.0, + "step": 7751 + }, + { + "epoch": 0.9861340796336344, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.474130630493164, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8777828216552734, + "num_tokens": 300907728.0, + "step": 7752 + }, + { + "epoch": 0.986261289912225, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.439533233642578, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8788756132125854, + "num_tokens": 300951367.0, + "step": 7753 + }, + { + "epoch": 0.9863885001908154, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.482298851013184, + "learning_rate": 1e-06, + "loss": 0.3854, + "mean_token_accuracy": 0.8671985864639282, + "num_tokens": 300991225.0, + "step": 7754 + }, + { + "epoch": 0.9865157104694059, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.434402465820312, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.894519567489624, + "num_tokens": 301028989.0, + "step": 7755 + }, + { + "epoch": 0.9866429207479964, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.421175956726074, + "learning_rate": 1e-06, + "loss": 0.3858, + "mean_token_accuracy": 0.8674121499061584, + "num_tokens": 301072532.0, + "step": 7756 + }, + { + "epoch": 0.986770131026587, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.446096420288086, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8871555328369141, + "num_tokens": 301112132.0, + "step": 7757 + }, + { + "epoch": 0.9868973413051775, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.46695327758789, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8813136219978333, + "num_tokens": 301153799.0, + "step": 7758 + }, + { + "epoch": 0.987024551583768, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.458561897277832, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8835805058479309, + "num_tokens": 301189554.0, + "step": 7759 + }, + { + "epoch": 0.9871517618623584, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.523186683654785, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8774400949478149, + "num_tokens": 301227829.0, + "step": 7760 + }, + { + "epoch": 0.987278972140949, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.524682998657227, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8740999698638916, + "num_tokens": 301266027.0, + "step": 7761 + }, + { + "epoch": 0.9874061824195395, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.400338172912598, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8871078491210938, + "num_tokens": 301309515.0, + "step": 7762 + }, + { + "epoch": 0.98753339269813, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.493051528930664, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8736587762832642, + "num_tokens": 301352217.0, + "step": 7763 + }, + { + "epoch": 0.9876606029767205, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.466875076293945, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8662471771240234, + "num_tokens": 301397308.0, + "step": 7764 + }, + { + "epoch": 0.9877878132553111, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.506505966186523, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8722174167633057, + "num_tokens": 301436720.0, + "step": 7765 + }, + { + "epoch": 0.9879150235339016, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.567712783813477, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8717104196548462, + "num_tokens": 301471214.0, + "step": 7766 + }, + { + "epoch": 0.988042233812492, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.508747100830078, + "learning_rate": 1e-06, + "loss": 0.3909, + "mean_token_accuracy": 0.8703925609588623, + "num_tokens": 301512012.0, + "step": 7767 + }, + { + "epoch": 0.9881694440910825, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.445823669433594, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8794103264808655, + "num_tokens": 301552482.0, + "step": 7768 + }, + { + "epoch": 0.9882966543696731, + "ewc_loss": 0.01263427734375, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.78223705291748, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8844701051712036, + "num_tokens": 301592227.0, + "step": 7769 + }, + { + "epoch": 0.9884238646482636, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.361701965332031e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.533965110778809, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8759908676147461, + "num_tokens": 301628035.0, + "step": 7770 + }, + { + "epoch": 0.9885510749268541, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.532706260681152, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8737761378288269, + "num_tokens": 301664930.0, + "step": 7771 + }, + { + "epoch": 0.9886782852054447, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.453635215759277, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8862199783325195, + "num_tokens": 301705233.0, + "step": 7772 + }, + { + "epoch": 0.9888054954840351, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.59534740447998, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8759568333625793, + "num_tokens": 301735848.0, + "step": 7773 + }, + { + "epoch": 0.9889327057626256, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.596948623657227, + "learning_rate": 1e-06, + "loss": 0.3811, + "mean_token_accuracy": 0.8706709742546082, + "num_tokens": 301771243.0, + "step": 7774 + }, + { + "epoch": 0.9890599160412161, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.48640251159668, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8813817501068115, + "num_tokens": 301811936.0, + "step": 7775 + }, + { + "epoch": 0.9891871263198067, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.45495319366455, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8793981671333313, + "num_tokens": 301856886.0, + "step": 7776 + }, + { + "epoch": 0.9893143365983972, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.552659034729004, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8705509901046753, + "num_tokens": 301894305.0, + "step": 7777 + }, + { + "epoch": 0.9894415468769877, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.55683422088623, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.878544807434082, + "num_tokens": 301930631.0, + "step": 7778 + }, + { + "epoch": 0.9895687571555781, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.524933815002441, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.882321834564209, + "num_tokens": 301965756.0, + "step": 7779 + }, + { + "epoch": 0.9896959674341687, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.528428077697754, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8781576156616211, + "num_tokens": 302004962.0, + "step": 7780 + }, + { + "epoch": 0.9898231777127592, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.514323234558105, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8809916973114014, + "num_tokens": 302039695.0, + "step": 7781 + }, + { + "epoch": 0.9899503879913497, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.46633529663086, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.882682204246521, + "num_tokens": 302076039.0, + "step": 7782 + }, + { + "epoch": 0.9900775982699402, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.574922561645508, + "learning_rate": 1e-06, + "loss": 0.4141, + "mean_token_accuracy": 0.8604113459587097, + "num_tokens": 302113100.0, + "step": 7783 + }, + { + "epoch": 0.9902048085485308, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.450323104858398, + "learning_rate": 1e-06, + "loss": 0.3922, + "mean_token_accuracy": 0.8701401948928833, + "num_tokens": 302156821.0, + "step": 7784 + }, + { + "epoch": 0.9903320188271212, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.503334045410156, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8923453688621521, + "num_tokens": 302193638.0, + "step": 7785 + }, + { + "epoch": 0.9904592291057117, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.40534782409668, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.8720895051956177, + "num_tokens": 302240611.0, + "step": 7786 + }, + { + "epoch": 0.9905864393843022, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.600248336791992, + "learning_rate": 1e-06, + "loss": 0.4105, + "mean_token_accuracy": 0.8624199628829956, + "num_tokens": 302281263.0, + "step": 7787 + }, + { + "epoch": 0.9907136496628928, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.441949844360352, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.885528564453125, + "num_tokens": 302323880.0, + "step": 7788 + }, + { + "epoch": 0.9908408599414833, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.508706092834473, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8773849010467529, + "num_tokens": 302362693.0, + "step": 7789 + }, + { + "epoch": 0.9909680702200738, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.466499328613281, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.9008596539497375, + "num_tokens": 302401707.0, + "step": 7790 + }, + { + "epoch": 0.9910952804986642, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.523904800415039, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8876738548278809, + "num_tokens": 302438130.0, + "step": 7791 + }, + { + "epoch": 0.9912224907772548, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.648999214172363, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8745898604393005, + "num_tokens": 302474117.0, + "step": 7792 + }, + { + "epoch": 0.9913497010558453, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.394037246704102, + "learning_rate": 1e-06, + "loss": 0.3753, + "mean_token_accuracy": 0.8695019483566284, + "num_tokens": 302521484.0, + "step": 7793 + }, + { + "epoch": 0.9914769113344358, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.556890487670898, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8845033645629883, + "num_tokens": 302558872.0, + "step": 7794 + }, + { + "epoch": 0.9916041216130264, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.670366287231445, + "learning_rate": 1e-06, + "loss": 0.4131, + "mean_token_accuracy": 0.8606088161468506, + "num_tokens": 302593927.0, + "step": 7795 + }, + { + "epoch": 0.9917313318916169, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.498046875, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8728330731391907, + "num_tokens": 302634259.0, + "step": 7796 + }, + { + "epoch": 0.9918585421702073, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.477570533752441, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8865960836410522, + "num_tokens": 302671809.0, + "step": 7797 + }, + { + "epoch": 0.9919857524487978, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.536958694458008, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8845089077949524, + "num_tokens": 302713025.0, + "step": 7798 + }, + { + "epoch": 0.9921129627273884, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.575302124023438, + "learning_rate": 1e-06, + "loss": 0.407, + "mean_token_accuracy": 0.8653998374938965, + "num_tokens": 302749918.0, + "step": 7799 + }, + { + "epoch": 0.9922401730059789, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.509176254272461, + "learning_rate": 1e-06, + "loss": 0.3804, + "mean_token_accuracy": 0.8713275194168091, + "num_tokens": 302795802.0, + "step": 7800 + }, + { + "epoch": 0.9923673832845694, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.44530963897705, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8819567561149597, + "num_tokens": 302838339.0, + "step": 7801 + }, + { + "epoch": 0.9924945935631599, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.396340370178223, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8930615782737732, + "num_tokens": 302878920.0, + "step": 7802 + }, + { + "epoch": 0.9926218038417504, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.50197696685791, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8835839033126831, + "num_tokens": 302916914.0, + "step": 7803 + }, + { + "epoch": 0.9927490141203409, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.466419219970703, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.8764346837997437, + "num_tokens": 302956997.0, + "step": 7804 + }, + { + "epoch": 0.9928762243989314, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.609045028686523, + "learning_rate": 1e-06, + "loss": 0.3892, + "mean_token_accuracy": 0.8660093545913696, + "num_tokens": 302993028.0, + "step": 7805 + }, + { + "epoch": 0.993003434677522, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.568450927734375, + "learning_rate": 1e-06, + "loss": 0.4006, + "mean_token_accuracy": 0.8644834160804749, + "num_tokens": 303027829.0, + "step": 7806 + }, + { + "epoch": 0.9931306449561125, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.44482135772705, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8827619552612305, + "num_tokens": 303067868.0, + "step": 7807 + }, + { + "epoch": 0.993257855234703, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.545858383178711, + "learning_rate": 1e-06, + "loss": 0.3654, + "mean_token_accuracy": 0.8755043745040894, + "num_tokens": 303105347.0, + "step": 7808 + }, + { + "epoch": 0.9933850655132934, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.547955513000488, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.8778765201568604, + "num_tokens": 303146522.0, + "step": 7809 + }, + { + "epoch": 0.993512275791884, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.439281463623047, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8806763887405396, + "num_tokens": 303193360.0, + "step": 7810 + }, + { + "epoch": 0.9936394860704745, + "ewc_loss": 0.0126953125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.298324584960938e-06, + "grad_norm": 8.498050689697266, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8796933889389038, + "num_tokens": 303231469.0, + "step": 7811 + }, + { + "epoch": 0.993766696349065, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.587770462036133, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8732858300209045, + "num_tokens": 303270256.0, + "step": 7812 + }, + { + "epoch": 0.9938939066276555, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.522307395935059, + "learning_rate": 1e-06, + "loss": 0.3905, + "mean_token_accuracy": 0.866878092288971, + "num_tokens": 303311689.0, + "step": 7813 + }, + { + "epoch": 0.9940211169062461, + "ewc_loss": 0.01275634765625, + "ewc_loss_diag": 3.409385681152344e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.496524810791016, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.872839093208313, + "num_tokens": 303355436.0, + "step": 7814 + }, + { + "epoch": 0.9941483271848366, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.471990585327148, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8784605264663696, + "num_tokens": 303397119.0, + "step": 7815 + }, + { + "epoch": 0.994275537463427, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.504124641418457, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8747438192367554, + "num_tokens": 303434483.0, + "step": 7816 + }, + { + "epoch": 0.9944027477420175, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.586119651794434, + "learning_rate": 1e-06, + "loss": 0.415, + "mean_token_accuracy": 0.8593980073928833, + "num_tokens": 303471483.0, + "step": 7817 + }, + { + "epoch": 0.9945299580206081, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.5822114944458, + "learning_rate": 1e-06, + "loss": 0.3848, + "mean_token_accuracy": 0.8709187507629395, + "num_tokens": 303506796.0, + "step": 7818 + }, + { + "epoch": 0.9946571682991986, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.505894660949707, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.8706305027008057, + "num_tokens": 303545829.0, + "step": 7819 + }, + { + "epoch": 0.9947843785777891, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.532929420471191, + "learning_rate": 1e-06, + "loss": 0.3475, + "mean_token_accuracy": 0.8833621740341187, + "num_tokens": 303582679.0, + "step": 7820 + }, + { + "epoch": 0.9949115888563796, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.480847358703613, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.881568193435669, + "num_tokens": 303620792.0, + "step": 7821 + }, + { + "epoch": 0.9950387991349701, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.544126510620117, + "learning_rate": 1e-06, + "loss": 0.3998, + "mean_token_accuracy": 0.8655394315719604, + "num_tokens": 303658479.0, + "step": 7822 + }, + { + "epoch": 0.9951660094135606, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.518365859985352, + "learning_rate": 1e-06, + "loss": 0.4275, + "mean_token_accuracy": 0.8556185960769653, + "num_tokens": 303703305.0, + "step": 7823 + }, + { + "epoch": 0.9952932196921511, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.454874038696289, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8692269325256348, + "num_tokens": 303747455.0, + "step": 7824 + }, + { + "epoch": 0.9954204299707416, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.638603210449219, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8696094155311584, + "num_tokens": 303781017.0, + "step": 7825 + }, + { + "epoch": 0.9955476402493322, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.450075149536133, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8768727779388428, + "num_tokens": 303826083.0, + "step": 7826 + }, + { + "epoch": 0.9956748505279227, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.500227928161621, + "learning_rate": 1e-06, + "loss": 0.4038, + "mean_token_accuracy": 0.8653733730316162, + "num_tokens": 303864440.0, + "step": 7827 + }, + { + "epoch": 0.9958020608065131, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.597962379455566, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8765742778778076, + "num_tokens": 303897075.0, + "step": 7828 + }, + { + "epoch": 0.9959292710851037, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.429173469543457, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8796847462654114, + "num_tokens": 303940888.0, + "step": 7829 + }, + { + "epoch": 0.9960564813636942, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.438162803649902, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.879432737827301, + "num_tokens": 303980890.0, + "step": 7830 + }, + { + "epoch": 0.9961836916422847, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.47605037689209, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8990451097488403, + "num_tokens": 304019981.0, + "step": 7831 + }, + { + "epoch": 0.9963109019208752, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.603918075561523, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8802484273910522, + "num_tokens": 304053946.0, + "step": 7832 + }, + { + "epoch": 0.9964381121994658, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.520527839660645, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8753979802131653, + "num_tokens": 304090109.0, + "step": 7833 + }, + { + "epoch": 0.9965653224780562, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.620367050170898, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.8759951591491699, + "num_tokens": 304121413.0, + "step": 7834 + }, + { + "epoch": 0.9966925327566467, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.45163631439209, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8716689348220825, + "num_tokens": 304160315.0, + "step": 7835 + }, + { + "epoch": 0.9968197430352372, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.52799129486084, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8901565074920654, + "num_tokens": 304193847.0, + "step": 7836 + }, + { + "epoch": 0.9969469533138278, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.509017944335938, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8741185665130615, + "num_tokens": 304234750.0, + "step": 7837 + }, + { + "epoch": 0.9970741635924183, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.564924240112305, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8868345618247986, + "num_tokens": 304268914.0, + "step": 7838 + }, + { + "epoch": 0.9972013738710088, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.575875282287598, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.8669639825820923, + "num_tokens": 304306323.0, + "step": 7839 + }, + { + "epoch": 0.9973285841495992, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.450080871582031, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8818686008453369, + "num_tokens": 304348243.0, + "step": 7840 + }, + { + "epoch": 0.9974557944281898, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.549886703491211, + "learning_rate": 1e-06, + "loss": 0.4182, + "mean_token_accuracy": 0.8590822219848633, + "num_tokens": 304386959.0, + "step": 7841 + }, + { + "epoch": 0.9975830047067803, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.471172332763672, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8869165182113647, + "num_tokens": 304425649.0, + "step": 7842 + }, + { + "epoch": 0.9977102149853708, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.542512893676758, + "learning_rate": 1e-06, + "loss": 0.3752, + "mean_token_accuracy": 0.8738495111465454, + "num_tokens": 304467908.0, + "step": 7843 + }, + { + "epoch": 0.9978374252639614, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.536508560180664, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8855835199356079, + "num_tokens": 304504425.0, + "step": 7844 + }, + { + "epoch": 0.9979646355425519, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.40916633605957, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8870542049407959, + "num_tokens": 304544849.0, + "step": 7845 + }, + { + "epoch": 0.9980918458211423, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.560590744018555, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9042829275131226, + "num_tokens": 304580153.0, + "step": 7846 + }, + { + "epoch": 0.9982190560997328, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.556754112243652, + "learning_rate": 1e-06, + "loss": 0.3597, + "mean_token_accuracy": 0.8790149688720703, + "num_tokens": 304613733.0, + "step": 7847 + }, + { + "epoch": 0.9983462663783234, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.544849395751953, + "learning_rate": 1e-06, + "loss": 0.4068, + "mean_token_accuracy": 0.8608983159065247, + "num_tokens": 304656873.0, + "step": 7848 + }, + { + "epoch": 0.9984734766569139, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.522590637207031, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8822913765907288, + "num_tokens": 304696138.0, + "step": 7849 + }, + { + "epoch": 0.9986006869355044, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.440789222717285, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8789526224136353, + "num_tokens": 304742026.0, + "step": 7850 + }, + { + "epoch": 0.9987278972140949, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.489179611206055, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8919987082481384, + "num_tokens": 304780263.0, + "step": 7851 + }, + { + "epoch": 0.9988551074926854, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.566903114318848, + "learning_rate": 1e-06, + "loss": 0.3974, + "mean_token_accuracy": 0.8669639825820923, + "num_tokens": 304817842.0, + "step": 7852 + }, + { + "epoch": 0.9989823177712759, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.528058052062988, + "learning_rate": 1e-06, + "loss": 0.3774, + "mean_token_accuracy": 0.8759554624557495, + "num_tokens": 304854939.0, + "step": 7853 + }, + { + "epoch": 0.9991095280498664, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.453486442565918, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8898115158081055, + "num_tokens": 304892041.0, + "step": 7854 + }, + { + "epoch": 0.9992367383284569, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.44002914428711, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.879328191280365, + "num_tokens": 304931467.0, + "step": 7855 + }, + { + "epoch": 0.9993639486070475, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.526765823364258, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8754921555519104, + "num_tokens": 304972372.0, + "step": 7856 + }, + { + "epoch": 0.999491158885638, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.509397506713867, + "learning_rate": 1e-06, + "loss": 0.3943, + "mean_token_accuracy": 0.8668830394744873, + "num_tokens": 305013567.0, + "step": 7857 + }, + { + "epoch": 0.9996183691642284, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.51767349243164, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8899878859519958, + "num_tokens": 305049912.0, + "step": 7858 + }, + { + "epoch": 0.9997455794428189, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.505992889404297, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8870272636413574, + "num_tokens": 305085010.0, + "step": 7859 + }, + { + "epoch": 0.9998727897214095, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.586355209350586, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8782707452774048, + "num_tokens": 305121764.0, + "step": 7860 + }, + { + "epoch": 1.0, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.545458793640137, + "learning_rate": 1e-06, + "loss": 0.4277, + "mean_token_accuracy": 0.8610551357269287, + "num_tokens": 305159571.0, + "step": 7861 + }, + { + "epoch": 1.0001272102785905, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.479997634887695, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8850845098495483, + "num_tokens": 305200690.0, + "step": 7862 + }, + { + "epoch": 1.000254420557181, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.436365127563477, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8823286294937134, + "num_tokens": 305241984.0, + "step": 7863 + }, + { + "epoch": 1.0003816308357716, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.460022926330566, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8956584334373474, + "num_tokens": 305279635.0, + "step": 7864 + }, + { + "epoch": 1.0005088411143621, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.518078804016113, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8815644383430481, + "num_tokens": 305316936.0, + "step": 7865 + }, + { + "epoch": 1.0006360513929526, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.451812744140625, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8978737592697144, + "num_tokens": 305355303.0, + "step": 7866 + }, + { + "epoch": 1.0007632616715432, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.594269752502441, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8960952162742615, + "num_tokens": 305392456.0, + "step": 7867 + }, + { + "epoch": 1.0008904719501335, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.448881149291992, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8889545202255249, + "num_tokens": 305437077.0, + "step": 7868 + }, + { + "epoch": 1.001017682228724, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.611327171325684, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8868914246559143, + "num_tokens": 305473424.0, + "step": 7869 + }, + { + "epoch": 1.0011448925073145, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.543282508850098, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8791536092758179, + "num_tokens": 305514322.0, + "step": 7870 + }, + { + "epoch": 1.001272102785905, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.457069396972656e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.529006958007812, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8904547691345215, + "num_tokens": 305552403.0, + "step": 7871 + }, + { + "epoch": 1.0013993130644956, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.687582015991211, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8753914833068848, + "num_tokens": 305586008.0, + "step": 7872 + }, + { + "epoch": 1.0015265233430861, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.551514625549316, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8866052627563477, + "num_tokens": 305618563.0, + "step": 7873 + }, + { + "epoch": 1.0016537336216766, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.562224388122559, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8784384727478027, + "num_tokens": 305657944.0, + "step": 7874 + }, + { + "epoch": 1.0017809439002672, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.410160064697266, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9021329879760742, + "num_tokens": 305699864.0, + "step": 7875 + }, + { + "epoch": 1.0019081541788577, + "ewc_loss": 0.0128173828125, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.357929229736328e-06, + "grad_norm": 8.454996109008789, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8873275518417358, + "num_tokens": 305742399.0, + "step": 7876 + }, + { + "epoch": 1.0020353644574482, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.555489540100098, + "learning_rate": 1e-06, + "loss": 0.3963, + "mean_token_accuracy": 0.8640185594558716, + "num_tokens": 305784339.0, + "step": 7877 + }, + { + "epoch": 1.0021625747360388, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.559463500976562, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8981192111968994, + "num_tokens": 305815304.0, + "step": 7878 + }, + { + "epoch": 1.0022897850146293, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.537849426269531, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8789953589439392, + "num_tokens": 305859130.0, + "step": 7879 + }, + { + "epoch": 1.0024169952932196, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.59980297088623, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8845614194869995, + "num_tokens": 305892703.0, + "step": 7880 + }, + { + "epoch": 1.0025442055718101, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.491424560546875, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.876977801322937, + "num_tokens": 305932532.0, + "step": 7881 + }, + { + "epoch": 1.0026714158504006, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.540348052978516, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8860450983047485, + "num_tokens": 305974610.0, + "step": 7882 + }, + { + "epoch": 1.0027986261289912, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.561293601989746, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8965764045715332, + "num_tokens": 306010738.0, + "step": 7883 + }, + { + "epoch": 1.0029258364075817, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.565536499023438, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8884528279304504, + "num_tokens": 306045544.0, + "step": 7884 + }, + { + "epoch": 1.0030530466861722, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.522489547729492, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8907418251037598, + "num_tokens": 306081236.0, + "step": 7885 + }, + { + "epoch": 1.0031802569647628, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.487606048583984, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8916621208190918, + "num_tokens": 306121796.0, + "step": 7886 + }, + { + "epoch": 1.0033074672433533, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.504891395568848, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8910226821899414, + "num_tokens": 306160689.0, + "step": 7887 + }, + { + "epoch": 1.0034346775219438, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.626197814941406, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8811889886856079, + "num_tokens": 306198332.0, + "step": 7888 + }, + { + "epoch": 1.0035618878005343, + "ewc_loss": 0.01300048828125, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.567276000976562, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.891892671585083, + "num_tokens": 306235506.0, + "step": 7889 + }, + { + "epoch": 1.0036890980791249, + "ewc_loss": 0.01287841796875, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.50222110748291, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8899219036102295, + "num_tokens": 306275747.0, + "step": 7890 + }, + { + "epoch": 1.0038163083577154, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.504753112792969e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.509282112121582, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.883543848991394, + "num_tokens": 306323604.0, + "step": 7891 + }, + { + "epoch": 1.0039435186363057, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.504753112792969e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.537056922912598, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.878771185874939, + "num_tokens": 306362235.0, + "step": 7892 + }, + { + "epoch": 1.0040707289148962, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.504753112792969e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.566197395324707, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8853528499603271, + "num_tokens": 306403394.0, + "step": 7893 + }, + { + "epoch": 1.0041979391934868, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.504753112792969e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.481870651245117, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8889660835266113, + "num_tokens": 306445418.0, + "step": 7894 + }, + { + "epoch": 1.0043251494720773, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.504753112792969e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.533278465270996, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8857847452163696, + "num_tokens": 306486725.0, + "step": 7895 + }, + { + "epoch": 1.0044523597506678, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.504753112792969e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.579903602600098, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8857893347740173, + "num_tokens": 306522880.0, + "step": 7896 + }, + { + "epoch": 1.0045795700292584, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.504753112792969e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.657857894897461, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8936920166015625, + "num_tokens": 306554626.0, + "step": 7897 + }, + { + "epoch": 1.0047067803078489, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.504753112792969e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.570656776428223, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8862466812133789, + "num_tokens": 306591555.0, + "step": 7898 + }, + { + "epoch": 1.0048339905864394, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.460009574890137, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8909051418304443, + "num_tokens": 306634014.0, + "step": 7899 + }, + { + "epoch": 1.00496120086503, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.632994651794434, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8884412050247192, + "num_tokens": 306665921.0, + "step": 7900 + }, + { + "epoch": 1.0050884111436205, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.574514389038086, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8838353753089905, + "num_tokens": 306702052.0, + "step": 7901 + }, + { + "epoch": 1.005215621422211, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.548232078552246, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8749428391456604, + "num_tokens": 306743142.0, + "step": 7902 + }, + { + "epoch": 1.0053428317008015, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.495415687561035, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.886410653591156, + "num_tokens": 306783625.0, + "step": 7903 + }, + { + "epoch": 1.0054700419793918, + "ewc_loss": 0.01300048828125, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.542583465576172, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8813108205795288, + "num_tokens": 306823371.0, + "step": 7904 + }, + { + "epoch": 1.0055972522579824, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.434962272644043, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8938903212547302, + "num_tokens": 306865702.0, + "step": 7905 + }, + { + "epoch": 1.0057244625365729, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.551681518554688, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8837786316871643, + "num_tokens": 306904776.0, + "step": 7906 + }, + { + "epoch": 1.0058516728151634, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.623671531677246, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.885633111000061, + "num_tokens": 306938447.0, + "step": 7907 + }, + { + "epoch": 1.005978883093754, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.682467460632324, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8786634802818298, + "num_tokens": 306977781.0, + "step": 7908 + }, + { + "epoch": 1.0061060933723445, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.568697929382324, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8794737458229065, + "num_tokens": 307013306.0, + "step": 7909 + }, + { + "epoch": 1.006233303650935, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.443096160888672, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8848503828048706, + "num_tokens": 307057659.0, + "step": 7910 + }, + { + "epoch": 1.0063605139295255, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.51864242553711, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8918027877807617, + "num_tokens": 307097402.0, + "step": 7911 + }, + { + "epoch": 1.006487724208116, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.475831031799316, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8955191373825073, + "num_tokens": 307137970.0, + "step": 7912 + }, + { + "epoch": 1.0066149344867066, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.536911964416504, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8900645971298218, + "num_tokens": 307177214.0, + "step": 7913 + }, + { + "epoch": 1.006742144765297, + "ewc_loss": 0.01300048828125, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.583306312561035, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8977618217468262, + "num_tokens": 307210290.0, + "step": 7914 + }, + { + "epoch": 1.0068693550438876, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.584397315979004, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8821790218353271, + "num_tokens": 307249671.0, + "step": 7915 + }, + { + "epoch": 1.0069965653224782, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.61122989654541, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8926554918289185, + "num_tokens": 307286942.0, + "step": 7916 + }, + { + "epoch": 1.0071237756010685, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.567892074584961, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8841229677200317, + "num_tokens": 307327510.0, + "step": 7917 + }, + { + "epoch": 1.007250985879659, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.571033477783203, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8790690302848816, + "num_tokens": 307371562.0, + "step": 7918 + }, + { + "epoch": 1.0073781961582495, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.67950439453125, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8847604990005493, + "num_tokens": 307409035.0, + "step": 7919 + }, + { + "epoch": 1.00750540643684, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.505684852600098, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8910396099090576, + "num_tokens": 307447014.0, + "step": 7920 + }, + { + "epoch": 1.0076326167154306, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.61008071899414, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8848963975906372, + "num_tokens": 307482690.0, + "step": 7921 + }, + { + "epoch": 1.0077598269940211, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.644649505615234, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8818356990814209, + "num_tokens": 307516399.0, + "step": 7922 + }, + { + "epoch": 1.0078870372726116, + "ewc_loss": 0.01312255859375, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.52302360534668, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8947348594665527, + "num_tokens": 307554648.0, + "step": 7923 + }, + { + "epoch": 1.0080142475512022, + "ewc_loss": 0.01300048828125, + "ewc_loss_diag": 3.4809112548828125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.621940612792969, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.878650963306427, + "num_tokens": 307591367.0, + "step": 7924 + }, + { + "epoch": 1.0081414578297927, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.472274780273438, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.90289705991745, + "num_tokens": 307631004.0, + "step": 7925 + }, + { + "epoch": 1.0082686681083832, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.541410446166992, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8812642693519592, + "num_tokens": 307673042.0, + "step": 7926 + }, + { + "epoch": 1.0083958783869738, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.552878379821777, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8966977000236511, + "num_tokens": 307712369.0, + "step": 7927 + }, + { + "epoch": 1.0085230886655643, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.71181869506836, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8882477283477783, + "num_tokens": 307748120.0, + "step": 7928 + }, + { + "epoch": 1.0086502989441546, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.561220169067383, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8818694949150085, + "num_tokens": 307786977.0, + "step": 7929 + }, + { + "epoch": 1.0087775092227451, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.528594970703125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.62527084350586, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.888689398765564, + "num_tokens": 307823379.0, + "step": 7930 + }, + { + "epoch": 1.0089047195013356, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.611394882202148, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.895106852054596, + "num_tokens": 307860910.0, + "step": 7931 + }, + { + "epoch": 1.0090319297799262, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.532685279846191, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8883413076400757, + "num_tokens": 307905509.0, + "step": 7932 + }, + { + "epoch": 1.0091591400585167, + "ewc_loss": 0.01300048828125, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.723383903503418, + "learning_rate": 1e-06, + "loss": 0.3937, + "mean_token_accuracy": 0.8677693605422974, + "num_tokens": 307941122.0, + "step": 7933 + }, + { + "epoch": 1.0092863503371072, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.593606948852539, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8946431875228882, + "num_tokens": 307978205.0, + "step": 7934 + }, + { + "epoch": 1.0094135606156978, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.446338653564453, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9022381901741028, + "num_tokens": 308021196.0, + "step": 7935 + }, + { + "epoch": 1.0095407708942883, + "ewc_loss": 0.01300048828125, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.594609260559082, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8747034072875977, + "num_tokens": 308058792.0, + "step": 7936 + }, + { + "epoch": 1.0096679811728788, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.552774429321289, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8952251672744751, + "num_tokens": 308099993.0, + "step": 7937 + }, + { + "epoch": 1.0097951914514693, + "ewc_loss": 0.01312255859375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.645495414733887, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.887665867805481, + "num_tokens": 308135770.0, + "step": 7938 + }, + { + "epoch": 1.0099224017300599, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.55134105682373, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8700475692749023, + "num_tokens": 308179358.0, + "step": 7939 + }, + { + "epoch": 1.0100496120086504, + "ewc_loss": 0.01312255859375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.542319297790527, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8875478506088257, + "num_tokens": 308220101.0, + "step": 7940 + }, + { + "epoch": 1.0101768222872407, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.60877513885498, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8859636187553406, + "num_tokens": 308258467.0, + "step": 7941 + }, + { + "epoch": 1.0103040325658312, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.491207122802734, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8840270042419434, + "num_tokens": 308303411.0, + "step": 7942 + }, + { + "epoch": 1.0104312428444218, + "ewc_loss": 0.01312255859375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.593870162963867, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.876393735408783, + "num_tokens": 308341063.0, + "step": 7943 + }, + { + "epoch": 1.0105584531230123, + "ewc_loss": 0.01312255859375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.47790813446045, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8857513070106506, + "num_tokens": 308388277.0, + "step": 7944 + }, + { + "epoch": 1.0106856634016028, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.563299179077148, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8769425749778748, + "num_tokens": 308429264.0, + "step": 7945 + }, + { + "epoch": 1.0108128736801933, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.653308868408203, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8937738537788391, + "num_tokens": 308462976.0, + "step": 7946 + }, + { + "epoch": 1.0109400839587839, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.518977165222168, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8942204713821411, + "num_tokens": 308502139.0, + "step": 7947 + }, + { + "epoch": 1.0110672942373744, + "ewc_loss": 0.01300048828125, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.533273696899414, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8883973360061646, + "num_tokens": 308542564.0, + "step": 7948 + }, + { + "epoch": 1.011194504515965, + "ewc_loss": 0.01300048828125, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.510690689086914, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8969153761863708, + "num_tokens": 308580916.0, + "step": 7949 + }, + { + "epoch": 1.0113217147945555, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.580251693725586, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.879104733467102, + "num_tokens": 308621747.0, + "step": 7950 + }, + { + "epoch": 1.011448925073146, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.58030891418457, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8827697038650513, + "num_tokens": 308658819.0, + "step": 7951 + }, + { + "epoch": 1.0115761353517365, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.517334938049316, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8819072246551514, + "num_tokens": 308703819.0, + "step": 7952 + }, + { + "epoch": 1.0117033456303268, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.654522895812988, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8872362375259399, + "num_tokens": 308737157.0, + "step": 7953 + }, + { + "epoch": 1.0118305559089174, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.587146759033203, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8971862196922302, + "num_tokens": 308772760.0, + "step": 7954 + }, + { + "epoch": 1.0119577661875079, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.703381538391113, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8864459991455078, + "num_tokens": 308804719.0, + "step": 7955 + }, + { + "epoch": 1.0120849764660984, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.633138656616211, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8855322003364563, + "num_tokens": 308841958.0, + "step": 7956 + }, + { + "epoch": 1.012212186744689, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.586939811706543, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8852366805076599, + "num_tokens": 308877198.0, + "step": 7957 + }, + { + "epoch": 1.0123393970232795, + "ewc_loss": 0.01312255859375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.559937477111816, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8977225422859192, + "num_tokens": 308915431.0, + "step": 7958 + }, + { + "epoch": 1.01246660730187, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.525444984436035, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8869245648384094, + "num_tokens": 308958223.0, + "step": 7959 + }, + { + "epoch": 1.0125938175804605, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.568987846374512, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8875281810760498, + "num_tokens": 308998495.0, + "step": 7960 + }, + { + "epoch": 1.012721027859051, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.59111213684082, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8858091831207275, + "num_tokens": 309036848.0, + "step": 7961 + }, + { + "epoch": 1.0128482381376416, + "ewc_loss": 0.01312255859375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.506601333618164, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8984445333480835, + "num_tokens": 309077782.0, + "step": 7962 + }, + { + "epoch": 1.012975448416232, + "ewc_loss": 0.01312255859375, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.631815910339355, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8792749643325806, + "num_tokens": 309116305.0, + "step": 7963 + }, + { + "epoch": 1.0131026586948226, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.552436828613281e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.610370635986328, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8857678174972534, + "num_tokens": 309156753.0, + "step": 7964 + }, + { + "epoch": 1.0132298689734132, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.5762786865234375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.54296875, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.889071524143219, + "num_tokens": 309200017.0, + "step": 7965 + }, + { + "epoch": 1.0133570792520035, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.5762786865234375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.559657096862793, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8833432197570801, + "num_tokens": 309241196.0, + "step": 7966 + }, + { + "epoch": 1.013484289530594, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.5762786865234375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.640338897705078, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8803205490112305, + "num_tokens": 309281647.0, + "step": 7967 + }, + { + "epoch": 1.0136114998091845, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.5762786865234375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.54796028137207, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8981677293777466, + "num_tokens": 309318458.0, + "step": 7968 + }, + { + "epoch": 1.013738710087775, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.5762786865234375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.555500984191895, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8901818990707397, + "num_tokens": 309357665.0, + "step": 7969 + }, + { + "epoch": 1.0138659203663656, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.5762786865234375e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.654964447021484, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8772541284561157, + "num_tokens": 309394710.0, + "step": 7970 + }, + { + "epoch": 1.013993130644956, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.5762786865234375e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.504007339477539, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.9025793671607971, + "num_tokens": 309432396.0, + "step": 7971 + }, + { + "epoch": 1.0141203409235466, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.5762786865234375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.600744247436523, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8879446983337402, + "num_tokens": 309470196.0, + "step": 7972 + }, + { + "epoch": 1.0142475512021372, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.5762786865234375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.445350646972656, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8873519897460938, + "num_tokens": 309514435.0, + "step": 7973 + }, + { + "epoch": 1.0143747614807277, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.5762786865234375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.595040321350098, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8776800632476807, + "num_tokens": 309553926.0, + "step": 7974 + }, + { + "epoch": 1.0145019717593182, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.5762786865234375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.51679515838623, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9018624424934387, + "num_tokens": 309593731.0, + "step": 7975 + }, + { + "epoch": 1.0146291820379088, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.5762786865234375e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.529999732971191, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8892176747322083, + "num_tokens": 309631577.0, + "step": 7976 + }, + { + "epoch": 1.0147563923164993, + "ewc_loss": 0.012939453125, + "ewc_loss_diag": 3.5762786865234375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.609725952148438, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8906170725822449, + "num_tokens": 309669003.0, + "step": 7977 + }, + { + "epoch": 1.0148836025950896, + "ewc_loss": 0.01312255859375, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.62035083770752, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8888099193572998, + "num_tokens": 309705965.0, + "step": 7978 + }, + { + "epoch": 1.0150108128736801, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.499417304992676, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9046032428741455, + "num_tokens": 309742480.0, + "step": 7979 + }, + { + "epoch": 1.0151380231522706, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.579453468322754, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.885289192199707, + "num_tokens": 309778021.0, + "step": 7980 + }, + { + "epoch": 1.0152652334308612, + "ewc_loss": 0.01300048828125, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.654350280761719, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8799943327903748, + "num_tokens": 309816606.0, + "step": 7981 + }, + { + "epoch": 1.0153924437094517, + "ewc_loss": 0.01312255859375, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.583632469177246, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8867963552474976, + "num_tokens": 309851494.0, + "step": 7982 + }, + { + "epoch": 1.0155196539880422, + "ewc_loss": 0.01300048828125, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.67585277557373, + "learning_rate": 1e-06, + "loss": 0.3963, + "mean_token_accuracy": 0.8662710189819336, + "num_tokens": 309889670.0, + "step": 7983 + }, + { + "epoch": 1.0156468642666328, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.611006736755371, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8823040723800659, + "num_tokens": 309926084.0, + "step": 7984 + }, + { + "epoch": 1.0157740745452233, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.535479545593262, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8943707942962646, + "num_tokens": 309962855.0, + "step": 7985 + }, + { + "epoch": 1.0159012848238138, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.532323837280273, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8887872099876404, + "num_tokens": 310003422.0, + "step": 7986 + }, + { + "epoch": 1.0160284951024043, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.631619453430176, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8927669525146484, + "num_tokens": 310037863.0, + "step": 7987 + }, + { + "epoch": 1.0161557053809949, + "ewc_loss": 0.01312255859375, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.540961265563965, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8822868466377258, + "num_tokens": 310078414.0, + "step": 7988 + }, + { + "epoch": 1.0162829156595854, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.596064567565918, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8841971158981323, + "num_tokens": 310115019.0, + "step": 7989 + }, + { + "epoch": 1.0164101259381757, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.602813720703125, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8895206451416016, + "num_tokens": 310153117.0, + "step": 7990 + }, + { + "epoch": 1.0165373362167662, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.527645111083984, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9042839407920837, + "num_tokens": 310185902.0, + "step": 7991 + }, + { + "epoch": 1.0166645464953568, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.592222213745117, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8945028185844421, + "num_tokens": 310226263.0, + "step": 7992 + }, + { + "epoch": 1.0167917567739473, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.572832107543945, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8859679698944092, + "num_tokens": 310267176.0, + "step": 7993 + }, + { + "epoch": 1.0169189670525378, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.443397521972656, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.9004688262939453, + "num_tokens": 310309327.0, + "step": 7994 + }, + { + "epoch": 1.0170461773311283, + "ewc_loss": 0.01312255859375, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.723345756530762, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8869287967681885, + "num_tokens": 310345074.0, + "step": 7995 + }, + { + "epoch": 1.0171733876097189, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.66263484954834, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8848701119422913, + "num_tokens": 310379950.0, + "step": 7996 + }, + { + "epoch": 1.0173005978883094, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.576309204101562, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8852686882019043, + "num_tokens": 310416412.0, + "step": 7997 + }, + { + "epoch": 1.0174278081669, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.670411109924316, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.8670661449432373, + "num_tokens": 310457289.0, + "step": 7998 + }, + { + "epoch": 1.0175550184454905, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.600120544433594e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.564888000488281, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8965463638305664, + "num_tokens": 310491437.0, + "step": 7999 + }, + { + "epoch": 1.017682228724081, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.61134147644043, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8866261839866638, + "num_tokens": 310525770.0, + "step": 8000 + }, + { + "epoch": 1.0178094390026715, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.572686195373535, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8874064683914185, + "num_tokens": 310567281.0, + "step": 8001 + }, + { + "epoch": 1.0179366492812618, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.637393951416016, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8887847661972046, + "num_tokens": 310597295.0, + "step": 8002 + }, + { + "epoch": 1.0180638595598523, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.487691879272461, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8903995752334595, + "num_tokens": 310643159.0, + "step": 8003 + }, + { + "epoch": 1.0181910698384429, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.653641700744629, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8722361326217651, + "num_tokens": 310681143.0, + "step": 8004 + }, + { + "epoch": 1.0183182801170334, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.499767303466797, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8914494514465332, + "num_tokens": 310720799.0, + "step": 8005 + }, + { + "epoch": 1.018445490395624, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.575439453125, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8868786692619324, + "num_tokens": 310758675.0, + "step": 8006 + }, + { + "epoch": 1.0185727006742145, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.515830993652344, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.892181932926178, + "num_tokens": 310795074.0, + "step": 8007 + }, + { + "epoch": 1.018699910952805, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.581156730651855, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8871903419494629, + "num_tokens": 310831756.0, + "step": 8008 + }, + { + "epoch": 1.0188271212313955, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.601110458374023, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8906691074371338, + "num_tokens": 310867686.0, + "step": 8009 + }, + { + "epoch": 1.018954331509986, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.7139310836792, + "learning_rate": 1e-06, + "loss": 0.4095, + "mean_token_accuracy": 0.8665307760238647, + "num_tokens": 310904372.0, + "step": 8010 + }, + { + "epoch": 1.0190815417885766, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.43681812286377, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8928109407424927, + "num_tokens": 310951036.0, + "step": 8011 + }, + { + "epoch": 1.019208752067167, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.609536170959473, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8801126480102539, + "num_tokens": 310985733.0, + "step": 8012 + }, + { + "epoch": 1.0193359623457576, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.456417083740234, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.9022173881530762, + "num_tokens": 311028553.0, + "step": 8013 + }, + { + "epoch": 1.0194631726243482, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.562461853027344, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8991950750350952, + "num_tokens": 311060532.0, + "step": 8014 + }, + { + "epoch": 1.0195903829029385, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.62172794342041, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8851522207260132, + "num_tokens": 311098810.0, + "step": 8015 + }, + { + "epoch": 1.019717593181529, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.614253044128418, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8727869987487793, + "num_tokens": 311132677.0, + "step": 8016 + }, + { + "epoch": 1.0198448034601195, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.580842018127441, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8784487247467041, + "num_tokens": 311171929.0, + "step": 8017 + }, + { + "epoch": 1.01997201373871, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.563716888427734, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8887635469436646, + "num_tokens": 311209882.0, + "step": 8018 + }, + { + "epoch": 1.0200992240173006, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.498892784118652, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.900863528251648, + "num_tokens": 311247696.0, + "step": 8019 + }, + { + "epoch": 1.020226434295891, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.535834312438965, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8923647999763489, + "num_tokens": 311283783.0, + "step": 8020 + }, + { + "epoch": 1.0203536445744816, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.521004676818848, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8972212672233582, + "num_tokens": 311321875.0, + "step": 8021 + }, + { + "epoch": 1.0204808548530722, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.48701000213623, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.89176344871521, + "num_tokens": 311364820.0, + "step": 8022 + }, + { + "epoch": 1.0206080651316627, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.629319190979004, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.887710690498352, + "num_tokens": 311397003.0, + "step": 8023 + }, + { + "epoch": 1.0207352754102532, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.505510330200195, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8864967226982117, + "num_tokens": 311439565.0, + "step": 8024 + }, + { + "epoch": 1.0208624856888437, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.582870483398438, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.882483720779419, + "num_tokens": 311480496.0, + "step": 8025 + }, + { + "epoch": 1.0209896959674343, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.600773811340332, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8888983130455017, + "num_tokens": 311520404.0, + "step": 8026 + }, + { + "epoch": 1.0211169062460246, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.602799415588379, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8919846415519714, + "num_tokens": 311555977.0, + "step": 8027 + }, + { + "epoch": 1.021244116524615, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.49374008178711, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8835947513580322, + "num_tokens": 311600118.0, + "step": 8028 + }, + { + "epoch": 1.0213713268032056, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.580094337463379, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8852302432060242, + "num_tokens": 311639454.0, + "step": 8029 + }, + { + "epoch": 1.0214985370817962, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.55383586883545, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8922823667526245, + "num_tokens": 311680881.0, + "step": 8030 + }, + { + "epoch": 1.0216257473603867, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.57528018951416, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.898949146270752, + "num_tokens": 311712064.0, + "step": 8031 + }, + { + "epoch": 1.0217529576389772, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.698358535766602, + "learning_rate": 1e-06, + "loss": 0.419, + "mean_token_accuracy": 0.8539215326309204, + "num_tokens": 311753841.0, + "step": 8032 + }, + { + "epoch": 1.0218801679175677, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.748481750488281, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8868533372879028, + "num_tokens": 311790041.0, + "step": 8033 + }, + { + "epoch": 1.0220073781961583, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.573603630065918, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8822070360183716, + "num_tokens": 311827309.0, + "step": 8034 + }, + { + "epoch": 1.0221345884747488, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.479134559631348, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8911260366439819, + "num_tokens": 311873399.0, + "step": 8035 + }, + { + "epoch": 1.0222617987533393, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.635879516601562, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8896569013595581, + "num_tokens": 311905632.0, + "step": 8036 + }, + { + "epoch": 1.0223890090319299, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.62396240234375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.618995666503906, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8958690166473389, + "num_tokens": 311940730.0, + "step": 8037 + }, + { + "epoch": 1.0225162193105204, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.601089477539062, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8813555240631104, + "num_tokens": 311981066.0, + "step": 8038 + }, + { + "epoch": 1.0226434295891107, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.559270858764648, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8869747519493103, + "num_tokens": 312021117.0, + "step": 8039 + }, + { + "epoch": 1.0227706398677012, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.527583122253418, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.9048326015472412, + "num_tokens": 312060442.0, + "step": 8040 + }, + { + "epoch": 1.0228978501462918, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.576703071594238, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8770056366920471, + "num_tokens": 312098421.0, + "step": 8041 + }, + { + "epoch": 1.0230250604248823, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.639242172241211, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.8703073859214783, + "num_tokens": 312135782.0, + "step": 8042 + }, + { + "epoch": 1.0231522707034728, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.535565376281738, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8828495740890503, + "num_tokens": 312177736.0, + "step": 8043 + }, + { + "epoch": 1.0232794809820633, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.866003036499023, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8866037726402283, + "num_tokens": 312211183.0, + "step": 8044 + }, + { + "epoch": 1.0234066912606539, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.507791519165039, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8905438184738159, + "num_tokens": 312249200.0, + "step": 8045 + }, + { + "epoch": 1.0235339015392444, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.60933780670166, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8918167948722839, + "num_tokens": 312281496.0, + "step": 8046 + }, + { + "epoch": 1.023661111817835, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.596416473388672, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8853069543838501, + "num_tokens": 312320049.0, + "step": 8047 + }, + { + "epoch": 1.0237883220964255, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.82880687713623, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8926089406013489, + "num_tokens": 312349385.0, + "step": 8048 + }, + { + "epoch": 1.023915532375016, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.549062728881836, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8903247117996216, + "num_tokens": 312389528.0, + "step": 8049 + }, + { + "epoch": 1.0240427426536065, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.529936790466309, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8846345543861389, + "num_tokens": 312431725.0, + "step": 8050 + }, + { + "epoch": 1.0241699529321968, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.563143730163574, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8956022262573242, + "num_tokens": 312468182.0, + "step": 8051 + }, + { + "epoch": 1.0242971632107873, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.53406047821045, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8822134733200073, + "num_tokens": 312510302.0, + "step": 8052 + }, + { + "epoch": 1.0244243734893779, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.655799865722656, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8911756277084351, + "num_tokens": 312547162.0, + "step": 8053 + }, + { + "epoch": 1.0245515837679684, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.474349021911621, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8900161981582642, + "num_tokens": 312588342.0, + "step": 8054 + }, + { + "epoch": 1.024678794046559, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.646110534667969, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8819069862365723, + "num_tokens": 312625761.0, + "step": 8055 + }, + { + "epoch": 1.0248060043251495, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.601140022277832, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8782748579978943, + "num_tokens": 312663171.0, + "step": 8056 + }, + { + "epoch": 1.02493321460374, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.516828536987305, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8890173435211182, + "num_tokens": 312710291.0, + "step": 8057 + }, + { + "epoch": 1.0250604248823305, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.553104400634766, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8946014642715454, + "num_tokens": 312749057.0, + "step": 8058 + }, + { + "epoch": 1.025187635160921, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.65419864654541, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8851562738418579, + "num_tokens": 312782915.0, + "step": 8059 + }, + { + "epoch": 1.0253148454395116, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.56502628326416, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8921877145767212, + "num_tokens": 312819737.0, + "step": 8060 + }, + { + "epoch": 1.025442055718102, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.588228225708008, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8860576152801514, + "num_tokens": 312860306.0, + "step": 8061 + }, + { + "epoch": 1.0255692659966926, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.648117065429688, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8795450925827026, + "num_tokens": 312897679.0, + "step": 8062 + }, + { + "epoch": 1.0256964762752832, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.53465747833252, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8940554857254028, + "num_tokens": 312937153.0, + "step": 8063 + }, + { + "epoch": 1.0258236865538735, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.532478332519531, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8842850923538208, + "num_tokens": 312977897.0, + "step": 8064 + }, + { + "epoch": 1.025950896832464, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.647804260253906e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.666267395019531, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.875136137008667, + "num_tokens": 313012353.0, + "step": 8065 + }, + { + "epoch": 1.0260781071110545, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.60600757598877, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8837323188781738, + "num_tokens": 313046759.0, + "step": 8066 + }, + { + "epoch": 1.026205317389645, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.536176681518555, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8807423114776611, + "num_tokens": 313087716.0, + "step": 8067 + }, + { + "epoch": 1.0263325276682356, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.607285499572754, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8878176808357239, + "num_tokens": 313123490.0, + "step": 8068 + }, + { + "epoch": 1.026459737946826, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.567959785461426, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8742610216140747, + "num_tokens": 313163146.0, + "step": 8069 + }, + { + "epoch": 1.0265869482254166, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.470739364624023, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8948967456817627, + "num_tokens": 313205060.0, + "step": 8070 + }, + { + "epoch": 1.0267141585040072, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.641546249389648, + "learning_rate": 1e-06, + "loss": 0.3776, + "mean_token_accuracy": 0.8722546100616455, + "num_tokens": 313245576.0, + "step": 8071 + }, + { + "epoch": 1.0268413687825977, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.552726745605469, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8859858512878418, + "num_tokens": 313284669.0, + "step": 8072 + }, + { + "epoch": 1.0269685790611882, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.53252124786377, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.898459792137146, + "num_tokens": 313322486.0, + "step": 8073 + }, + { + "epoch": 1.0270957893397787, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.56896686553955, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8861916065216064, + "num_tokens": 313360601.0, + "step": 8074 + }, + { + "epoch": 1.0272229996183693, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.59142780303955, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8873906135559082, + "num_tokens": 313397896.0, + "step": 8075 + }, + { + "epoch": 1.0273502098969596, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.569586753845215, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8988293409347534, + "num_tokens": 313432976.0, + "step": 8076 + }, + { + "epoch": 1.02747742017555, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.46247386932373, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8949671387672424, + "num_tokens": 313482633.0, + "step": 8077 + }, + { + "epoch": 1.0276046304541406, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.590961456298828, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8851640820503235, + "num_tokens": 313523363.0, + "step": 8078 + }, + { + "epoch": 1.0277318407327312, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.690498352050781, + "learning_rate": 1e-06, + "loss": 0.3922, + "mean_token_accuracy": 0.8748166561126709, + "num_tokens": 313559762.0, + "step": 8079 + }, + { + "epoch": 1.0278590510113217, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.585441589355469, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8869999647140503, + "num_tokens": 313597749.0, + "step": 8080 + }, + { + "epoch": 1.0279862612899122, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.521806716918945, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8738372325897217, + "num_tokens": 313637975.0, + "step": 8081 + }, + { + "epoch": 1.0281134715685027, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.561963081359863, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8995950222015381, + "num_tokens": 313677602.0, + "step": 8082 + }, + { + "epoch": 1.0282406818470933, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.597566604614258, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8903762102127075, + "num_tokens": 313714373.0, + "step": 8083 + }, + { + "epoch": 1.0283678921256838, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.540124893188477, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9049907922744751, + "num_tokens": 313751933.0, + "step": 8084 + }, + { + "epoch": 1.0284951024042743, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.668620109558105, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8871302604675293, + "num_tokens": 313784813.0, + "step": 8085 + }, + { + "epoch": 1.0286223126828649, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.531474113464355, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8895615339279175, + "num_tokens": 313826716.0, + "step": 8086 + }, + { + "epoch": 1.0287495229614554, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.572623252868652, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8828644156455994, + "num_tokens": 313862960.0, + "step": 8087 + }, + { + "epoch": 1.0288767332400457, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.529401779174805, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8937139511108398, + "num_tokens": 313902978.0, + "step": 8088 + }, + { + "epoch": 1.0290039435186362, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.609055519104004, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8809786438941956, + "num_tokens": 313942484.0, + "step": 8089 + }, + { + "epoch": 1.0291311537972267, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.687798500061035, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8721197843551636, + "num_tokens": 313975703.0, + "step": 8090 + }, + { + "epoch": 1.0292583640758173, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.492164611816406, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.891527533531189, + "num_tokens": 314018404.0, + "step": 8091 + }, + { + "epoch": 1.0293855743544078, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.61440372467041, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8870570659637451, + "num_tokens": 314056821.0, + "step": 8092 + }, + { + "epoch": 1.0295127846329983, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.445832252502441, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8924137353897095, + "num_tokens": 314102030.0, + "step": 8093 + }, + { + "epoch": 1.0296399949115889, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.577606201171875, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8890479803085327, + "num_tokens": 314144569.0, + "step": 8094 + }, + { + "epoch": 1.0297672051901794, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.588004112243652, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8806816339492798, + "num_tokens": 314184649.0, + "step": 8095 + }, + { + "epoch": 1.02989441546877, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.556097030639648, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8935199975967407, + "num_tokens": 314219637.0, + "step": 8096 + }, + { + "epoch": 1.0300216257473604, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.611085891723633, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8880054950714111, + "num_tokens": 314261155.0, + "step": 8097 + }, + { + "epoch": 1.030148836025951, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.570919036865234, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8798418045043945, + "num_tokens": 314302388.0, + "step": 8098 + }, + { + "epoch": 1.0302760463045415, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.568488121032715, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8981711268424988, + "num_tokens": 314344218.0, + "step": 8099 + }, + { + "epoch": 1.0304032565831318, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.558418273925781, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8953015804290771, + "num_tokens": 314383198.0, + "step": 8100 + }, + { + "epoch": 1.0305304668617223, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.527772903442383, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8996075391769409, + "num_tokens": 314425326.0, + "step": 8101 + }, + { + "epoch": 1.0306576771403129, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.569637298583984, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9069567322731018, + "num_tokens": 314462355.0, + "step": 8102 + }, + { + "epoch": 1.0307848874189034, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.619693756103516, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8847663402557373, + "num_tokens": 314498319.0, + "step": 8103 + }, + { + "epoch": 1.030912097697494, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.631146430969238, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8756319880485535, + "num_tokens": 314536806.0, + "step": 8104 + }, + { + "epoch": 1.0310393079760845, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.48370361328125, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8874231576919556, + "num_tokens": 314581534.0, + "step": 8105 + }, + { + "epoch": 1.031166518254675, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.545758247375488, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8954371213912964, + "num_tokens": 314620737.0, + "step": 8106 + }, + { + "epoch": 1.0312937285332655, + "ewc_loss": 0.0130615234375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.626340866088867, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8783191442489624, + "num_tokens": 314661594.0, + "step": 8107 + }, + { + "epoch": 1.031420938811856, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.657160758972168, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8743780255317688, + "num_tokens": 314699940.0, + "step": 8108 + }, + { + "epoch": 1.0315481490904466, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.653170585632324, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8862638473510742, + "num_tokens": 314734271.0, + "step": 8109 + }, + { + "epoch": 1.031675359369037, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.548416137695312, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8920391798019409, + "num_tokens": 314775671.0, + "step": 8110 + }, + { + "epoch": 1.0318025696476276, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.579309463500977, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8993767499923706, + "num_tokens": 314814260.0, + "step": 8111 + }, + { + "epoch": 1.0319297799262181, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.64089584350586, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8854180574417114, + "num_tokens": 314854242.0, + "step": 8112 + }, + { + "epoch": 1.0320569902048085, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.784127235412598, + "learning_rate": 1e-06, + "loss": 0.3823, + "mean_token_accuracy": 0.8710561990737915, + "num_tokens": 314889912.0, + "step": 8113 + }, + { + "epoch": 1.032184200483399, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.586755752563477, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.881089448928833, + "num_tokens": 314936334.0, + "step": 8114 + }, + { + "epoch": 1.0323114107619895, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.521089553833008, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8892661929130554, + "num_tokens": 314976733.0, + "step": 8115 + }, + { + "epoch": 1.03243862104058, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.655478477478027, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8756710290908813, + "num_tokens": 315016831.0, + "step": 8116 + }, + { + "epoch": 1.0325658313191706, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.556949615478516, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8916295170783997, + "num_tokens": 315056567.0, + "step": 8117 + }, + { + "epoch": 1.032693041597761, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.564762115478516, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8733365535736084, + "num_tokens": 315101905.0, + "step": 8118 + }, + { + "epoch": 1.0328202518763516, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.62402629852295, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8764603137969971, + "num_tokens": 315142112.0, + "step": 8119 + }, + { + "epoch": 1.0329474621549422, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.517505645751953, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8934804797172546, + "num_tokens": 315180941.0, + "step": 8120 + }, + { + "epoch": 1.0330746724335327, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.649776458740234, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8764960765838623, + "num_tokens": 315217957.0, + "step": 8121 + }, + { + "epoch": 1.0332018827121232, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.75540542602539, + "learning_rate": 1e-06, + "loss": 0.3933, + "mean_token_accuracy": 0.8651310205459595, + "num_tokens": 315249567.0, + "step": 8122 + }, + { + "epoch": 1.0333290929907137, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.581576347351074, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.891395092010498, + "num_tokens": 315285511.0, + "step": 8123 + }, + { + "epoch": 1.0334563032693043, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.551942825317383, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8895922899246216, + "num_tokens": 315323794.0, + "step": 8124 + }, + { + "epoch": 1.0335835135478946, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.667835235595703, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.900730550289154, + "num_tokens": 315356509.0, + "step": 8125 + }, + { + "epoch": 1.033710723826485, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.499521255493164, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8913266062736511, + "num_tokens": 315396542.0, + "step": 8126 + }, + { + "epoch": 1.0338379341050756, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.563735961914062, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9054854512214661, + "num_tokens": 315435497.0, + "step": 8127 + }, + { + "epoch": 1.0339651443836662, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.638833999633789, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8762211799621582, + "num_tokens": 315474378.0, + "step": 8128 + }, + { + "epoch": 1.0340923546622567, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.613129615783691, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8848053812980652, + "num_tokens": 315511472.0, + "step": 8129 + }, + { + "epoch": 1.0342195649408472, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.650084495544434, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8978615999221802, + "num_tokens": 315548754.0, + "step": 8130 + }, + { + "epoch": 1.0343467752194377, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.630342483520508, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.871157169342041, + "num_tokens": 315585593.0, + "step": 8131 + }, + { + "epoch": 1.0344739854980283, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.573098182678223, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8814022541046143, + "num_tokens": 315627470.0, + "step": 8132 + }, + { + "epoch": 1.0346011957766188, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.653576850891113, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8739315271377563, + "num_tokens": 315664428.0, + "step": 8133 + }, + { + "epoch": 1.0347284060552093, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.778890609741211, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8795543313026428, + "num_tokens": 315694178.0, + "step": 8134 + }, + { + "epoch": 1.0348556163337999, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.597261428833008, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8799076080322266, + "num_tokens": 315735358.0, + "step": 8135 + }, + { + "epoch": 1.0349828266123904, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.695487976074219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.687467575073242, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8853071928024292, + "num_tokens": 315773103.0, + "step": 8136 + }, + { + "epoch": 1.0351100368909807, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.629646301269531, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8918384313583374, + "num_tokens": 315810464.0, + "step": 8137 + }, + { + "epoch": 1.0352372471695712, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.589195251464844, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8878124952316284, + "num_tokens": 315848450.0, + "step": 8138 + }, + { + "epoch": 1.0353644574481617, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.665031433105469, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.878682017326355, + "num_tokens": 315886470.0, + "step": 8139 + }, + { + "epoch": 1.0354916677267523, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.583049774169922, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8844329118728638, + "num_tokens": 315925118.0, + "step": 8140 + }, + { + "epoch": 1.0356188780053428, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.582119941711426, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8985227346420288, + "num_tokens": 315961900.0, + "step": 8141 + }, + { + "epoch": 1.0357460882839333, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.595586776733398, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8894903659820557, + "num_tokens": 315997615.0, + "step": 8142 + }, + { + "epoch": 1.0358732985625239, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.621682167053223, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8860476613044739, + "num_tokens": 316037544.0, + "step": 8143 + }, + { + "epoch": 1.0360005088411144, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.609588623046875, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8845350742340088, + "num_tokens": 316071568.0, + "step": 8144 + }, + { + "epoch": 1.036127719119705, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.636360168457031, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8861420154571533, + "num_tokens": 316113220.0, + "step": 8145 + }, + { + "epoch": 1.0362549293982954, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.53653621673584, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8966686725616455, + "num_tokens": 316153460.0, + "step": 8146 + }, + { + "epoch": 1.036382139676886, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.621051788330078, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8881717324256897, + "num_tokens": 316191685.0, + "step": 8147 + }, + { + "epoch": 1.0365093499554765, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.684915542602539, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8793020248413086, + "num_tokens": 316225375.0, + "step": 8148 + }, + { + "epoch": 1.0366365602340668, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.521011352539062, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.888671875, + "num_tokens": 316263301.0, + "step": 8149 + }, + { + "epoch": 1.0367637705126573, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.677705764770508, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8777924180030823, + "num_tokens": 316298720.0, + "step": 8150 + }, + { + "epoch": 1.0368909807912479, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.641900062561035, + "learning_rate": 1e-06, + "loss": 0.3903, + "mean_token_accuracy": 0.8692575693130493, + "num_tokens": 316339030.0, + "step": 8151 + }, + { + "epoch": 1.0370181910698384, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.616105079650879, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8912591934204102, + "num_tokens": 316379083.0, + "step": 8152 + }, + { + "epoch": 1.037145401348429, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.636699676513672, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8871088027954102, + "num_tokens": 316417411.0, + "step": 8153 + }, + { + "epoch": 1.0372726116270194, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.525437355041504, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.887500524520874, + "num_tokens": 316460872.0, + "step": 8154 + }, + { + "epoch": 1.03739982190561, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.51928997039795, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8930058479309082, + "num_tokens": 316500444.0, + "step": 8155 + }, + { + "epoch": 1.0375270321842005, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.5768404006958, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8822356462478638, + "num_tokens": 316540414.0, + "step": 8156 + }, + { + "epoch": 1.037654242462791, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.52424144744873, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8942462801933289, + "num_tokens": 316579200.0, + "step": 8157 + }, + { + "epoch": 1.0377814527413816, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.534852981567383, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.8717128038406372, + "num_tokens": 316624881.0, + "step": 8158 + }, + { + "epoch": 1.037908663019972, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.523557662963867, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8903542757034302, + "num_tokens": 316669041.0, + "step": 8159 + }, + { + "epoch": 1.0380358732985626, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.54430103302002, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8907917737960815, + "num_tokens": 316708638.0, + "step": 8160 + }, + { + "epoch": 1.0381630835771531, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.655977249145508, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8879120349884033, + "num_tokens": 316742747.0, + "step": 8161 + }, + { + "epoch": 1.0382902938557435, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.571451187133789, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8930031061172485, + "num_tokens": 316781454.0, + "step": 8162 + }, + { + "epoch": 1.038417504134334, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.671992301940918, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8742697834968567, + "num_tokens": 316819194.0, + "step": 8163 + }, + { + "epoch": 1.0385447144129245, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.549120903015137, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8859882950782776, + "num_tokens": 316857901.0, + "step": 8164 + }, + { + "epoch": 1.038671924691515, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.656510353088379, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.883203387260437, + "num_tokens": 316891028.0, + "step": 8165 + }, + { + "epoch": 1.0387991349701056, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.719329833984375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.590611457824707, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.8749402761459351, + "num_tokens": 316932667.0, + "step": 8166 + }, + { + "epoch": 1.038926345248696, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.591946601867676, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8797796964645386, + "num_tokens": 316967713.0, + "step": 8167 + }, + { + "epoch": 1.0390535555272866, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.71595287322998, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8730961084365845, + "num_tokens": 317001889.0, + "step": 8168 + }, + { + "epoch": 1.0391807658058771, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.50489616394043, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9024791717529297, + "num_tokens": 317040158.0, + "step": 8169 + }, + { + "epoch": 1.0393079760844677, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.582435607910156, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8805619478225708, + "num_tokens": 317078391.0, + "step": 8170 + }, + { + "epoch": 1.0394351863630582, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.612929344177246, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8793283700942993, + "num_tokens": 317123181.0, + "step": 8171 + }, + { + "epoch": 1.0395623966416487, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.677586555480957, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.8711415529251099, + "num_tokens": 317155910.0, + "step": 8172 + }, + { + "epoch": 1.0396896069202393, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.653641700744629, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8841232061386108, + "num_tokens": 317192467.0, + "step": 8173 + }, + { + "epoch": 1.0398168171988296, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.671010971069336, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.8684507608413696, + "num_tokens": 317228415.0, + "step": 8174 + }, + { + "epoch": 1.03994402747742, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.552908897399902, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8842437267303467, + "num_tokens": 317269641.0, + "step": 8175 + }, + { + "epoch": 1.0400712377560106, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.594630241394043, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8953714370727539, + "num_tokens": 317306979.0, + "step": 8176 + }, + { + "epoch": 1.0401984480346012, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.70830249786377, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8836134672164917, + "num_tokens": 317340121.0, + "step": 8177 + }, + { + "epoch": 1.0403256583131917, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.621597290039062, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8924112319946289, + "num_tokens": 317375084.0, + "step": 8178 + }, + { + "epoch": 1.0404528685917822, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.67054271697998, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8849800825119019, + "num_tokens": 317405843.0, + "step": 8179 + }, + { + "epoch": 1.0405800788703727, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.60004711151123, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.886011004447937, + "num_tokens": 317443448.0, + "step": 8180 + }, + { + "epoch": 1.0407072891489633, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.683825492858887, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.9007291197776794, + "num_tokens": 317479329.0, + "step": 8181 + }, + { + "epoch": 1.0408344994275538, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.657349586486816, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8858714699745178, + "num_tokens": 317513844.0, + "step": 8182 + }, + { + "epoch": 1.0409617097061443, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.62744426727295, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8867056369781494, + "num_tokens": 317549296.0, + "step": 8183 + }, + { + "epoch": 1.0410889199847349, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.564128875732422, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8804839849472046, + "num_tokens": 317589306.0, + "step": 8184 + }, + { + "epoch": 1.0412161302633254, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.573410987854004, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8803812265396118, + "num_tokens": 317628375.0, + "step": 8185 + }, + { + "epoch": 1.0413433405419157, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.600686073303223, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8831430673599243, + "num_tokens": 317666349.0, + "step": 8186 + }, + { + "epoch": 1.0414705508205062, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.630485534667969, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8803234100341797, + "num_tokens": 317705205.0, + "step": 8187 + }, + { + "epoch": 1.0415977610990967, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.638623237609863, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8971310257911682, + "num_tokens": 317743025.0, + "step": 8188 + }, + { + "epoch": 1.0417249713776873, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.529815673828125, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8787316679954529, + "num_tokens": 317783735.0, + "step": 8189 + }, + { + "epoch": 1.0418521816562778, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.594558715820312, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8688159584999084, + "num_tokens": 317825154.0, + "step": 8190 + }, + { + "epoch": 1.0419793919348683, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.643918991088867, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8830983638763428, + "num_tokens": 317863752.0, + "step": 8191 + }, + { + "epoch": 1.0421066022134589, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.592572212219238, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.887369692325592, + "num_tokens": 317903615.0, + "step": 8192 + }, + { + "epoch": 1.0422338124920494, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.590958595275879, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8836893439292908, + "num_tokens": 317942099.0, + "step": 8193 + }, + { + "epoch": 1.04236102277064, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.710153579711914, + "learning_rate": 1e-06, + "loss": 0.3996, + "mean_token_accuracy": 0.8649250268936157, + "num_tokens": 317983445.0, + "step": 8194 + }, + { + "epoch": 1.0424882330492304, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.584548950195312, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.875923752784729, + "num_tokens": 318025719.0, + "step": 8195 + }, + { + "epoch": 1.042615443327821, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.556685447692871, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8877637386322021, + "num_tokens": 318062618.0, + "step": 8196 + }, + { + "epoch": 1.0427426536064115, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.580489158630371, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8974368572235107, + "num_tokens": 318104890.0, + "step": 8197 + }, + { + "epoch": 1.0428698638850018, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.508286476135254, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8870837688446045, + "num_tokens": 318149796.0, + "step": 8198 + }, + { + "epoch": 1.0429970741635923, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.720916748046875, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.8755987882614136, + "num_tokens": 318187165.0, + "step": 8199 + }, + { + "epoch": 1.0431242844421829, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.601766586303711, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8886591196060181, + "num_tokens": 318222713.0, + "step": 8200 + }, + { + "epoch": 1.0432514947207734, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.633903503417969, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8830040693283081, + "num_tokens": 318258693.0, + "step": 8201 + }, + { + "epoch": 1.043378704999364, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.706775665283203, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.8696959018707275, + "num_tokens": 318295210.0, + "step": 8202 + }, + { + "epoch": 1.0435059152779544, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.57519245147705, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.9003634452819824, + "num_tokens": 318327212.0, + "step": 8203 + }, + { + "epoch": 1.043633125556545, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.546749114990234, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8894140124320984, + "num_tokens": 318375459.0, + "step": 8204 + }, + { + "epoch": 1.0437603358351355, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.57663631439209, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8925672769546509, + "num_tokens": 318412687.0, + "step": 8205 + }, + { + "epoch": 1.043887546113726, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.571915626525879, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8858538269996643, + "num_tokens": 318453518.0, + "step": 8206 + }, + { + "epoch": 1.0440147563923166, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.717601776123047, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8788790702819824, + "num_tokens": 318488951.0, + "step": 8207 + }, + { + "epoch": 1.044141966670907, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.604222297668457, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8876832723617554, + "num_tokens": 318525365.0, + "step": 8208 + }, + { + "epoch": 1.0442691769494976, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.64186954498291, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8801458477973938, + "num_tokens": 318567888.0, + "step": 8209 + }, + { + "epoch": 1.0443963872280881, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.555091857910156, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.896994411945343, + "num_tokens": 318612914.0, + "step": 8210 + }, + { + "epoch": 1.0445235975066784, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.535367965698242, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8823488354682922, + "num_tokens": 318657511.0, + "step": 8211 + }, + { + "epoch": 1.044650807785269, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.784674644470215, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8742111921310425, + "num_tokens": 318693659.0, + "step": 8212 + }, + { + "epoch": 1.0447780180638595, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.533727645874023, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.89778071641922, + "num_tokens": 318735200.0, + "step": 8213 + }, + { + "epoch": 1.04490522834245, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.581589698791504, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8807483911514282, + "num_tokens": 318776248.0, + "step": 8214 + }, + { + "epoch": 1.0450324386210406, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.57351016998291, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8846215009689331, + "num_tokens": 318826119.0, + "step": 8215 + }, + { + "epoch": 1.045159648899631, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.618934631347656, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.886292576789856, + "num_tokens": 318865250.0, + "step": 8216 + }, + { + "epoch": 1.0452868591782216, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.69400691986084, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8873912692070007, + "num_tokens": 318899960.0, + "step": 8217 + }, + { + "epoch": 1.0454140694568121, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.646366119384766, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8943259716033936, + "num_tokens": 318935234.0, + "step": 8218 + }, + { + "epoch": 1.0455412797354027, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.613995552062988, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8796364068984985, + "num_tokens": 318976518.0, + "step": 8219 + }, + { + "epoch": 1.0456684900139932, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.748068809509277, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.873559296131134, + "num_tokens": 319011996.0, + "step": 8220 + }, + { + "epoch": 1.0457957002925837, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.701154708862305, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8810230493545532, + "num_tokens": 319046659.0, + "step": 8221 + }, + { + "epoch": 1.0459229105711743, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.743171691894531e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.654867172241211, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8840542435646057, + "num_tokens": 319084985.0, + "step": 8222 + }, + { + "epoch": 1.0460501208497646, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.597535133361816, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8782498240470886, + "num_tokens": 319124386.0, + "step": 8223 + }, + { + "epoch": 1.046177331128355, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.61975383758545, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8806366920471191, + "num_tokens": 319163151.0, + "step": 8224 + }, + { + "epoch": 1.0463045414069456, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.572433471679688, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8923583030700684, + "num_tokens": 319202334.0, + "step": 8225 + }, + { + "epoch": 1.0464317516855361, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.613449096679688, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8973323702812195, + "num_tokens": 319242990.0, + "step": 8226 + }, + { + "epoch": 1.0465589619641267, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.695518493652344, + "learning_rate": 1e-06, + "loss": 0.3864, + "mean_token_accuracy": 0.8716562986373901, + "num_tokens": 319282011.0, + "step": 8227 + }, + { + "epoch": 1.0466861722427172, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.616369247436523, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8899684548377991, + "num_tokens": 319326014.0, + "step": 8228 + }, + { + "epoch": 1.0468133825213077, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.606720924377441, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.8760781288146973, + "num_tokens": 319367540.0, + "step": 8229 + }, + { + "epoch": 1.0469405927998983, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.716527938842773, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9040648937225342, + "num_tokens": 319399111.0, + "step": 8230 + }, + { + "epoch": 1.0470678030784888, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.69538688659668, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8921113014221191, + "num_tokens": 319432484.0, + "step": 8231 + }, + { + "epoch": 1.0471950133570793, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.625944137573242, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8912938833236694, + "num_tokens": 319467851.0, + "step": 8232 + }, + { + "epoch": 1.0473222236356698, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.790873527526855, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8769457340240479, + "num_tokens": 319500432.0, + "step": 8233 + }, + { + "epoch": 1.0474494339142604, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.585602760314941, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8849156498908997, + "num_tokens": 319546464.0, + "step": 8234 + }, + { + "epoch": 1.0475766441928507, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.55171012878418, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8812267780303955, + "num_tokens": 319586819.0, + "step": 8235 + }, + { + "epoch": 1.0477038544714412, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.642434120178223, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8851816654205322, + "num_tokens": 319623922.0, + "step": 8236 + }, + { + "epoch": 1.0478310647500317, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.777251243591309, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8812963366508484, + "num_tokens": 319655206.0, + "step": 8237 + }, + { + "epoch": 1.0479582750286223, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.655840873718262, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8746169805526733, + "num_tokens": 319689127.0, + "step": 8238 + }, + { + "epoch": 1.0480854853072128, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.592108726501465, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.880290687084198, + "num_tokens": 319729478.0, + "step": 8239 + }, + { + "epoch": 1.0482126955858033, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.65268325805664, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8817270994186401, + "num_tokens": 319763045.0, + "step": 8240 + }, + { + "epoch": 1.0483399058643939, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.684483528137207, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8854901194572449, + "num_tokens": 319796629.0, + "step": 8241 + }, + { + "epoch": 1.0484671161429844, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.523231506347656, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8807339072227478, + "num_tokens": 319841730.0, + "step": 8242 + }, + { + "epoch": 1.048594326421575, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.655028343200684, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8795638084411621, + "num_tokens": 319878171.0, + "step": 8243 + }, + { + "epoch": 1.0487215367001654, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.531014442443848, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8967668414115906, + "num_tokens": 319915711.0, + "step": 8244 + }, + { + "epoch": 1.048848746978756, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.605454444885254, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8835670351982117, + "num_tokens": 319957208.0, + "step": 8245 + }, + { + "epoch": 1.0489759572573465, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.651216506958008, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8852478265762329, + "num_tokens": 319992431.0, + "step": 8246 + }, + { + "epoch": 1.0491031675359368, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.590315818786621, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.9002174735069275, + "num_tokens": 320028838.0, + "step": 8247 + }, + { + "epoch": 1.0492303778145273, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.628856658935547, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8838773965835571, + "num_tokens": 320070161.0, + "step": 8248 + }, + { + "epoch": 1.0493575880931179, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.675883293151855, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.892130970954895, + "num_tokens": 320104176.0, + "step": 8249 + }, + { + "epoch": 1.0494847983717084, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.568239212036133, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.886063814163208, + "num_tokens": 320144828.0, + "step": 8250 + }, + { + "epoch": 1.049612008650299, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.601092338562012, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8854662179946899, + "num_tokens": 320187808.0, + "step": 8251 + }, + { + "epoch": 1.0497392189288894, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.605998039245605, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8844042420387268, + "num_tokens": 320232807.0, + "step": 8252 + }, + { + "epoch": 1.04986642920748, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.603312492370605, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8866279721260071, + "num_tokens": 320273092.0, + "step": 8253 + }, + { + "epoch": 1.0499936394860705, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.636820793151855, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8917621374130249, + "num_tokens": 320309194.0, + "step": 8254 + }, + { + "epoch": 1.050120849764661, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.67686653137207, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8831728100776672, + "num_tokens": 320343603.0, + "step": 8255 + }, + { + "epoch": 1.0502480600432516, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.655701637268066, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8908741474151611, + "num_tokens": 320382245.0, + "step": 8256 + }, + { + "epoch": 1.050375270321842, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.638113021850586, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8750449419021606, + "num_tokens": 320420907.0, + "step": 8257 + }, + { + "epoch": 1.0505024806004326, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.693365097045898, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8801397085189819, + "num_tokens": 320457960.0, + "step": 8258 + }, + { + "epoch": 1.0506296908790231, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.540916442871094, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8892006874084473, + "num_tokens": 320499093.0, + "step": 8259 + }, + { + "epoch": 1.0507569011576134, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.661233901977539, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8768700361251831, + "num_tokens": 320539699.0, + "step": 8260 + }, + { + "epoch": 1.050884111436204, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.6475248336792, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8829655647277832, + "num_tokens": 320577562.0, + "step": 8261 + }, + { + "epoch": 1.0510113217147945, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.573711395263672, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8829086422920227, + "num_tokens": 320617535.0, + "step": 8262 + }, + { + "epoch": 1.051138531993385, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.641806602478027, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8728179931640625, + "num_tokens": 320656282.0, + "step": 8263 + }, + { + "epoch": 1.0512657422719756, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.645313262939453, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8945022821426392, + "num_tokens": 320690934.0, + "step": 8264 + }, + { + "epoch": 1.051392952550566, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.512292861938477, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8897689580917358, + "num_tokens": 320733779.0, + "step": 8265 + }, + { + "epoch": 1.0515201628291566, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.665287017822266, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.872268795967102, + "num_tokens": 320773648.0, + "step": 8266 + }, + { + "epoch": 1.0516473731077471, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.624463081359863, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8788180947303772, + "num_tokens": 320812587.0, + "step": 8267 + }, + { + "epoch": 1.0517745833863377, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.58671760559082, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9045214056968689, + "num_tokens": 320849454.0, + "step": 8268 + }, + { + "epoch": 1.0519017936649282, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.731524467468262, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8787603378295898, + "num_tokens": 320887365.0, + "step": 8269 + }, + { + "epoch": 1.0520290039435187, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.54072380065918, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9010968208312988, + "num_tokens": 320924170.0, + "step": 8270 + }, + { + "epoch": 1.0521562142221093, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.564785957336426, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8941996693611145, + "num_tokens": 320964016.0, + "step": 8271 + }, + { + "epoch": 1.0522834245006996, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.694500923156738, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8953129053115845, + "num_tokens": 320995504.0, + "step": 8272 + }, + { + "epoch": 1.05241063477929, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.730500221252441, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8897218704223633, + "num_tokens": 321029409.0, + "step": 8273 + }, + { + "epoch": 1.0525378450578806, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.562949180603027, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8921890258789062, + "num_tokens": 321068945.0, + "step": 8274 + }, + { + "epoch": 1.0526650553364711, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.580406188964844, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8995760083198547, + "num_tokens": 321107093.0, + "step": 8275 + }, + { + "epoch": 1.0527922656150617, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.56922721862793, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8852554559707642, + "num_tokens": 321145045.0, + "step": 8276 + }, + { + "epoch": 1.0529194758936522, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.786112785339355, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8762516975402832, + "num_tokens": 321174615.0, + "step": 8277 + }, + { + "epoch": 1.0530466861722427, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.530431747436523, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8937045335769653, + "num_tokens": 321217484.0, + "step": 8278 + }, + { + "epoch": 1.0531738964508333, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.611470222473145, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8886411786079407, + "num_tokens": 321253690.0, + "step": 8279 + }, + { + "epoch": 1.0533011067294238, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.679760932922363, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8934957981109619, + "num_tokens": 321295724.0, + "step": 8280 + }, + { + "epoch": 1.0534283170080143, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.626350402832031, + "learning_rate": 1e-06, + "loss": 0.3911, + "mean_token_accuracy": 0.8651243448257446, + "num_tokens": 321339525.0, + "step": 8281 + }, + { + "epoch": 1.0535555272866048, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.714949607849121, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8808935880661011, + "num_tokens": 321376329.0, + "step": 8282 + }, + { + "epoch": 1.0536827375651954, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.631436347961426, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8963388800621033, + "num_tokens": 321411421.0, + "step": 8283 + }, + { + "epoch": 1.0538099478437857, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.576212882995605, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8860752582550049, + "num_tokens": 321449070.0, + "step": 8284 + }, + { + "epoch": 1.0539371581223762, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.640215873718262, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8889051675796509, + "num_tokens": 321489385.0, + "step": 8285 + }, + { + "epoch": 1.0540643684009667, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.562820434570312, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8826265335083008, + "num_tokens": 321531172.0, + "step": 8286 + }, + { + "epoch": 1.0541915786795573, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.604138374328613, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8980710506439209, + "num_tokens": 321576799.0, + "step": 8287 + }, + { + "epoch": 1.0543187889581478, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.646772384643555, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8867066502571106, + "num_tokens": 321620618.0, + "step": 8288 + }, + { + "epoch": 1.0544459992367383, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.577301979064941, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8906440734863281, + "num_tokens": 321663774.0, + "step": 8289 + }, + { + "epoch": 1.0545732095153288, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.674036979675293, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8760933876037598, + "num_tokens": 321709615.0, + "step": 8290 + }, + { + "epoch": 1.0547004197939194, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.587322235107422, + "learning_rate": 1e-06, + "loss": 0.3631, + "mean_token_accuracy": 0.8774173259735107, + "num_tokens": 321753037.0, + "step": 8291 + }, + { + "epoch": 1.05482763007251, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.603621482849121, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8905457854270935, + "num_tokens": 321795226.0, + "step": 8292 + }, + { + "epoch": 1.0549548403511004, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.637533187866211, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8918763995170593, + "num_tokens": 321830739.0, + "step": 8293 + }, + { + "epoch": 1.055082050629691, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.646687507629395, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8871155977249146, + "num_tokens": 321870275.0, + "step": 8294 + }, + { + "epoch": 1.0552092609082815, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.601210594177246, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8896955847740173, + "num_tokens": 321908999.0, + "step": 8295 + }, + { + "epoch": 1.0553364711868718, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.609546661376953, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8876916766166687, + "num_tokens": 321945818.0, + "step": 8296 + }, + { + "epoch": 1.0554636814654623, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.618253707885742, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8887559175491333, + "num_tokens": 321987024.0, + "step": 8297 + }, + { + "epoch": 1.0555908917440529, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.521472930908203, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8961304426193237, + "num_tokens": 322032813.0, + "step": 8298 + }, + { + "epoch": 1.0557181020226434, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.619218826293945, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.9007166624069214, + "num_tokens": 322069165.0, + "step": 8299 + }, + { + "epoch": 1.055845312301234, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.672307968139648, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8707108497619629, + "num_tokens": 322111497.0, + "step": 8300 + }, + { + "epoch": 1.0559725225798244, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.494678497314453, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9008891582489014, + "num_tokens": 322156491.0, + "step": 8301 + }, + { + "epoch": 1.056099732858415, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.70088005065918, + "learning_rate": 1e-06, + "loss": 0.3916, + "mean_token_accuracy": 0.8678845167160034, + "num_tokens": 322199488.0, + "step": 8302 + }, + { + "epoch": 1.0562269431370055, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.56742000579834, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8990412950515747, + "num_tokens": 322240212.0, + "step": 8303 + }, + { + "epoch": 1.056354153415596, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.703067779541016, + "learning_rate": 1e-06, + "loss": 0.3974, + "mean_token_accuracy": 0.8667236566543579, + "num_tokens": 322276463.0, + "step": 8304 + }, + { + "epoch": 1.0564813636941865, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.574807167053223, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8836722373962402, + "num_tokens": 322322146.0, + "step": 8305 + }, + { + "epoch": 1.056608573972777, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.521953582763672, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8953018188476562, + "num_tokens": 322368534.0, + "step": 8306 + }, + { + "epoch": 1.0567357842513676, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.63199520111084, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.894763708114624, + "num_tokens": 322408572.0, + "step": 8307 + }, + { + "epoch": 1.0568629945299581, + "ewc_loss": 0.01318359375, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.555789947509766, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8824945092201233, + "num_tokens": 322448626.0, + "step": 8308 + }, + { + "epoch": 1.0569902048085484, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.594613075256348, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8969680070877075, + "num_tokens": 322488973.0, + "step": 8309 + }, + { + "epoch": 1.057117415087139, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.790855407714844e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.688231468200684, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8802106380462646, + "num_tokens": 322529105.0, + "step": 8310 + }, + { + "epoch": 1.0572446253657295, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.681000709533691, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8845243453979492, + "num_tokens": 322566130.0, + "step": 8311 + }, + { + "epoch": 1.05737183564432, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.545146942138672, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8894994258880615, + "num_tokens": 322607462.0, + "step": 8312 + }, + { + "epoch": 1.0574990459229106, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.653552055358887, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.881829559803009, + "num_tokens": 322648149.0, + "step": 8313 + }, + { + "epoch": 1.057626256201501, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.597219467163086, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9051058292388916, + "num_tokens": 322680104.0, + "step": 8314 + }, + { + "epoch": 1.0577534664800916, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.724164962768555, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8812502026557922, + "num_tokens": 322719497.0, + "step": 8315 + }, + { + "epoch": 1.0578806767586821, + "ewc_loss": 0.0133056640625, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.47713851928711e-06, + "grad_norm": 8.701905250549316, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.886486291885376, + "num_tokens": 322749346.0, + "step": 8316 + }, + { + "epoch": 1.0580078870372727, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.59516429901123, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8915965557098389, + "num_tokens": 322790588.0, + "step": 8317 + }, + { + "epoch": 1.0581350973158632, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.61345100402832, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8802111148834229, + "num_tokens": 322834767.0, + "step": 8318 + }, + { + "epoch": 1.0582623075944537, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.581523895263672, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8847737312316895, + "num_tokens": 322877048.0, + "step": 8319 + }, + { + "epoch": 1.058389517873044, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.603647232055664, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8956717252731323, + "num_tokens": 322914359.0, + "step": 8320 + }, + { + "epoch": 1.0585167281516346, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.627357482910156, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8746787905693054, + "num_tokens": 322954808.0, + "step": 8321 + }, + { + "epoch": 1.058643938430225, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.614550590515137, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8829914927482605, + "num_tokens": 322994842.0, + "step": 8322 + }, + { + "epoch": 1.0587711487088156, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.597803115844727, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8933264017105103, + "num_tokens": 323035883.0, + "step": 8323 + }, + { + "epoch": 1.0588983589874061, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.560508728027344, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8929524421691895, + "num_tokens": 323076695.0, + "step": 8324 + }, + { + "epoch": 1.0590255692659967, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.677770614624023, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8843569755554199, + "num_tokens": 323117705.0, + "step": 8325 + }, + { + "epoch": 1.0591527795445872, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.669106483459473, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8836525082588196, + "num_tokens": 323153744.0, + "step": 8326 + }, + { + "epoch": 1.0592799898231777, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.58137035369873, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9022074937820435, + "num_tokens": 323192364.0, + "step": 8327 + }, + { + "epoch": 1.0594072001017683, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.678608894348145, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8736301064491272, + "num_tokens": 323233419.0, + "step": 8328 + }, + { + "epoch": 1.0595344103803588, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.613007545471191, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8845793604850769, + "num_tokens": 323272289.0, + "step": 8329 + }, + { + "epoch": 1.0596616206589493, + "ewc_loss": 0.01324462890625, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.417533874511719e-06, + "grad_norm": 8.548686027526855, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8984484672546387, + "num_tokens": 323312892.0, + "step": 8330 + }, + { + "epoch": 1.0597888309375398, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.604869842529297, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.893722653388977, + "num_tokens": 323350993.0, + "step": 8331 + }, + { + "epoch": 1.0599160412161304, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.61188793182373, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8924148082733154, + "num_tokens": 323390431.0, + "step": 8332 + }, + { + "epoch": 1.0600432514947207, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.587164878845215, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8901082277297974, + "num_tokens": 323426163.0, + "step": 8333 + }, + { + "epoch": 1.0601704617733112, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.640860557556152, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8860421180725098, + "num_tokens": 323461947.0, + "step": 8334 + }, + { + "epoch": 1.0602976720519017, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.615167617797852, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8874655961990356, + "num_tokens": 323505093.0, + "step": 8335 + }, + { + "epoch": 1.0604248823304923, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.687983512878418, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8816193342208862, + "num_tokens": 323539615.0, + "step": 8336 + }, + { + "epoch": 1.0605520926090828, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.658478736877441, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.896968424320221, + "num_tokens": 323576615.0, + "step": 8337 + }, + { + "epoch": 1.0606793028876733, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.53181266784668, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9019999504089355, + "num_tokens": 323616767.0, + "step": 8338 + }, + { + "epoch": 1.0608065131662638, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.639342308044434, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8899134397506714, + "num_tokens": 323659418.0, + "step": 8339 + }, + { + "epoch": 1.0609337234448544, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.627076148986816, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.9019619226455688, + "num_tokens": 323694157.0, + "step": 8340 + }, + { + "epoch": 1.061060933723445, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.672107696533203, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8773138523101807, + "num_tokens": 323729120.0, + "step": 8341 + }, + { + "epoch": 1.0611881440020354, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.53901481628418, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8891518712043762, + "num_tokens": 323772489.0, + "step": 8342 + }, + { + "epoch": 1.061315354280626, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.642985343933105, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.8757180571556091, + "num_tokens": 323813393.0, + "step": 8343 + }, + { + "epoch": 1.0614425645592165, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.677616119384766, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8888705372810364, + "num_tokens": 323850409.0, + "step": 8344 + }, + { + "epoch": 1.0615697748378068, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.628562927246094, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.892940878868103, + "num_tokens": 323891367.0, + "step": 8345 + }, + { + "epoch": 1.0616969851163973, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.696305274963379, + "learning_rate": 1e-06, + "loss": 0.3671, + "mean_token_accuracy": 0.8722018599510193, + "num_tokens": 323934497.0, + "step": 8346 + }, + { + "epoch": 1.0618241953949878, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.702234268188477, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8870483636856079, + "num_tokens": 323969235.0, + "step": 8347 + }, + { + "epoch": 1.0619514056735784, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.563908576965332, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8988335132598877, + "num_tokens": 324005494.0, + "step": 8348 + }, + { + "epoch": 1.062078615952169, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.588409423828125, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.89508056640625, + "num_tokens": 324046998.0, + "step": 8349 + }, + { + "epoch": 1.0622058262307594, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.69633960723877, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8759207129478455, + "num_tokens": 324085988.0, + "step": 8350 + }, + { + "epoch": 1.06233303650935, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.683505058288574, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8886679410934448, + "num_tokens": 324120122.0, + "step": 8351 + }, + { + "epoch": 1.0624602467879405, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.652153968811035, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8787699937820435, + "num_tokens": 324160199.0, + "step": 8352 + }, + { + "epoch": 1.062587457066531, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.61136245727539, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8932881951332092, + "num_tokens": 324198745.0, + "step": 8353 + }, + { + "epoch": 1.0627146673451215, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.670413970947266, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.8990845680236816, + "num_tokens": 324233967.0, + "step": 8354 + }, + { + "epoch": 1.062841877623712, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.56180191040039, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8911405801773071, + "num_tokens": 324276317.0, + "step": 8355 + }, + { + "epoch": 1.0629690879023026, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.732115745544434, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8881663084030151, + "num_tokens": 324309753.0, + "step": 8356 + }, + { + "epoch": 1.0630962981808931, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.622267723083496, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8959107995033264, + "num_tokens": 324351735.0, + "step": 8357 + }, + { + "epoch": 1.0632235084594834, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.74885082244873, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8823643922805786, + "num_tokens": 324388768.0, + "step": 8358 + }, + { + "epoch": 1.063350718738074, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.629861831665039, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8788890838623047, + "num_tokens": 324429606.0, + "step": 8359 + }, + { + "epoch": 1.0634779290166645, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.68437385559082, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.888178825378418, + "num_tokens": 324467233.0, + "step": 8360 + }, + { + "epoch": 1.063605139295255, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.643752098083496, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8737562894821167, + "num_tokens": 324506880.0, + "step": 8361 + }, + { + "epoch": 1.0637323495738455, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.7144136428833, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8911378383636475, + "num_tokens": 324541659.0, + "step": 8362 + }, + { + "epoch": 1.063859559852436, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.67551040649414, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.896866500377655, + "num_tokens": 324572995.0, + "step": 8363 + }, + { + "epoch": 1.0639867701310266, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.756728172302246, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8864516019821167, + "num_tokens": 324604514.0, + "step": 8364 + }, + { + "epoch": 1.0641139804096171, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.624056816101074, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8768298625946045, + "num_tokens": 324642212.0, + "step": 8365 + }, + { + "epoch": 1.0642411906882077, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.612398147583008, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.9048644304275513, + "num_tokens": 324681043.0, + "step": 8366 + }, + { + "epoch": 1.0643684009667982, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.541814804077148, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8985327482223511, + "num_tokens": 324720730.0, + "step": 8367 + }, + { + "epoch": 1.0644956112453887, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.675999641418457, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8757163286209106, + "num_tokens": 324757361.0, + "step": 8368 + }, + { + "epoch": 1.064622821523979, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.725215911865234, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8979754447937012, + "num_tokens": 324789139.0, + "step": 8369 + }, + { + "epoch": 1.0647500318025696, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.838539123535156e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.569439888000488, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8873863220214844, + "num_tokens": 324826922.0, + "step": 8370 + }, + { + "epoch": 1.06487724208116, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.617081642150879, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8936062455177307, + "num_tokens": 324867418.0, + "step": 8371 + }, + { + "epoch": 1.0650044523597506, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.580788612365723, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8958800435066223, + "num_tokens": 324912450.0, + "step": 8372 + }, + { + "epoch": 1.0651316626383411, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.559646606445312, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.899057149887085, + "num_tokens": 324950430.0, + "step": 8373 + }, + { + "epoch": 1.0652588729169317, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.6222505569458, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8929668664932251, + "num_tokens": 324994100.0, + "step": 8374 + }, + { + "epoch": 1.0653860831955222, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.547390937805176, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8932428359985352, + "num_tokens": 325036952.0, + "step": 8375 + }, + { + "epoch": 1.0655132934741127, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.658158302307129, + "learning_rate": 1e-06, + "loss": 0.4012, + "mean_token_accuracy": 0.8632469177246094, + "num_tokens": 325082107.0, + "step": 8376 + }, + { + "epoch": 1.0656405037527032, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.585205078125, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8952658772468567, + "num_tokens": 325123770.0, + "step": 8377 + }, + { + "epoch": 1.0657677140312938, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.688828468322754, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8835790157318115, + "num_tokens": 325160249.0, + "step": 8378 + }, + { + "epoch": 1.0658949243098843, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.619762420654297, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8816600441932678, + "num_tokens": 325202582.0, + "step": 8379 + }, + { + "epoch": 1.0660221345884748, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.633788108825684, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8720625638961792, + "num_tokens": 325244904.0, + "step": 8380 + }, + { + "epoch": 1.0661493448670654, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.648966789245605, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8898595571517944, + "num_tokens": 325280829.0, + "step": 8381 + }, + { + "epoch": 1.0662765551456557, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.578839302062988, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8926215171813965, + "num_tokens": 325322583.0, + "step": 8382 + }, + { + "epoch": 1.0664037654242462, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.696340560913086, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8842094540596008, + "num_tokens": 325358692.0, + "step": 8383 + }, + { + "epoch": 1.0665309757028367, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.580827713012695, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8857711553573608, + "num_tokens": 325405805.0, + "step": 8384 + }, + { + "epoch": 1.0666581859814273, + "ewc_loss": 0.01336669921875, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.680639266967773, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.880544900894165, + "num_tokens": 325440694.0, + "step": 8385 + }, + { + "epoch": 1.0667853962600178, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.647233963012695, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8871644735336304, + "num_tokens": 325481452.0, + "step": 8386 + }, + { + "epoch": 1.0669126065386083, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.682840347290039, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.874527096748352, + "num_tokens": 325520845.0, + "step": 8387 + }, + { + "epoch": 1.0670398168171988, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.728084564208984, + "learning_rate": 1e-06, + "loss": 0.3717, + "mean_token_accuracy": 0.8728892803192139, + "num_tokens": 325561648.0, + "step": 8388 + }, + { + "epoch": 1.0671670270957894, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.593913078308105, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8848557472229004, + "num_tokens": 325605087.0, + "step": 8389 + }, + { + "epoch": 1.06729423737438, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.708649635314941, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8806837201118469, + "num_tokens": 325641537.0, + "step": 8390 + }, + { + "epoch": 1.0674214476529704, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.551124572753906, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8889560699462891, + "num_tokens": 325687593.0, + "step": 8391 + }, + { + "epoch": 1.067548657931561, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.6449556350708, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8805776834487915, + "num_tokens": 325730202.0, + "step": 8392 + }, + { + "epoch": 1.0676758682101515, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.8623809814453125e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.654419898986816, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8993161916732788, + "num_tokens": 325767336.0, + "step": 8393 + }, + { + "epoch": 1.0678030784887418, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.699016571044922, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8722412586212158, + "num_tokens": 325803779.0, + "step": 8394 + }, + { + "epoch": 1.0679302887673323, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.736268997192383, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8772842884063721, + "num_tokens": 325842638.0, + "step": 8395 + }, + { + "epoch": 1.0680574990459228, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.613694190979004, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8785502910614014, + "num_tokens": 325882716.0, + "step": 8396 + }, + { + "epoch": 1.0681847093245134, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.73046875, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8775234818458557, + "num_tokens": 325917974.0, + "step": 8397 + }, + { + "epoch": 1.068311919603104, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.642330169677734, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8859484791755676, + "num_tokens": 325951367.0, + "step": 8398 + }, + { + "epoch": 1.0684391298816944, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.665348052978516, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.887313723564148, + "num_tokens": 325989131.0, + "step": 8399 + }, + { + "epoch": 1.068566340160285, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.695298194885254, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8857071995735168, + "num_tokens": 326025486.0, + "step": 8400 + }, + { + "epoch": 1.0686935504388755, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.716633796691895, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8875349760055542, + "num_tokens": 326060964.0, + "step": 8401 + }, + { + "epoch": 1.068820760717466, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.604108810424805, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8979290723800659, + "num_tokens": 326099410.0, + "step": 8402 + }, + { + "epoch": 1.0689479709960565, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.669575691223145, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.8679441809654236, + "num_tokens": 326139858.0, + "step": 8403 + }, + { + "epoch": 1.069075181274647, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.6123685836792, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9025497436523438, + "num_tokens": 326176082.0, + "step": 8404 + }, + { + "epoch": 1.0692023915532376, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.679056167602539, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8879482746124268, + "num_tokens": 326217401.0, + "step": 8405 + }, + { + "epoch": 1.0693296018318281, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.751365661621094, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8878965377807617, + "num_tokens": 326248399.0, + "step": 8406 + }, + { + "epoch": 1.0694568121104184, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.542613983154297, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8978688716888428, + "num_tokens": 326292567.0, + "step": 8407 + }, + { + "epoch": 1.069584022389009, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.673733711242676, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8922325372695923, + "num_tokens": 326325333.0, + "step": 8408 + }, + { + "epoch": 1.0697112326675995, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.758173942565918, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8952888250350952, + "num_tokens": 326358840.0, + "step": 8409 + }, + { + "epoch": 1.06983844294619, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.643808364868164, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.878795862197876, + "num_tokens": 326401275.0, + "step": 8410 + }, + { + "epoch": 1.0699656532247805, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.682297706604004, + "learning_rate": 1e-06, + "loss": 0.4028, + "mean_token_accuracy": 0.8662451505661011, + "num_tokens": 326439109.0, + "step": 8411 + }, + { + "epoch": 1.070092863503371, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.567180633544922, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8901838660240173, + "num_tokens": 326478906.0, + "step": 8412 + }, + { + "epoch": 1.0702200737819616, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.568763732910156, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8849166631698608, + "num_tokens": 326522613.0, + "step": 8413 + }, + { + "epoch": 1.0703472840605521, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.587759017944336, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8940888047218323, + "num_tokens": 326559845.0, + "step": 8414 + }, + { + "epoch": 1.0704744943391427, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.694607734680176, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.894545316696167, + "num_tokens": 326595293.0, + "step": 8415 + }, + { + "epoch": 1.0706017046177332, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.577476501464844, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8833588361740112, + "num_tokens": 326639004.0, + "step": 8416 + }, + { + "epoch": 1.0707289148963237, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.671948432922363, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8839566707611084, + "num_tokens": 326683592.0, + "step": 8417 + }, + { + "epoch": 1.070856125174914, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.618618965148926, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8910163640975952, + "num_tokens": 326721595.0, + "step": 8418 + }, + { + "epoch": 1.0709833354535045, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.575881958007812, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8929961919784546, + "num_tokens": 326762583.0, + "step": 8419 + }, + { + "epoch": 1.071110545732095, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.652932167053223, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8782212734222412, + "num_tokens": 326801779.0, + "step": 8420 + }, + { + "epoch": 1.0712377560106856, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.65833568572998, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.884408712387085, + "num_tokens": 326842326.0, + "step": 8421 + }, + { + "epoch": 1.0713649662892761, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.705416679382324, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.880370020866394, + "num_tokens": 326881796.0, + "step": 8422 + }, + { + "epoch": 1.0714921765678667, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.62129020690918, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8845747709274292, + "num_tokens": 326921426.0, + "step": 8423 + }, + { + "epoch": 1.0716193868464572, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.700175285339355, + "learning_rate": 1e-06, + "loss": 0.3643, + "mean_token_accuracy": 0.8721493482589722, + "num_tokens": 326957001.0, + "step": 8424 + }, + { + "epoch": 1.0717465971250477, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.737346649169922, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8821873068809509, + "num_tokens": 326988956.0, + "step": 8425 + }, + { + "epoch": 1.0718738074036382, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.672699928283691, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8911058902740479, + "num_tokens": 327022473.0, + "step": 8426 + }, + { + "epoch": 1.0720010176822288, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.554359436035156, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.8996185064315796, + "num_tokens": 327062303.0, + "step": 8427 + }, + { + "epoch": 1.0721282279608193, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.740440368652344, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8847391605377197, + "num_tokens": 327093320.0, + "step": 8428 + }, + { + "epoch": 1.0722554382394098, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.647418022155762, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8923001289367676, + "num_tokens": 327135269.0, + "step": 8429 + }, + { + "epoch": 1.0723826485180004, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.577917098999023, + "learning_rate": 1e-06, + "loss": 0.3475, + "mean_token_accuracy": 0.8822469711303711, + "num_tokens": 327176234.0, + "step": 8430 + }, + { + "epoch": 1.0725098587965907, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.65390682220459, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8867133855819702, + "num_tokens": 327212555.0, + "step": 8431 + }, + { + "epoch": 1.0726370690751812, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.712949752807617, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8810762166976929, + "num_tokens": 327246827.0, + "step": 8432 + }, + { + "epoch": 1.0727642793537717, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.670938491821289, + "learning_rate": 1e-06, + "loss": 0.3854, + "mean_token_accuracy": 0.869882345199585, + "num_tokens": 327282084.0, + "step": 8433 + }, + { + "epoch": 1.0728914896323622, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.586913108825684, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8859614133834839, + "num_tokens": 327321329.0, + "step": 8434 + }, + { + "epoch": 1.0730186999109528, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.615273475646973, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8948378562927246, + "num_tokens": 327359773.0, + "step": 8435 + }, + { + "epoch": 1.0731459101895433, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.669454574584961, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8827308416366577, + "num_tokens": 327398398.0, + "step": 8436 + }, + { + "epoch": 1.0732731204681338, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.602925300598145, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8896153569221497, + "num_tokens": 327437017.0, + "step": 8437 + }, + { + "epoch": 1.0734003307467244, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.611184120178223, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.881821870803833, + "num_tokens": 327474549.0, + "step": 8438 + }, + { + "epoch": 1.073527541025315, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.569416046142578, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8845810890197754, + "num_tokens": 327515115.0, + "step": 8439 + }, + { + "epoch": 1.0736547513039054, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.622167587280273, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.893166184425354, + "num_tokens": 327552069.0, + "step": 8440 + }, + { + "epoch": 1.073781961582496, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.590825080871582, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8861504793167114, + "num_tokens": 327594268.0, + "step": 8441 + }, + { + "epoch": 1.0739091718610865, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.636388778686523, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8888553380966187, + "num_tokens": 327631758.0, + "step": 8442 + }, + { + "epoch": 1.0740363821396768, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.611479759216309, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.88905930519104, + "num_tokens": 327670539.0, + "step": 8443 + }, + { + "epoch": 1.0741635924182673, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.61451244354248, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8823453187942505, + "num_tokens": 327714363.0, + "step": 8444 + }, + { + "epoch": 1.0742908026968578, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.886222839355469e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.684974670410156, + "learning_rate": 1e-06, + "loss": 0.3798, + "mean_token_accuracy": 0.8699796199798584, + "num_tokens": 327752987.0, + "step": 8445 + }, + { + "epoch": 1.0744180129754484, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.806974411010742, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.9037874341011047, + "num_tokens": 327777822.0, + "step": 8446 + }, + { + "epoch": 1.074545223254039, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.657423973083496, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8814529180526733, + "num_tokens": 327812794.0, + "step": 8447 + }, + { + "epoch": 1.0746724335326294, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.649565696716309, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8837265968322754, + "num_tokens": 327851927.0, + "step": 8448 + }, + { + "epoch": 1.07479964381122, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.589624404907227, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8942976593971252, + "num_tokens": 327890259.0, + "step": 8449 + }, + { + "epoch": 1.0749268540898105, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.592405319213867, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8931679129600525, + "num_tokens": 327931035.0, + "step": 8450 + }, + { + "epoch": 1.075054064368401, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.683536529541016, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8889372944831848, + "num_tokens": 327964361.0, + "step": 8451 + }, + { + "epoch": 1.0751812746469915, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.612260818481445, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8886248469352722, + "num_tokens": 328006779.0, + "step": 8452 + }, + { + "epoch": 1.075308484925582, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.570467948913574, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8913071751594543, + "num_tokens": 328046131.0, + "step": 8453 + }, + { + "epoch": 1.0754356952041726, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.656614303588867, + "learning_rate": 1e-06, + "loss": 0.4215, + "mean_token_accuracy": 0.8559139966964722, + "num_tokens": 328091688.0, + "step": 8454 + }, + { + "epoch": 1.0755629054827631, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.54985237121582, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9036164283752441, + "num_tokens": 328132887.0, + "step": 8455 + }, + { + "epoch": 1.0756901157613534, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.550867080688477, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.891315221786499, + "num_tokens": 328179948.0, + "step": 8456 + }, + { + "epoch": 1.075817326039944, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.612801551818848, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8881124258041382, + "num_tokens": 328220760.0, + "step": 8457 + }, + { + "epoch": 1.0759445363185345, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.644771575927734, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8811414241790771, + "num_tokens": 328256975.0, + "step": 8458 + }, + { + "epoch": 1.076071746597125, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.654586791992188, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8719688653945923, + "num_tokens": 328300594.0, + "step": 8459 + }, + { + "epoch": 1.0761989568757155, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.703048706054688, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8843322396278381, + "num_tokens": 328336976.0, + "step": 8460 + }, + { + "epoch": 1.076326167154306, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.624263763427734, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8893554210662842, + "num_tokens": 328373661.0, + "step": 8461 + }, + { + "epoch": 1.0764533774328966, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.621559143066406, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.892125129699707, + "num_tokens": 328412727.0, + "step": 8462 + }, + { + "epoch": 1.0765805877114871, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.743903160095215, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.875946581363678, + "num_tokens": 328447906.0, + "step": 8463 + }, + { + "epoch": 1.0767077979900777, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.697632789611816, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8849890232086182, + "num_tokens": 328485120.0, + "step": 8464 + }, + { + "epoch": 1.0768350082686682, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.695549011230469, + "learning_rate": 1e-06, + "loss": 0.3922, + "mean_token_accuracy": 0.8660253286361694, + "num_tokens": 328523898.0, + "step": 8465 + }, + { + "epoch": 1.0769622185472587, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.712135314941406, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.887810468673706, + "num_tokens": 328556612.0, + "step": 8466 + }, + { + "epoch": 1.077089428825849, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.652902603149414, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8869274258613586, + "num_tokens": 328592104.0, + "step": 8467 + }, + { + "epoch": 1.0772166391044395, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.5776948928833, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8967956304550171, + "num_tokens": 328634136.0, + "step": 8468 + }, + { + "epoch": 1.07734384938303, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.585311889648438, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8846228122711182, + "num_tokens": 328677607.0, + "step": 8469 + }, + { + "epoch": 1.0774710596616206, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.594765663146973, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8969025015830994, + "num_tokens": 328718816.0, + "step": 8470 + }, + { + "epoch": 1.0775982699402111, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.666004180908203, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8803640604019165, + "num_tokens": 328756655.0, + "step": 8471 + }, + { + "epoch": 1.0777254802188017, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.732196807861328, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.9003398418426514, + "num_tokens": 328790943.0, + "step": 8472 + }, + { + "epoch": 1.0778526904973922, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.630681037902832, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8926128149032593, + "num_tokens": 328833583.0, + "step": 8473 + }, + { + "epoch": 1.0779799007759827, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.657988548278809, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8873596787452698, + "num_tokens": 328869500.0, + "step": 8474 + }, + { + "epoch": 1.0781071110545732, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.704075813293457, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8832293748855591, + "num_tokens": 328908418.0, + "step": 8475 + }, + { + "epoch": 1.0782343213331638, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.625763893127441, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8870537281036377, + "num_tokens": 328946068.0, + "step": 8476 + }, + { + "epoch": 1.0783615316117543, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.637791633605957, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.9010776281356812, + "num_tokens": 328983245.0, + "step": 8477 + }, + { + "epoch": 1.0784887418903448, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.681320190429688, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8820169568061829, + "num_tokens": 329019189.0, + "step": 8478 + }, + { + "epoch": 1.0786159521689354, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.604145050048828, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8815222978591919, + "num_tokens": 329060519.0, + "step": 8479 + }, + { + "epoch": 1.0787431624475257, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.631348609924316, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8933409452438354, + "num_tokens": 329096722.0, + "step": 8480 + }, + { + "epoch": 1.0788703727261162, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.649484634399414, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8836753368377686, + "num_tokens": 329134319.0, + "step": 8481 + }, + { + "epoch": 1.0789975830047067, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.56802749633789, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.888683557510376, + "num_tokens": 329175094.0, + "step": 8482 + }, + { + "epoch": 1.0791247932832972, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.600428581237793, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8820090293884277, + "num_tokens": 329218417.0, + "step": 8483 + }, + { + "epoch": 1.0792520035618878, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.646187782287598, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8895976543426514, + "num_tokens": 329256154.0, + "step": 8484 + }, + { + "epoch": 1.0793792138404783, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.569689750671387, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8955170512199402, + "num_tokens": 329294252.0, + "step": 8485 + }, + { + "epoch": 1.0795064241190688, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.58979606628418, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8856507539749146, + "num_tokens": 329334997.0, + "step": 8486 + }, + { + "epoch": 1.0796336343976594, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.719874382019043, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8818714022636414, + "num_tokens": 329375102.0, + "step": 8487 + }, + { + "epoch": 1.0797608446762499, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.541581153869629, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8986490964889526, + "num_tokens": 329417897.0, + "step": 8488 + }, + { + "epoch": 1.0798880549548404, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.59589672088623, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8919788599014282, + "num_tokens": 329459953.0, + "step": 8489 + }, + { + "epoch": 1.080015265233431, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.694913864135742, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8947298526763916, + "num_tokens": 329500150.0, + "step": 8490 + }, + { + "epoch": 1.0801424755120215, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.584000587463379, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8858969211578369, + "num_tokens": 329543036.0, + "step": 8491 + }, + { + "epoch": 1.0802696857906118, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.664610862731934, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.895763099193573, + "num_tokens": 329583593.0, + "step": 8492 + }, + { + "epoch": 1.0803968960692023, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.601036071777344, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8927831649780273, + "num_tokens": 329623492.0, + "step": 8493 + }, + { + "epoch": 1.0805241063477928, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.720184326171875, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8779181241989136, + "num_tokens": 329658897.0, + "step": 8494 + }, + { + "epoch": 1.0806513166263834, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.649136543273926, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8843583464622498, + "num_tokens": 329695075.0, + "step": 8495 + }, + { + "epoch": 1.080778526904974, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.652495384216309, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8987584114074707, + "num_tokens": 329731218.0, + "step": 8496 + }, + { + "epoch": 1.0809057371835644, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.619531631469727, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8749139308929443, + "num_tokens": 329776181.0, + "step": 8497 + }, + { + "epoch": 1.081032947462155, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.610207557678223, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.890734851360321, + "num_tokens": 329813892.0, + "step": 8498 + }, + { + "epoch": 1.0811601577407455, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.690560340881348, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8874190449714661, + "num_tokens": 329845637.0, + "step": 8499 + }, + { + "epoch": 1.081287368019336, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.739896774291992, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.8725389242172241, + "num_tokens": 329880761.0, + "step": 8500 + }, + { + "epoch": 1.0814145782979265, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.682074546813965, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8887574076652527, + "num_tokens": 329919389.0, + "step": 8501 + }, + { + "epoch": 1.081541788576517, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.571341514587402, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8918296098709106, + "num_tokens": 329964131.0, + "step": 8502 + }, + { + "epoch": 1.0816689988551076, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.604491233825684, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8956313133239746, + "num_tokens": 330005227.0, + "step": 8503 + }, + { + "epoch": 1.0817962091336981, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.617578506469727, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8726441860198975, + "num_tokens": 330045885.0, + "step": 8504 + }, + { + "epoch": 1.0819234194122884, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.686934471130371, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.882605254650116, + "num_tokens": 330082908.0, + "step": 8505 + }, + { + "epoch": 1.082050629690879, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.709000587463379, + "learning_rate": 1e-06, + "loss": 0.3897, + "mean_token_accuracy": 0.8645752668380737, + "num_tokens": 330125763.0, + "step": 8506 + }, + { + "epoch": 1.0821778399694695, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.627035140991211, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8967540860176086, + "num_tokens": 330162606.0, + "step": 8507 + }, + { + "epoch": 1.08230505024806, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.639710426330566, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8883559703826904, + "num_tokens": 330199948.0, + "step": 8508 + }, + { + "epoch": 1.0824322605266505, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.731890678405762, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8868972063064575, + "num_tokens": 330234306.0, + "step": 8509 + }, + { + "epoch": 1.082559470805241, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.624613761901855, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8769902586936951, + "num_tokens": 330274236.0, + "step": 8510 + }, + { + "epoch": 1.0826866810838316, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.6181001663208, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8915647268295288, + "num_tokens": 330312960.0, + "step": 8511 + }, + { + "epoch": 1.0828138913624221, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.679654121398926, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8783718943595886, + "num_tokens": 330354483.0, + "step": 8512 + }, + { + "epoch": 1.0829411016410126, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.566473007202148, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8885207772254944, + "num_tokens": 330394571.0, + "step": 8513 + }, + { + "epoch": 1.0830683119196032, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.583316802978516, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8872100114822388, + "num_tokens": 330440872.0, + "step": 8514 + }, + { + "epoch": 1.0831955221981937, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.675311088562012, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8982160687446594, + "num_tokens": 330477956.0, + "step": 8515 + }, + { + "epoch": 1.083322732476784, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.577019691467285, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8775320053100586, + "num_tokens": 330522061.0, + "step": 8516 + }, + { + "epoch": 1.0834499427553745, + "ewc_loss": 0.01348876953125, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.647294998168945, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8866645097732544, + "num_tokens": 330562991.0, + "step": 8517 + }, + { + "epoch": 1.083577153033965, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.685449600219727, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8825408220291138, + "num_tokens": 330600165.0, + "step": 8518 + }, + { + "epoch": 1.0837043633125556, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.704392433166504, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8863058686256409, + "num_tokens": 330637845.0, + "step": 8519 + }, + { + "epoch": 1.0838315735911461, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.70431900024414, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8911256790161133, + "num_tokens": 330669782.0, + "step": 8520 + }, + { + "epoch": 1.0839587838697367, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.62474250793457, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8888756632804871, + "num_tokens": 330707800.0, + "step": 8521 + }, + { + "epoch": 1.0840859941483272, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.63305950164795, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8781548738479614, + "num_tokens": 330750057.0, + "step": 8522 + }, + { + "epoch": 1.0842132044269177, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.665812492370605, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8893697261810303, + "num_tokens": 330791453.0, + "step": 8523 + }, + { + "epoch": 1.0843404147055082, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.714827537536621, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.874361515045166, + "num_tokens": 330828534.0, + "step": 8524 + }, + { + "epoch": 1.0844676249840988, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.65167236328125, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9018230438232422, + "num_tokens": 330865723.0, + "step": 8525 + }, + { + "epoch": 1.0845948352626893, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.4967622756958, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8960115909576416, + "num_tokens": 330910530.0, + "step": 8526 + }, + { + "epoch": 1.0847220455412798, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.711989402770996, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.888594925403595, + "num_tokens": 330943807.0, + "step": 8527 + }, + { + "epoch": 1.0848492558198704, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.655348777770996, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8848464488983154, + "num_tokens": 330985422.0, + "step": 8528 + }, + { + "epoch": 1.0849764660984607, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.697773933410645, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8983734846115112, + "num_tokens": 331015117.0, + "step": 8529 + }, + { + "epoch": 1.0851036763770512, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.644582748413086, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8876063823699951, + "num_tokens": 331050475.0, + "step": 8530 + }, + { + "epoch": 1.0852308866556417, + "ewc_loss": 0.013427734375, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.76318359375, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.8738964796066284, + "num_tokens": 331084622.0, + "step": 8531 + }, + { + "epoch": 1.0853580969342322, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.596330642700195, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.9043771624565125, + "num_tokens": 331124236.0, + "step": 8532 + }, + { + "epoch": 1.0854853072128228, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.598409652709961, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8937914371490479, + "num_tokens": 331163725.0, + "step": 8533 + }, + { + "epoch": 1.0856125174914133, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.910064697265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.784966468811035, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8707993626594543, + "num_tokens": 331199516.0, + "step": 8534 + }, + { + "epoch": 1.0857397277700038, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.602625846862793, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8954693078994751, + "num_tokens": 331238907.0, + "step": 8535 + }, + { + "epoch": 1.0858669380485944, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.700455665588379, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8967193961143494, + "num_tokens": 331273313.0, + "step": 8536 + }, + { + "epoch": 1.0859941483271849, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.638084411621094, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8843412399291992, + "num_tokens": 331313562.0, + "step": 8537 + }, + { + "epoch": 1.0861213586057754, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.80635929107666, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8788758516311646, + "num_tokens": 331345661.0, + "step": 8538 + }, + { + "epoch": 1.086248568884366, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.561893463134766, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8870891332626343, + "num_tokens": 331389509.0, + "step": 8539 + }, + { + "epoch": 1.0863757791629565, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.650412559509277, + "learning_rate": 1e-06, + "loss": 0.3681, + "mean_token_accuracy": 0.8765724897384644, + "num_tokens": 331429041.0, + "step": 8540 + }, + { + "epoch": 1.0865029894415468, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.682185173034668, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8924407362937927, + "num_tokens": 331465031.0, + "step": 8541 + }, + { + "epoch": 1.0866301997201373, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.547586441040039, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8930132985115051, + "num_tokens": 331505544.0, + "step": 8542 + }, + { + "epoch": 1.0867574099987278, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.735251426696777, + "learning_rate": 1e-06, + "loss": 0.398, + "mean_token_accuracy": 0.8648239374160767, + "num_tokens": 331546443.0, + "step": 8543 + }, + { + "epoch": 1.0868846202773184, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.673849105834961, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8936564922332764, + "num_tokens": 331584734.0, + "step": 8544 + }, + { + "epoch": 1.0870118305559089, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.589200973510742, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8855949640274048, + "num_tokens": 331627125.0, + "step": 8545 + }, + { + "epoch": 1.0871390408344994, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.761283874511719, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8798410296440125, + "num_tokens": 331664354.0, + "step": 8546 + }, + { + "epoch": 1.08726625111309, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.618569374084473, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8912691473960876, + "num_tokens": 331702528.0, + "step": 8547 + }, + { + "epoch": 1.0873934613916805, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.766694068908691, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8928311467170715, + "num_tokens": 331738706.0, + "step": 8548 + }, + { + "epoch": 1.087520671670271, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.665947914123535, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8891812562942505, + "num_tokens": 331775727.0, + "step": 8549 + }, + { + "epoch": 1.0876478819488615, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.775922775268555, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8913435935974121, + "num_tokens": 331807459.0, + "step": 8550 + }, + { + "epoch": 1.087775092227452, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.771056175231934, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8811297416687012, + "num_tokens": 331843365.0, + "step": 8551 + }, + { + "epoch": 1.0879023025060426, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.60297679901123, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8878699541091919, + "num_tokens": 331879690.0, + "step": 8552 + }, + { + "epoch": 1.0880295127846331, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.724310874938965, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.882663369178772, + "num_tokens": 331920380.0, + "step": 8553 + }, + { + "epoch": 1.0881567230632234, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.602673530578613, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8886915445327759, + "num_tokens": 331963562.0, + "step": 8554 + }, + { + "epoch": 1.088283933341814, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.577301025390625, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.900348424911499, + "num_tokens": 332005340.0, + "step": 8555 + }, + { + "epoch": 1.0884111436204045, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.747074127197266, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8790477514266968, + "num_tokens": 332043790.0, + "step": 8556 + }, + { + "epoch": 1.088538353898995, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.718031883239746, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8844650983810425, + "num_tokens": 332081958.0, + "step": 8557 + }, + { + "epoch": 1.0886655641775855, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.764168739318848, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8870071172714233, + "num_tokens": 332117859.0, + "step": 8558 + }, + { + "epoch": 1.088792774456176, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.737712860107422, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8826217651367188, + "num_tokens": 332157084.0, + "step": 8559 + }, + { + "epoch": 1.0889199847347666, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.658621788024902, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8888067007064819, + "num_tokens": 332196307.0, + "step": 8560 + }, + { + "epoch": 1.0890471950133571, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.732833862304688, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8853908777236938, + "num_tokens": 332231337.0, + "step": 8561 + }, + { + "epoch": 1.0891744052919476, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.827184677124023, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8958449959754944, + "num_tokens": 332263337.0, + "step": 8562 + }, + { + "epoch": 1.0893016155705382, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.59998607635498, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8925718665122986, + "num_tokens": 332301517.0, + "step": 8563 + }, + { + "epoch": 1.0894288258491287, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.65262222290039, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8826272487640381, + "num_tokens": 332342192.0, + "step": 8564 + }, + { + "epoch": 1.089556036127719, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.678756713867188, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8802007436752319, + "num_tokens": 332386387.0, + "step": 8565 + }, + { + "epoch": 1.0896832464063095, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.675604820251465, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8873087167739868, + "num_tokens": 332423265.0, + "step": 8566 + }, + { + "epoch": 1.0898104566849, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.67997932434082, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8847850561141968, + "num_tokens": 332460504.0, + "step": 8567 + }, + { + "epoch": 1.0899376669634906, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.569326400756836, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8874922394752502, + "num_tokens": 332505551.0, + "step": 8568 + }, + { + "epoch": 1.0900648772420811, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.565462112426758, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8900353312492371, + "num_tokens": 332548419.0, + "step": 8569 + }, + { + "epoch": 1.0901920875206716, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.576617240905762, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8863093852996826, + "num_tokens": 332591107.0, + "step": 8570 + }, + { + "epoch": 1.0903192977992622, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.741084098815918, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8777917623519897, + "num_tokens": 332629831.0, + "step": 8571 + }, + { + "epoch": 1.0904465080778527, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.933906555175781e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.640470504760742, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8774147033691406, + "num_tokens": 332673589.0, + "step": 8572 + }, + { + "epoch": 1.0905737183564432, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.707322120666504, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8917321562767029, + "num_tokens": 332711339.0, + "step": 8573 + }, + { + "epoch": 1.0907009286350338, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.777806282043457, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8840250372886658, + "num_tokens": 332741144.0, + "step": 8574 + }, + { + "epoch": 1.0908281389136243, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.672110557556152, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8886629343032837, + "num_tokens": 332780994.0, + "step": 8575 + }, + { + "epoch": 1.0909553491922148, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.657958030700684, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8817788362503052, + "num_tokens": 332821112.0, + "step": 8576 + }, + { + "epoch": 1.0910825594708053, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.629315376281738, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8913307189941406, + "num_tokens": 332858294.0, + "step": 8577 + }, + { + "epoch": 1.0912097697493957, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.649566650390625, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.89296555519104, + "num_tokens": 332897751.0, + "step": 8578 + }, + { + "epoch": 1.0913369800279862, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.680877685546875, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.8734566569328308, + "num_tokens": 332940263.0, + "step": 8579 + }, + { + "epoch": 1.0914641903065767, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.676074981689453, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8802664279937744, + "num_tokens": 332978254.0, + "step": 8580 + }, + { + "epoch": 1.0915914005851672, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.585304260253906, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8998391628265381, + "num_tokens": 333018622.0, + "step": 8581 + }, + { + "epoch": 1.0917186108637578, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.601200103759766, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8885462880134583, + "num_tokens": 333058782.0, + "step": 8582 + }, + { + "epoch": 1.0918458211423483, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.707049369812012, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.89408278465271, + "num_tokens": 333099224.0, + "step": 8583 + }, + { + "epoch": 1.0919730314209388, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.723179817199707, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8842096328735352, + "num_tokens": 333136032.0, + "step": 8584 + }, + { + "epoch": 1.0921002416995294, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.645828247070312, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8870725631713867, + "num_tokens": 333178119.0, + "step": 8585 + }, + { + "epoch": 1.0922274519781199, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.592504501342773, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8965727090835571, + "num_tokens": 333220327.0, + "step": 8586 + }, + { + "epoch": 1.0923546622567104, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.743769645690918, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8845533132553101, + "num_tokens": 333258578.0, + "step": 8587 + }, + { + "epoch": 1.092481872535301, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.653721809387207, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8889234662055969, + "num_tokens": 333301067.0, + "step": 8588 + }, + { + "epoch": 1.0926090828138915, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.703442573547363, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8899117708206177, + "num_tokens": 333338221.0, + "step": 8589 + }, + { + "epoch": 1.0927362930924818, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.630363464355469, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8756788969039917, + "num_tokens": 333383719.0, + "step": 8590 + }, + { + "epoch": 1.0928635033710723, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.636632919311523, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8803873062133789, + "num_tokens": 333422322.0, + "step": 8591 + }, + { + "epoch": 1.0929907136496628, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.585668563842773, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8905025720596313, + "num_tokens": 333462892.0, + "step": 8592 + }, + { + "epoch": 1.0931179239282534, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.668487548828125, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.8756435513496399, + "num_tokens": 333503032.0, + "step": 8593 + }, + { + "epoch": 1.0932451342068439, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.640576362609863, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8837355375289917, + "num_tokens": 333545656.0, + "step": 8594 + }, + { + "epoch": 1.0933723444854344, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.75536823272705, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8769593834877014, + "num_tokens": 333581851.0, + "step": 8595 + }, + { + "epoch": 1.093499554764025, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.658935546875, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.887071430683136, + "num_tokens": 333619053.0, + "step": 8596 + }, + { + "epoch": 1.0936267650426155, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.59145736694336, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8793761730194092, + "num_tokens": 333661314.0, + "step": 8597 + }, + { + "epoch": 1.093753975321206, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.594871520996094, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8834009170532227, + "num_tokens": 333700021.0, + "step": 8598 + }, + { + "epoch": 1.0938811855997965, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.643646240234375, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8938791751861572, + "num_tokens": 333738199.0, + "step": 8599 + }, + { + "epoch": 1.094008395878387, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.617170333862305, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8831740021705627, + "num_tokens": 333781895.0, + "step": 8600 + }, + { + "epoch": 1.0941356061569776, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.633923530578613, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8893252015113831, + "num_tokens": 333822138.0, + "step": 8601 + }, + { + "epoch": 1.094262816435568, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.700624465942383, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8835049867630005, + "num_tokens": 333860950.0, + "step": 8602 + }, + { + "epoch": 1.0943900267141584, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.619186401367188, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8892824649810791, + "num_tokens": 333902477.0, + "step": 8603 + }, + { + "epoch": 1.094517236992749, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.724508285522461, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8829387426376343, + "num_tokens": 333940654.0, + "step": 8604 + }, + { + "epoch": 1.0946444472713395, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.79841423034668, + "learning_rate": 1e-06, + "loss": 0.4056, + "mean_token_accuracy": 0.8615971207618713, + "num_tokens": 333976524.0, + "step": 8605 + }, + { + "epoch": 1.09477165754993, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.705099105834961, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8869642615318298, + "num_tokens": 334010956.0, + "step": 8606 + }, + { + "epoch": 1.0948988678285205, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.700338363647461, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8723957538604736, + "num_tokens": 334047314.0, + "step": 8607 + }, + { + "epoch": 1.095026078107111, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.642391204833984, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.884657621383667, + "num_tokens": 334086456.0, + "step": 8608 + }, + { + "epoch": 1.0951532883857016, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.69434642791748, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8867216110229492, + "num_tokens": 334124303.0, + "step": 8609 + }, + { + "epoch": 1.0952804986642921, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.65947437286377, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8911235332489014, + "num_tokens": 334166709.0, + "step": 8610 + }, + { + "epoch": 1.0954077089428826, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.700824737548828, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8922941088676453, + "num_tokens": 334202728.0, + "step": 8611 + }, + { + "epoch": 1.0955349192214732, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.755500793457031, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8785768747329712, + "num_tokens": 334240041.0, + "step": 8612 + }, + { + "epoch": 1.0956621295000637, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.63477897644043, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8825050592422485, + "num_tokens": 334278042.0, + "step": 8613 + }, + { + "epoch": 1.095789339778654, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.705488204956055, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8838880062103271, + "num_tokens": 334314505.0, + "step": 8614 + }, + { + "epoch": 1.0959165500572445, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.699461936950684, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.877071738243103, + "num_tokens": 334352218.0, + "step": 8615 + }, + { + "epoch": 1.096043760335835, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.669912338256836, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8720052242279053, + "num_tokens": 334396181.0, + "step": 8616 + }, + { + "epoch": 1.0961709706144256, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.800667762756348, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8830575942993164, + "num_tokens": 334426609.0, + "step": 8617 + }, + { + "epoch": 1.0962981808930161, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.64876937866211, + "learning_rate": 1e-06, + "loss": 0.3859, + "mean_token_accuracy": 0.8703345060348511, + "num_tokens": 334465115.0, + "step": 8618 + }, + { + "epoch": 1.0964253911716066, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.633975982666016, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8945125937461853, + "num_tokens": 334499606.0, + "step": 8619 + }, + { + "epoch": 1.0965526014501972, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.666882514953613, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8752955794334412, + "num_tokens": 334539498.0, + "step": 8620 + }, + { + "epoch": 1.0966798117287877, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.646958351135254, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8849141597747803, + "num_tokens": 334576538.0, + "step": 8621 + }, + { + "epoch": 1.0968070220073782, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.691985130310059, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8964241147041321, + "num_tokens": 334610883.0, + "step": 8622 + }, + { + "epoch": 1.0969342322859688, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.7159423828125, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8836262226104736, + "num_tokens": 334648508.0, + "step": 8623 + }, + { + "epoch": 1.0970614425645593, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.65272331237793, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8887799978256226, + "num_tokens": 334686621.0, + "step": 8624 + }, + { + "epoch": 1.0971886528431498, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.818860054016113, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8782294392585754, + "num_tokens": 334719875.0, + "step": 8625 + }, + { + "epoch": 1.0973158631217403, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.58386516571045, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8959869742393494, + "num_tokens": 334760645.0, + "step": 8626 + }, + { + "epoch": 1.0974430734003306, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.750239372253418, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8877491354942322, + "num_tokens": 334797664.0, + "step": 8627 + }, + { + "epoch": 1.0975702836789212, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.737849235534668, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.8719093203544617, + "num_tokens": 334843714.0, + "step": 8628 + }, + { + "epoch": 1.0976974939575117, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 3.9577484130859375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.624781608581543, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8917081952095032, + "num_tokens": 334880217.0, + "step": 8629 + }, + { + "epoch": 1.0978247042361022, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.751241683959961, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8931242227554321, + "num_tokens": 334917646.0, + "step": 8630 + }, + { + "epoch": 1.0979519145146928, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.68503475189209, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9009409546852112, + "num_tokens": 334953036.0, + "step": 8631 + }, + { + "epoch": 1.0980791247932833, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.603689193725586, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.896108090877533, + "num_tokens": 334997448.0, + "step": 8632 + }, + { + "epoch": 1.0982063350718738, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.64107608795166, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8769134283065796, + "num_tokens": 335037414.0, + "step": 8633 + }, + { + "epoch": 1.0983335453504643, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.750662803649902, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8874797821044922, + "num_tokens": 335071835.0, + "step": 8634 + }, + { + "epoch": 1.0984607556290549, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.681800842285156, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8834472298622131, + "num_tokens": 335113978.0, + "step": 8635 + }, + { + "epoch": 1.0985879659076454, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.695172309875488, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8723649382591248, + "num_tokens": 335151568.0, + "step": 8636 + }, + { + "epoch": 1.098715176186236, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.70662784576416, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8787316679954529, + "num_tokens": 335188169.0, + "step": 8637 + }, + { + "epoch": 1.0988423864648265, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.621167182922363, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8870067596435547, + "num_tokens": 335227420.0, + "step": 8638 + }, + { + "epoch": 1.0989695967434168, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.59085750579834, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8980738520622253, + "num_tokens": 335266981.0, + "step": 8639 + }, + { + "epoch": 1.0990968070220073, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.720918655395508, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8847748041152954, + "num_tokens": 335306032.0, + "step": 8640 + }, + { + "epoch": 1.0992240173005978, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.71043872833252, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8886107206344604, + "num_tokens": 335343830.0, + "step": 8641 + }, + { + "epoch": 1.0993512275791884, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.650703430175781, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8856028318405151, + "num_tokens": 335384362.0, + "step": 8642 + }, + { + "epoch": 1.0994784378577789, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.692526817321777, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8929616808891296, + "num_tokens": 335420263.0, + "step": 8643 + }, + { + "epoch": 1.0996056481363694, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.728766441345215, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8826931118965149, + "num_tokens": 335456551.0, + "step": 8644 + }, + { + "epoch": 1.09973285841496, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.727910041809082, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8771852850914001, + "num_tokens": 335497712.0, + "step": 8645 + }, + { + "epoch": 1.0998600686935505, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.655413627624512, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8914052248001099, + "num_tokens": 335534389.0, + "step": 8646 + }, + { + "epoch": 1.099987278972141, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.766785621643066, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8782963752746582, + "num_tokens": 335574179.0, + "step": 8647 + }, + { + "epoch": 1.1001144892507315, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.602316856384277, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8918560147285461, + "num_tokens": 335616319.0, + "step": 8648 + }, + { + "epoch": 1.100241699529322, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.777496337890625, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8862310647964478, + "num_tokens": 335650590.0, + "step": 8649 + }, + { + "epoch": 1.1003689098079126, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.785805702209473, + "learning_rate": 1e-06, + "loss": 0.4035, + "mean_token_accuracy": 0.864648699760437, + "num_tokens": 335689445.0, + "step": 8650 + }, + { + "epoch": 1.100496120086503, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.645222663879395, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8804422616958618, + "num_tokens": 335733382.0, + "step": 8651 + }, + { + "epoch": 1.1006233303650934, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.633891105651855, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.896777331829071, + "num_tokens": 335771590.0, + "step": 8652 + }, + { + "epoch": 1.100750540643684, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.622990608215332, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8945961594581604, + "num_tokens": 335810119.0, + "step": 8653 + }, + { + "epoch": 1.1008777509222745, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.720980644226074, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8838643431663513, + "num_tokens": 335849952.0, + "step": 8654 + }, + { + "epoch": 1.101004961200865, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.613195419311523, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8824346661567688, + "num_tokens": 335889629.0, + "step": 8655 + }, + { + "epoch": 1.1011321714794555, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.63137149810791, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8909242153167725, + "num_tokens": 335928997.0, + "step": 8656 + }, + { + "epoch": 1.101259381758046, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.79111385345459, + "learning_rate": 1e-06, + "loss": 0.3769, + "mean_token_accuracy": 0.8707035779953003, + "num_tokens": 335965640.0, + "step": 8657 + }, + { + "epoch": 1.1013865920366366, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.689879417419434, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8893486261367798, + "num_tokens": 336000426.0, + "step": 8658 + }, + { + "epoch": 1.101513802315227, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.60999584197998, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.882880449295044, + "num_tokens": 336043242.0, + "step": 8659 + }, + { + "epoch": 1.1016410125938176, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.619980812072754, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.879670262336731, + "num_tokens": 336087467.0, + "step": 8660 + }, + { + "epoch": 1.1017682228724082, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.669464111328125, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8848460912704468, + "num_tokens": 336125210.0, + "step": 8661 + }, + { + "epoch": 1.1018954331509987, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.643465995788574, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8863794207572937, + "num_tokens": 336166506.0, + "step": 8662 + }, + { + "epoch": 1.102022643429589, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.5720796585083, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8988941311836243, + "num_tokens": 336209854.0, + "step": 8663 + }, + { + "epoch": 1.1021498537081795, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.657745361328125, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8845950365066528, + "num_tokens": 336250287.0, + "step": 8664 + }, + { + "epoch": 1.10227706398677, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.730756759643555, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8742353320121765, + "num_tokens": 336290926.0, + "step": 8665 + }, + { + "epoch": 1.1024042742653606, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.588967323303223, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8985334634780884, + "num_tokens": 336335255.0, + "step": 8666 + }, + { + "epoch": 1.1025314845439511, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.715507507324219, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.889485776424408, + "num_tokens": 336372665.0, + "step": 8667 + }, + { + "epoch": 1.1026586948225416, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.701536178588867, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8821455240249634, + "num_tokens": 336412984.0, + "step": 8668 + }, + { + "epoch": 1.1027859051011322, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.734888076782227, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8731217384338379, + "num_tokens": 336450256.0, + "step": 8669 + }, + { + "epoch": 1.1029131153797227, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.65677547454834, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8794656991958618, + "num_tokens": 336492450.0, + "step": 8670 + }, + { + "epoch": 1.1030403256583132, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.668598175048828, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8874453902244568, + "num_tokens": 336529856.0, + "step": 8671 + }, + { + "epoch": 1.1031675359369038, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.661824226379395, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9045590758323669, + "num_tokens": 336570579.0, + "step": 8672 + }, + { + "epoch": 1.1032947462154943, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.971600532531738, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8872864246368408, + "num_tokens": 336599707.0, + "step": 8673 + }, + { + "epoch": 1.1034219564940848, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.652762413024902, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8948763608932495, + "num_tokens": 336636078.0, + "step": 8674 + }, + { + "epoch": 1.1035491667726753, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.803691864013672, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8902356028556824, + "num_tokens": 336677344.0, + "step": 8675 + }, + { + "epoch": 1.1036763770512656, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.71914291381836, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8783138394355774, + "num_tokens": 336716203.0, + "step": 8676 + }, + { + "epoch": 1.1038035873298562, + "ewc_loss": 0.01361083984375, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.629018783569336, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8862338662147522, + "num_tokens": 336756546.0, + "step": 8677 + }, + { + "epoch": 1.1039307976084467, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.716580390930176, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.881615161895752, + "num_tokens": 336795339.0, + "step": 8678 + }, + { + "epoch": 1.1040580078870372, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.708105087280273, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8850135207176208, + "num_tokens": 336836158.0, + "step": 8679 + }, + { + "epoch": 1.1041852181656278, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 3.981590270996094e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.80027961730957, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8935678005218506, + "num_tokens": 336873446.0, + "step": 8680 + }, + { + "epoch": 1.1043124284442183, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.785852432250977, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.892626166343689, + "num_tokens": 336906073.0, + "step": 8681 + }, + { + "epoch": 1.1044396387228088, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.680636405944824, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8892834186553955, + "num_tokens": 336943211.0, + "step": 8682 + }, + { + "epoch": 1.1045668490013993, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.711624145507812, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8709067106246948, + "num_tokens": 336989455.0, + "step": 8683 + }, + { + "epoch": 1.1046940592799899, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.67210865020752, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8810902833938599, + "num_tokens": 337031244.0, + "step": 8684 + }, + { + "epoch": 1.1048212695585804, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.748865127563477, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8884949684143066, + "num_tokens": 337069617.0, + "step": 8685 + }, + { + "epoch": 1.104948479837171, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.787382125854492, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8858590126037598, + "num_tokens": 337104705.0, + "step": 8686 + }, + { + "epoch": 1.1050756901157615, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.691359519958496, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.876920223236084, + "num_tokens": 337150911.0, + "step": 8687 + }, + { + "epoch": 1.1052029003943518, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.787001609802246, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.887917160987854, + "num_tokens": 337189988.0, + "step": 8688 + }, + { + "epoch": 1.1053301106729423, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.683615684509277, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8911711573600769, + "num_tokens": 337235886.0, + "step": 8689 + }, + { + "epoch": 1.1054573209515328, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.715507507324219, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8882583379745483, + "num_tokens": 337274008.0, + "step": 8690 + }, + { + "epoch": 1.1055845312301233, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.743667602539062, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8769614696502686, + "num_tokens": 337311626.0, + "step": 8691 + }, + { + "epoch": 1.1057117415087139, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.6327486038208, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8991431593894958, + "num_tokens": 337351740.0, + "step": 8692 + }, + { + "epoch": 1.1058389517873044, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.75528621673584, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8797683715820312, + "num_tokens": 337387861.0, + "step": 8693 + }, + { + "epoch": 1.105966162065895, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.71212100982666, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8859561681747437, + "num_tokens": 337426117.0, + "step": 8694 + }, + { + "epoch": 1.1060933723444855, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.55515193939209, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8900320529937744, + "num_tokens": 337472233.0, + "step": 8695 + }, + { + "epoch": 1.106220582623076, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.716972351074219, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8851022720336914, + "num_tokens": 337508724.0, + "step": 8696 + }, + { + "epoch": 1.1063477929016665, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.754777908325195, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8786028623580933, + "num_tokens": 337552544.0, + "step": 8697 + }, + { + "epoch": 1.106475003180257, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.610931396484375, + "learning_rate": 1e-06, + "loss": 0.3955, + "mean_token_accuracy": 0.8675945997238159, + "num_tokens": 337595430.0, + "step": 8698 + }, + { + "epoch": 1.1066022134588476, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.723966598510742, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8896673321723938, + "num_tokens": 337631582.0, + "step": 8699 + }, + { + "epoch": 1.106729423737438, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.618517875671387, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.888342022895813, + "num_tokens": 337668321.0, + "step": 8700 + }, + { + "epoch": 1.1068566340160284, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.651727676391602, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8874826431274414, + "num_tokens": 337709682.0, + "step": 8701 + }, + { + "epoch": 1.106983844294619, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.63315486907959, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8821009993553162, + "num_tokens": 337754153.0, + "step": 8702 + }, + { + "epoch": 1.1071110545732095, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.677997589111328, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.894093930721283, + "num_tokens": 337792378.0, + "step": 8703 + }, + { + "epoch": 1.1072382648518, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.745331764221191, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9020869135856628, + "num_tokens": 337827676.0, + "step": 8704 + }, + { + "epoch": 1.1073654751303905, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.627653121948242, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8975013494491577, + "num_tokens": 337865529.0, + "step": 8705 + }, + { + "epoch": 1.107492685408981, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.622464179992676, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8931148052215576, + "num_tokens": 337906933.0, + "step": 8706 + }, + { + "epoch": 1.1076198956875716, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.59532642364502, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8989877700805664, + "num_tokens": 337946806.0, + "step": 8707 + }, + { + "epoch": 1.107747105966162, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.692387580871582, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8800023794174194, + "num_tokens": 337989180.0, + "step": 8708 + }, + { + "epoch": 1.1078743162447526, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.691314697265625, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8759539127349854, + "num_tokens": 338031910.0, + "step": 8709 + }, + { + "epoch": 1.1080015265233432, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.652521133422852, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8947901129722595, + "num_tokens": 338070856.0, + "step": 8710 + }, + { + "epoch": 1.1081287368019337, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.630194664001465, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8922542333602905, + "num_tokens": 338112052.0, + "step": 8711 + }, + { + "epoch": 1.108255947080524, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.648327827453613, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8840696811676025, + "num_tokens": 338154493.0, + "step": 8712 + }, + { + "epoch": 1.1083831573591145, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.574906349182129, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8873506784439087, + "num_tokens": 338205803.0, + "step": 8713 + }, + { + "epoch": 1.108510367637705, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.807175636291504, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8815350532531738, + "num_tokens": 338240352.0, + "step": 8714 + }, + { + "epoch": 1.1086375779162956, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.681979179382324, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8879414200782776, + "num_tokens": 338281819.0, + "step": 8715 + }, + { + "epoch": 1.108764788194886, + "ewc_loss": 0.0135498046875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.5367431640625e-06, + "grad_norm": 8.760815620422363, + "learning_rate": 1e-06, + "loss": 0.3896, + "mean_token_accuracy": 0.874140739440918, + "num_tokens": 338319837.0, + "step": 8716 + }, + { + "epoch": 1.1088919984734766, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.67786693572998, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8960886597633362, + "num_tokens": 338357392.0, + "step": 8717 + }, + { + "epoch": 1.1090192087520672, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.65597915649414, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8928694128990173, + "num_tokens": 338395011.0, + "step": 8718 + }, + { + "epoch": 1.1091464190306577, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.809112548828125, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8722713589668274, + "num_tokens": 338429841.0, + "step": 8719 + }, + { + "epoch": 1.1092736293092482, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.650464057922363, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.875015139579773, + "num_tokens": 338470659.0, + "step": 8720 + }, + { + "epoch": 1.1094008395878387, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.677397727966309, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8942208290100098, + "num_tokens": 338507877.0, + "step": 8721 + }, + { + "epoch": 1.1095280498664293, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.72082233428955, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8876152038574219, + "num_tokens": 338542831.0, + "step": 8722 + }, + { + "epoch": 1.1096552601450198, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.70008659362793, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8929511308670044, + "num_tokens": 338576558.0, + "step": 8723 + }, + { + "epoch": 1.1097824704236103, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.635109901428223, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8854891657829285, + "num_tokens": 338618500.0, + "step": 8724 + }, + { + "epoch": 1.1099096807022006, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.70573616027832, + "learning_rate": 1e-06, + "loss": 0.3892, + "mean_token_accuracy": 0.8669953346252441, + "num_tokens": 338657605.0, + "step": 8725 + }, + { + "epoch": 1.1100368909807912, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.66219425201416, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.883639395236969, + "num_tokens": 338700689.0, + "step": 8726 + }, + { + "epoch": 1.1101641012593817, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.814675331115723, + "learning_rate": 1e-06, + "loss": 0.3671, + "mean_token_accuracy": 0.8712232112884521, + "num_tokens": 338731925.0, + "step": 8727 + }, + { + "epoch": 1.1102913115379722, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.61960506439209, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8844956159591675, + "num_tokens": 338775365.0, + "step": 8728 + }, + { + "epoch": 1.1104185218165628, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.696372032165527, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8877614736557007, + "num_tokens": 338809825.0, + "step": 8729 + }, + { + "epoch": 1.1105457320951533, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.767660140991211, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8823592662811279, + "num_tokens": 338839485.0, + "step": 8730 + }, + { + "epoch": 1.1106729423737438, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.715994834899902, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8818684816360474, + "num_tokens": 338876884.0, + "step": 8731 + }, + { + "epoch": 1.1108001526523343, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.660601615905762, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8864818811416626, + "num_tokens": 338912693.0, + "step": 8732 + }, + { + "epoch": 1.1109273629309249, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.807311058044434, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8753111362457275, + "num_tokens": 338943572.0, + "step": 8733 + }, + { + "epoch": 1.1110545732095154, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.59395694732666, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8977932333946228, + "num_tokens": 338983641.0, + "step": 8734 + }, + { + "epoch": 1.111181783488106, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.65602970123291, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8891704082489014, + "num_tokens": 339020911.0, + "step": 8735 + }, + { + "epoch": 1.1113089937666965, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.625982284545898, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8960713148117065, + "num_tokens": 339062708.0, + "step": 8736 + }, + { + "epoch": 1.1114362040452868, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.810037612915039, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8911533951759338, + "num_tokens": 339098449.0, + "step": 8737 + }, + { + "epoch": 1.1115634143238773, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.817366600036621, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8850396871566772, + "num_tokens": 339130036.0, + "step": 8738 + }, + { + "epoch": 1.1116906246024678, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.667997360229492, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8873295783996582, + "num_tokens": 339168815.0, + "step": 8739 + }, + { + "epoch": 1.1118178348810583, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.556295394897461, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8903540372848511, + "num_tokens": 339213949.0, + "step": 8740 + }, + { + "epoch": 1.1119450451596489, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.758735656738281, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.885116696357727, + "num_tokens": 339249408.0, + "step": 8741 + }, + { + "epoch": 1.1120722554382394, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.68093204498291, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8912087678909302, + "num_tokens": 339284243.0, + "step": 8742 + }, + { + "epoch": 1.11219946571683, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.629256248474121, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8897988796234131, + "num_tokens": 339325311.0, + "step": 8743 + }, + { + "epoch": 1.1123266759954205, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.673500061035156, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8904451727867126, + "num_tokens": 339363875.0, + "step": 8744 + }, + { + "epoch": 1.112453886274011, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.739297866821289, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.892292857170105, + "num_tokens": 339399524.0, + "step": 8745 + }, + { + "epoch": 1.1125810965526015, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.73558235168457, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8878998756408691, + "num_tokens": 339437590.0, + "step": 8746 + }, + { + "epoch": 1.112708306831192, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.661176681518555, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8927045464515686, + "num_tokens": 339471778.0, + "step": 8747 + }, + { + "epoch": 1.1128355171097826, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.693028450012207, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8983904719352722, + "num_tokens": 339510300.0, + "step": 8748 + }, + { + "epoch": 1.112962727388373, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.533781051635742, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8809750080108643, + "num_tokens": 339557392.0, + "step": 8749 + }, + { + "epoch": 1.1130899376669634, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.727402687072754, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8892727494239807, + "num_tokens": 339594872.0, + "step": 8750 + }, + { + "epoch": 1.113217147945554, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.720551490783691, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8842707872390747, + "num_tokens": 339628930.0, + "step": 8751 + }, + { + "epoch": 1.1133443582241445, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.718070983886719, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8821007609367371, + "num_tokens": 339664326.0, + "step": 8752 + }, + { + "epoch": 1.113471568502735, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.66104507446289, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8887988328933716, + "num_tokens": 339705404.0, + "step": 8753 + }, + { + "epoch": 1.1135987787813255, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.680314064025879, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8776915669441223, + "num_tokens": 339745365.0, + "step": 8754 + }, + { + "epoch": 1.113725989059916, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.655454635620117, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8828405141830444, + "num_tokens": 339783374.0, + "step": 8755 + }, + { + "epoch": 1.1138531993385066, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.739816665649414, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8892543315887451, + "num_tokens": 339817194.0, + "step": 8756 + }, + { + "epoch": 1.113980409617097, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.577800750732422, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8985803127288818, + "num_tokens": 339859176.0, + "step": 8757 + }, + { + "epoch": 1.1141076198956876, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.639128684997559, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8836683630943298, + "num_tokens": 339905261.0, + "step": 8758 + }, + { + "epoch": 1.1142348301742782, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.735941886901855, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8927419185638428, + "num_tokens": 339941673.0, + "step": 8759 + }, + { + "epoch": 1.1143620404528687, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.599518775939941, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8951022624969482, + "num_tokens": 339983490.0, + "step": 8760 + }, + { + "epoch": 1.114489250731459, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.668456077575684, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8846619129180908, + "num_tokens": 340025134.0, + "step": 8761 + }, + { + "epoch": 1.1146164610100495, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.615427017211914, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8880126476287842, + "num_tokens": 340065007.0, + "step": 8762 + }, + { + "epoch": 1.11474367128864, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.756661415100098, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8877823948860168, + "num_tokens": 340100728.0, + "step": 8763 + }, + { + "epoch": 1.1148708815672306, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.683272361755371, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8788987398147583, + "num_tokens": 340143401.0, + "step": 8764 + }, + { + "epoch": 1.114998091845821, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.69168758392334, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8960790634155273, + "num_tokens": 340179646.0, + "step": 8765 + }, + { + "epoch": 1.1151253021244116, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.75644302368164, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8810211420059204, + "num_tokens": 340217870.0, + "step": 8766 + }, + { + "epoch": 1.1152525124030022, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.660100936889648, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8798304796218872, + "num_tokens": 340257090.0, + "step": 8767 + }, + { + "epoch": 1.1153797226815927, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.00543212890625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.688246726989746, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8897932767868042, + "num_tokens": 340293569.0, + "step": 8768 + }, + { + "epoch": 1.1155069329601832, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.656768798828125, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.9009160995483398, + "num_tokens": 340331090.0, + "step": 8769 + }, + { + "epoch": 1.1156341432387737, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.605681419372559, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8908296823501587, + "num_tokens": 340373719.0, + "step": 8770 + }, + { + "epoch": 1.1157613535173643, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.80722713470459, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8808202743530273, + "num_tokens": 340407977.0, + "step": 8771 + }, + { + "epoch": 1.1158885637959548, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.676612854003906, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8960433006286621, + "num_tokens": 340441828.0, + "step": 8772 + }, + { + "epoch": 1.1160157740745453, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.717470169067383, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.890433669090271, + "num_tokens": 340477838.0, + "step": 8773 + }, + { + "epoch": 1.1161429843531356, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.607876777648926, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8915495276451111, + "num_tokens": 340519295.0, + "step": 8774 + }, + { + "epoch": 1.1162701946317262, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.662093162536621, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8855284452438354, + "num_tokens": 340559377.0, + "step": 8775 + }, + { + "epoch": 1.1163974049103167, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.753499031066895, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8919639587402344, + "num_tokens": 340588963.0, + "step": 8776 + }, + { + "epoch": 1.1165246151889072, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.623544692993164, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9075146317481995, + "num_tokens": 340629636.0, + "step": 8777 + }, + { + "epoch": 1.1166518254674977, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.683079719543457, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8853322267532349, + "num_tokens": 340669550.0, + "step": 8778 + }, + { + "epoch": 1.1167790357460883, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.653654098510742, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8845130205154419, + "num_tokens": 340710955.0, + "step": 8779 + }, + { + "epoch": 1.1169062460246788, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.654488563537598, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.8804792165756226, + "num_tokens": 340755527.0, + "step": 8780 + }, + { + "epoch": 1.1170334563032693, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.773975372314453, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8857839107513428, + "num_tokens": 340788815.0, + "step": 8781 + }, + { + "epoch": 1.1171606665818599, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.717164993286133, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.887418270111084, + "num_tokens": 340822377.0, + "step": 8782 + }, + { + "epoch": 1.1172878768604504, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.65240478515625, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8949390053749084, + "num_tokens": 340862855.0, + "step": 8783 + }, + { + "epoch": 1.117415087139041, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.735527038574219, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8904348015785217, + "num_tokens": 340903594.0, + "step": 8784 + }, + { + "epoch": 1.1175422974176314, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.643577575683594, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8918057680130005, + "num_tokens": 340941463.0, + "step": 8785 + }, + { + "epoch": 1.1176695076962218, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.747367858886719, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.8740785717964172, + "num_tokens": 340976934.0, + "step": 8786 + }, + { + "epoch": 1.1177967179748123, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.773613929748535, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8972432017326355, + "num_tokens": 341012647.0, + "step": 8787 + }, + { + "epoch": 1.1179239282534028, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.648512840270996, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8880457282066345, + "num_tokens": 341052132.0, + "step": 8788 + }, + { + "epoch": 1.1180511385319933, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.603387832641602, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8870277404785156, + "num_tokens": 341103782.0, + "step": 8789 + }, + { + "epoch": 1.1181783488105839, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.636731147766113, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8981515765190125, + "num_tokens": 341141819.0, + "step": 8790 + }, + { + "epoch": 1.1183055590891744, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.029273986816406e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.678180694580078, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.895733118057251, + "num_tokens": 341176953.0, + "step": 8791 + }, + { + "epoch": 1.118432769367765, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.65226936340332, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.9011105298995972, + "num_tokens": 341218129.0, + "step": 8792 + }, + { + "epoch": 1.1185599796463555, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.712021827697754, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.879734992980957, + "num_tokens": 341257174.0, + "step": 8793 + }, + { + "epoch": 1.118687189924946, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.707013130187988, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8869346976280212, + "num_tokens": 341295443.0, + "step": 8794 + }, + { + "epoch": 1.1188144002035365, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.664151191711426, + "learning_rate": 1e-06, + "loss": 0.3631, + "mean_token_accuracy": 0.8788345456123352, + "num_tokens": 341335283.0, + "step": 8795 + }, + { + "epoch": 1.118941610482127, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.703200340270996, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8812443017959595, + "num_tokens": 341378315.0, + "step": 8796 + }, + { + "epoch": 1.1190688207607176, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.604995727539062, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8903685212135315, + "num_tokens": 341422770.0, + "step": 8797 + }, + { + "epoch": 1.119196031039308, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.659942626953125, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8892489075660706, + "num_tokens": 341461711.0, + "step": 8798 + }, + { + "epoch": 1.1193232413178984, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.855340003967285, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8758724927902222, + "num_tokens": 341492523.0, + "step": 8799 + }, + { + "epoch": 1.119450451596489, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.691943168640137, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8910008668899536, + "num_tokens": 341528240.0, + "step": 8800 + }, + { + "epoch": 1.1195776618750795, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.727330207824707, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.893455982208252, + "num_tokens": 341566540.0, + "step": 8801 + }, + { + "epoch": 1.11970487215367, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.70081615447998, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8928119540214539, + "num_tokens": 341608941.0, + "step": 8802 + }, + { + "epoch": 1.1198320824322605, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.60720443725586, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8817489147186279, + "num_tokens": 341652141.0, + "step": 8803 + }, + { + "epoch": 1.119959292710851, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.726211547851562, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8934242129325867, + "num_tokens": 341689354.0, + "step": 8804 + }, + { + "epoch": 1.1200865029894416, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.668802261352539, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8933764696121216, + "num_tokens": 341729665.0, + "step": 8805 + }, + { + "epoch": 1.120213713268032, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.572877883911133, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9041240215301514, + "num_tokens": 341769154.0, + "step": 8806 + }, + { + "epoch": 1.1203409235466226, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.752904891967773, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8846632838249207, + "num_tokens": 341805117.0, + "step": 8807 + }, + { + "epoch": 1.1204681338252132, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.94631290435791, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8997703194618225, + "num_tokens": 341844688.0, + "step": 8808 + }, + { + "epoch": 1.1205953441038037, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.69104290008545, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8871426582336426, + "num_tokens": 341883937.0, + "step": 8809 + }, + { + "epoch": 1.120722554382394, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.625347137451172, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8921800255775452, + "num_tokens": 341924594.0, + "step": 8810 + }, + { + "epoch": 1.1208497646609845, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.651517868041992, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8906065225601196, + "num_tokens": 341962695.0, + "step": 8811 + }, + { + "epoch": 1.120976974939575, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.610891342163086, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8859654664993286, + "num_tokens": 342006031.0, + "step": 8812 + }, + { + "epoch": 1.1211041852181656, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.615460395812988, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9096147418022156, + "num_tokens": 342046320.0, + "step": 8813 + }, + { + "epoch": 1.121231395496756, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.582030296325684, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8888704776763916, + "num_tokens": 342091237.0, + "step": 8814 + }, + { + "epoch": 1.1213586057753466, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.770404815673828, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8955299854278564, + "num_tokens": 342125208.0, + "step": 8815 + }, + { + "epoch": 1.1214858160539372, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.619169235229492, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8842818737030029, + "num_tokens": 342168637.0, + "step": 8816 + }, + { + "epoch": 1.1216130263325277, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.59475326538086, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8910924196243286, + "num_tokens": 342215262.0, + "step": 8817 + }, + { + "epoch": 1.1217402366111182, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.676508903503418, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8994127511978149, + "num_tokens": 342254539.0, + "step": 8818 + }, + { + "epoch": 1.1218674468897087, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.733147621154785, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.889936089515686, + "num_tokens": 342290490.0, + "step": 8819 + }, + { + "epoch": 1.1219946571682993, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.799184799194336, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8912107944488525, + "num_tokens": 342322626.0, + "step": 8820 + }, + { + "epoch": 1.1221218674468898, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.643439292907715, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8834285736083984, + "num_tokens": 342361739.0, + "step": 8821 + }, + { + "epoch": 1.1222490777254803, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.718893051147461, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8978044986724854, + "num_tokens": 342394326.0, + "step": 8822 + }, + { + "epoch": 1.1223762880040706, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.78381633758545, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8752546906471252, + "num_tokens": 342429498.0, + "step": 8823 + }, + { + "epoch": 1.1225034982826612, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.73107624053955, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8776029348373413, + "num_tokens": 342467347.0, + "step": 8824 + }, + { + "epoch": 1.1226307085612517, + "ewc_loss": 0.013671875, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.649245262145996, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8822252750396729, + "num_tokens": 342507245.0, + "step": 8825 + }, + { + "epoch": 1.1227579188398422, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.720623016357422, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8861162662506104, + "num_tokens": 342544090.0, + "step": 8826 + }, + { + "epoch": 1.1228851291184327, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.663460731506348, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8768657445907593, + "num_tokens": 342584657.0, + "step": 8827 + }, + { + "epoch": 1.1230123393970233, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.69437026977539, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.884842574596405, + "num_tokens": 342625455.0, + "step": 8828 + }, + { + "epoch": 1.1231395496756138, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.855803489685059, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8765358328819275, + "num_tokens": 342655944.0, + "step": 8829 + }, + { + "epoch": 1.1232667599542043, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.738253593444824, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8814351558685303, + "num_tokens": 342693378.0, + "step": 8830 + }, + { + "epoch": 1.1233939702327949, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.617329597473145, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8910229206085205, + "num_tokens": 342730274.0, + "step": 8831 + }, + { + "epoch": 1.1235211805113854, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.653114318847656, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8932496905326843, + "num_tokens": 342766517.0, + "step": 8832 + }, + { + "epoch": 1.123648390789976, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.631275177001953, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8924187421798706, + "num_tokens": 342808216.0, + "step": 8833 + }, + { + "epoch": 1.1237756010685664, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.6117525100708, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8960830569267273, + "num_tokens": 342848169.0, + "step": 8834 + }, + { + "epoch": 1.1239028113471567, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.74145793914795, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8957772254943848, + "num_tokens": 342880399.0, + "step": 8835 + }, + { + "epoch": 1.1240300216257473, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.0531158447265625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.76486587524414, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8766355514526367, + "num_tokens": 342918731.0, + "step": 8836 + }, + { + "epoch": 1.1241572319043378, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.6682710647583, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8823421597480774, + "num_tokens": 342955568.0, + "step": 8837 + }, + { + "epoch": 1.1242844421829283, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.687623977661133, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8917244672775269, + "num_tokens": 342996236.0, + "step": 8838 + }, + { + "epoch": 1.1244116524615189, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.674605369567871, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8847756385803223, + "num_tokens": 343039113.0, + "step": 8839 + }, + { + "epoch": 1.1245388627401094, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.86195182800293, + "learning_rate": 1e-06, + "loss": 0.3872, + "mean_token_accuracy": 0.8718583583831787, + "num_tokens": 343074065.0, + "step": 8840 + }, + { + "epoch": 1.1246660730187, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.698994636535645, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8917520046234131, + "num_tokens": 343112858.0, + "step": 8841 + }, + { + "epoch": 1.1247932832972904, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.711630821228027, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8791483640670776, + "num_tokens": 343153016.0, + "step": 8842 + }, + { + "epoch": 1.124920493575881, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.98547077178955, + "learning_rate": 1e-06, + "loss": 0.3904, + "mean_token_accuracy": 0.8709249496459961, + "num_tokens": 343183006.0, + "step": 8843 + }, + { + "epoch": 1.1250477038544715, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.668923377990723, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8862786293029785, + "num_tokens": 343223323.0, + "step": 8844 + }, + { + "epoch": 1.125174914133062, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.6318941116333, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8986006379127502, + "num_tokens": 343263608.0, + "step": 8845 + }, + { + "epoch": 1.1253021244116526, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.75053882598877, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8902412056922913, + "num_tokens": 343298350.0, + "step": 8846 + }, + { + "epoch": 1.125429334690243, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.85317611694336, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8776496648788452, + "num_tokens": 343329687.0, + "step": 8847 + }, + { + "epoch": 1.1255565449688334, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.63656234741211, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8811544179916382, + "num_tokens": 343371058.0, + "step": 8848 + }, + { + "epoch": 1.125683755247424, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.693782806396484, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8849947452545166, + "num_tokens": 343412159.0, + "step": 8849 + }, + { + "epoch": 1.1258109655260145, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.680002212524414, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8917660713195801, + "num_tokens": 343450035.0, + "step": 8850 + }, + { + "epoch": 1.125938175804605, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.678138732910156, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8795844912528992, + "num_tokens": 343492100.0, + "step": 8851 + }, + { + "epoch": 1.1260653860831955, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.72555160522461, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8865983486175537, + "num_tokens": 343526112.0, + "step": 8852 + }, + { + "epoch": 1.126192596361786, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.750916481018066, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.8741270303726196, + "num_tokens": 343564043.0, + "step": 8853 + }, + { + "epoch": 1.1263198066403766, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.659832954406738, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8818100690841675, + "num_tokens": 343603642.0, + "step": 8854 + }, + { + "epoch": 1.126447016918967, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.076957702636719e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.66217041015625, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.9014368057250977, + "num_tokens": 343641744.0, + "step": 8855 + }, + { + "epoch": 1.1265742271975576, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.717700958251953, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8757538795471191, + "num_tokens": 343679059.0, + "step": 8856 + }, + { + "epoch": 1.1267014374761481, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.540950775146484, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9020609855651855, + "num_tokens": 343723310.0, + "step": 8857 + }, + { + "epoch": 1.1268286477547387, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.713543891906738, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.9006472229957581, + "num_tokens": 343759166.0, + "step": 8858 + }, + { + "epoch": 1.126955858033329, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.794952392578125, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8934550881385803, + "num_tokens": 343792733.0, + "step": 8859 + }, + { + "epoch": 1.1270830683119195, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.683233261108398, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8825607299804688, + "num_tokens": 343829370.0, + "step": 8860 + }, + { + "epoch": 1.12721027859051, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.689556121826172, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8782617449760437, + "num_tokens": 343871306.0, + "step": 8861 + }, + { + "epoch": 1.1273374888691006, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.651413917541504, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8856668472290039, + "num_tokens": 343915246.0, + "step": 8862 + }, + { + "epoch": 1.127464699147691, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.732050895690918, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8934468030929565, + "num_tokens": 343954453.0, + "step": 8863 + }, + { + "epoch": 1.1275919094262816, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.77662467956543, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8832070231437683, + "num_tokens": 343991552.0, + "step": 8864 + }, + { + "epoch": 1.1277191197048722, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.741216659545898, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8848072290420532, + "num_tokens": 344028588.0, + "step": 8865 + }, + { + "epoch": 1.1278463299834627, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.804208755493164, + "learning_rate": 1e-06, + "loss": 0.3811, + "mean_token_accuracy": 0.871503472328186, + "num_tokens": 344068585.0, + "step": 8866 + }, + { + "epoch": 1.1279735402620532, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.62350082397461, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8953878879547119, + "num_tokens": 344108643.0, + "step": 8867 + }, + { + "epoch": 1.1281007505406437, + "ewc_loss": 0.01373291015625, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.734145164489746, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8860476016998291, + "num_tokens": 344142092.0, + "step": 8868 + }, + { + "epoch": 1.1282279608192343, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.773551940917969, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8990271687507629, + "num_tokens": 344174938.0, + "step": 8869 + }, + { + "epoch": 1.1283551710978248, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.653205871582031, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8923599720001221, + "num_tokens": 344218311.0, + "step": 8870 + }, + { + "epoch": 1.1284823813764153, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.788787841796875, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8841477632522583, + "num_tokens": 344255217.0, + "step": 8871 + }, + { + "epoch": 1.1286095916550056, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.670072555541992, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8847808837890625, + "num_tokens": 344294787.0, + "step": 8872 + }, + { + "epoch": 1.1287368019335962, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.827629089355469, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8917067050933838, + "num_tokens": 344328537.0, + "step": 8873 + }, + { + "epoch": 1.1288640122121867, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.719071388244629, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8926352858543396, + "num_tokens": 344363634.0, + "step": 8874 + }, + { + "epoch": 1.1289912224907772, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.747271537780762, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8729414939880371, + "num_tokens": 344406614.0, + "step": 8875 + }, + { + "epoch": 1.1291184327693677, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.581988334655762, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8779505491256714, + "num_tokens": 344457321.0, + "step": 8876 + }, + { + "epoch": 1.1292456430479583, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.750946998596191, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8893966674804688, + "num_tokens": 344490158.0, + "step": 8877 + }, + { + "epoch": 1.1293728533265488, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.77266788482666, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8911664485931396, + "num_tokens": 344525483.0, + "step": 8878 + }, + { + "epoch": 1.1295000636051393, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.645288467407227, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8924858570098877, + "num_tokens": 344565341.0, + "step": 8879 + }, + { + "epoch": 1.1296272738837299, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.809002876281738, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8769465088844299, + "num_tokens": 344603335.0, + "step": 8880 + }, + { + "epoch": 1.1297544841623204, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.708571434020996, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8867156505584717, + "num_tokens": 344642542.0, + "step": 8881 + }, + { + "epoch": 1.129881694440911, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.651434898376465, + "learning_rate": 1e-06, + "loss": 0.3475, + "mean_token_accuracy": 0.879096269607544, + "num_tokens": 344682220.0, + "step": 8882 + }, + { + "epoch": 1.1300089047195012, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.776302337646484, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8805557489395142, + "num_tokens": 344720062.0, + "step": 8883 + }, + { + "epoch": 1.1301361149980917, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.552629470825195, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9044827222824097, + "num_tokens": 344763534.0, + "step": 8884 + }, + { + "epoch": 1.1302633252766823, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.700028419494629, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8879055976867676, + "num_tokens": 344807602.0, + "step": 8885 + }, + { + "epoch": 1.1303905355552728, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.788647651672363, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8829683065414429, + "num_tokens": 344844438.0, + "step": 8886 + }, + { + "epoch": 1.1305177458338633, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.820184707641602, + "learning_rate": 1e-06, + "loss": 0.3861, + "mean_token_accuracy": 0.869289755821228, + "num_tokens": 344882573.0, + "step": 8887 + }, + { + "epoch": 1.1306449561124539, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.755770683288574, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8864446878433228, + "num_tokens": 344918779.0, + "step": 8888 + }, + { + "epoch": 1.1307721663910444, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.888643264770508, + "learning_rate": 1e-06, + "loss": 0.4001, + "mean_token_accuracy": 0.8684537410736084, + "num_tokens": 344953050.0, + "step": 8889 + }, + { + "epoch": 1.130899376669635, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.76286506652832, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8834810853004456, + "num_tokens": 344988656.0, + "step": 8890 + }, + { + "epoch": 1.1310265869482254, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.668500900268555, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8896757960319519, + "num_tokens": 345027262.0, + "step": 8891 + }, + { + "epoch": 1.131153797226816, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.79738712310791, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8889451622962952, + "num_tokens": 345061481.0, + "step": 8892 + }, + { + "epoch": 1.1312810075054065, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.7014741897583, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.895173966884613, + "num_tokens": 345099556.0, + "step": 8893 + }, + { + "epoch": 1.131408217783997, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.764917373657227, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8799601197242737, + "num_tokens": 345138984.0, + "step": 8894 + }, + { + "epoch": 1.1315354280625876, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.669492721557617, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.89745032787323, + "num_tokens": 345174210.0, + "step": 8895 + }, + { + "epoch": 1.131662638341178, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.67908000946045, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8898798227310181, + "num_tokens": 345216243.0, + "step": 8896 + }, + { + "epoch": 1.1317898486197684, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.671900749206543, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.890842080116272, + "num_tokens": 345259040.0, + "step": 8897 + }, + { + "epoch": 1.131917058898359, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.78562068939209, + "learning_rate": 1e-06, + "loss": 0.4115, + "mean_token_accuracy": 0.864825427532196, + "num_tokens": 345300325.0, + "step": 8898 + }, + { + "epoch": 1.1320442691769494, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.793654441833496, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8841230869293213, + "num_tokens": 345339096.0, + "step": 8899 + }, + { + "epoch": 1.13217147945554, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.651113510131836, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8833884000778198, + "num_tokens": 345382450.0, + "step": 8900 + }, + { + "epoch": 1.1322986897341305, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.62905216217041, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8979036808013916, + "num_tokens": 345425832.0, + "step": 8901 + }, + { + "epoch": 1.132425900012721, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.673333168029785, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8875244855880737, + "num_tokens": 345464099.0, + "step": 8902 + }, + { + "epoch": 1.1325531102913116, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.740446090698242, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8742191791534424, + "num_tokens": 345505248.0, + "step": 8903 + }, + { + "epoch": 1.132680320569902, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.706101417541504, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8785485029220581, + "num_tokens": 345544257.0, + "step": 8904 + }, + { + "epoch": 1.1328075308484926, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.674622535705566, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8997174501419067, + "num_tokens": 345584700.0, + "step": 8905 + }, + { + "epoch": 1.1329347411270831, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.636930465698242, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8955094814300537, + "num_tokens": 345625231.0, + "step": 8906 + }, + { + "epoch": 1.1330619514056737, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.724703788757324, + "learning_rate": 1e-06, + "loss": 0.3597, + "mean_token_accuracy": 0.8761433362960815, + "num_tokens": 345670011.0, + "step": 8907 + }, + { + "epoch": 1.133189161684264, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.650054931640625, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8912523984909058, + "num_tokens": 345713166.0, + "step": 8908 + }, + { + "epoch": 1.1333163719628545, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.677482604980469, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8779706954956055, + "num_tokens": 345758693.0, + "step": 8909 + }, + { + "epoch": 1.133443582241445, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.770957946777344, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8817145824432373, + "num_tokens": 345795947.0, + "step": 8910 + }, + { + "epoch": 1.1335707925200356, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.100799560546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.769464492797852, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8817287683486938, + "num_tokens": 345828920.0, + "step": 8911 + }, + { + "epoch": 1.133698002798626, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.69825553894043, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.876613974571228, + "num_tokens": 345872148.0, + "step": 8912 + }, + { + "epoch": 1.1338252130772166, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.726691246032715, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8876936435699463, + "num_tokens": 345909489.0, + "step": 8913 + }, + { + "epoch": 1.1339524233558071, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.654253005981445, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.871317982673645, + "num_tokens": 345952078.0, + "step": 8914 + }, + { + "epoch": 1.1340796336343977, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.692720413208008, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8888318538665771, + "num_tokens": 345990762.0, + "step": 8915 + }, + { + "epoch": 1.1342068439129882, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.697298049926758, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.881169319152832, + "num_tokens": 346034288.0, + "step": 8916 + }, + { + "epoch": 1.1343340541915787, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.633783340454102, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8943570256233215, + "num_tokens": 346072287.0, + "step": 8917 + }, + { + "epoch": 1.1344612644701693, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.65436840057373, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8866193294525146, + "num_tokens": 346117747.0, + "step": 8918 + }, + { + "epoch": 1.1345884747487598, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.743070602416992, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8868526220321655, + "num_tokens": 346154323.0, + "step": 8919 + }, + { + "epoch": 1.1347156850273503, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.695856094360352, + "learning_rate": 1e-06, + "loss": 0.3835, + "mean_token_accuracy": 0.8728740811347961, + "num_tokens": 346197954.0, + "step": 8920 + }, + { + "epoch": 1.1348428953059406, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.713005065917969, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8879198431968689, + "num_tokens": 346238464.0, + "step": 8921 + }, + { + "epoch": 1.1349701055845312, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.64377498626709, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8965091705322266, + "num_tokens": 346275183.0, + "step": 8922 + }, + { + "epoch": 1.1350973158631217, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.777030944824219, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8848801851272583, + "num_tokens": 346314611.0, + "step": 8923 + }, + { + "epoch": 1.1352245261417122, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.631084442138672, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8915599584579468, + "num_tokens": 346362018.0, + "step": 8924 + }, + { + "epoch": 1.1353517364203027, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.618491172790527, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8974626660346985, + "num_tokens": 346408092.0, + "step": 8925 + }, + { + "epoch": 1.1354789466988933, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.822937965393066, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8882549405097961, + "num_tokens": 346443730.0, + "step": 8926 + }, + { + "epoch": 1.1356061569774838, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.713098526000977, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8878942131996155, + "num_tokens": 346481676.0, + "step": 8927 + }, + { + "epoch": 1.1357333672560743, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.821978569030762, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8816715478897095, + "num_tokens": 346516943.0, + "step": 8928 + }, + { + "epoch": 1.1358605775346649, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.825285911560059, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8978877067565918, + "num_tokens": 346546402.0, + "step": 8929 + }, + { + "epoch": 1.1359877878132554, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.629929542541504, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.890448808670044, + "num_tokens": 346592211.0, + "step": 8930 + }, + { + "epoch": 1.136114998091846, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.814751625061035, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8774176836013794, + "num_tokens": 346627079.0, + "step": 8931 + }, + { + "epoch": 1.1362422083704362, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.834625244140625, + "learning_rate": 1e-06, + "loss": 0.3979, + "mean_token_accuracy": 0.8656104207038879, + "num_tokens": 346663098.0, + "step": 8932 + }, + { + "epoch": 1.1363694186490267, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.736288070678711, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8855938911437988, + "num_tokens": 346699598.0, + "step": 8933 + }, + { + "epoch": 1.1364966289276173, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.778687477111816, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8878848552703857, + "num_tokens": 346734723.0, + "step": 8934 + }, + { + "epoch": 1.1366238392062078, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.742940902709961, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8973618745803833, + "num_tokens": 346769588.0, + "step": 8935 + }, + { + "epoch": 1.1367510494847983, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.665949821472168, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.891409158706665, + "num_tokens": 346808474.0, + "step": 8936 + }, + { + "epoch": 1.1368782597633889, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.828701972961426, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8902965188026428, + "num_tokens": 346840761.0, + "step": 8937 + }, + { + "epoch": 1.1370054700419794, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.715882301330566, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8908361196517944, + "num_tokens": 346881043.0, + "step": 8938 + }, + { + "epoch": 1.13713268032057, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.690437316894531, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8832290172576904, + "num_tokens": 346920132.0, + "step": 8939 + }, + { + "epoch": 1.1372598905991604, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.779073715209961, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8870731592178345, + "num_tokens": 346955954.0, + "step": 8940 + }, + { + "epoch": 1.137387100877751, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.600934982299805, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8868514895439148, + "num_tokens": 347001600.0, + "step": 8941 + }, + { + "epoch": 1.1375143111563415, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.72747802734375, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8828827142715454, + "num_tokens": 347038016.0, + "step": 8942 + }, + { + "epoch": 1.137641521434932, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.82033920288086, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.886188805103302, + "num_tokens": 347068898.0, + "step": 8943 + }, + { + "epoch": 1.1377687317135226, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.64089298248291, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8890866041183472, + "num_tokens": 347113586.0, + "step": 8944 + }, + { + "epoch": 1.137895941992113, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.652241706848145, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8770428895950317, + "num_tokens": 347160500.0, + "step": 8945 + }, + { + "epoch": 1.1380231522707034, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.696516036987305, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8801337480545044, + "num_tokens": 347200639.0, + "step": 8946 + }, + { + "epoch": 1.138150362549294, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.855548858642578, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8805570006370544, + "num_tokens": 347234001.0, + "step": 8947 + }, + { + "epoch": 1.1382775728278844, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.670027732849121, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.874052882194519, + "num_tokens": 347279249.0, + "step": 8948 + }, + { + "epoch": 1.138404783106475, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.752643585205078, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8807133436203003, + "num_tokens": 347316611.0, + "step": 8949 + }, + { + "epoch": 1.1385319933850655, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.798227310180664, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8970019817352295, + "num_tokens": 347348044.0, + "step": 8950 + }, + { + "epoch": 1.138659203663656, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.647954940795898, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8866214752197266, + "num_tokens": 347393195.0, + "step": 8951 + }, + { + "epoch": 1.1387864139422466, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.65489387512207, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8921284675598145, + "num_tokens": 347432311.0, + "step": 8952 + }, + { + "epoch": 1.138913624220837, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.751813888549805, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8825942873954773, + "num_tokens": 347473493.0, + "step": 8953 + }, + { + "epoch": 1.1390408344994276, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.666070938110352, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8886739015579224, + "num_tokens": 347510355.0, + "step": 8954 + }, + { + "epoch": 1.1391680447780181, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.737656593322754, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8858209848403931, + "num_tokens": 347554748.0, + "step": 8955 + }, + { + "epoch": 1.1392952550566087, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.773388862609863, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8790225386619568, + "num_tokens": 347590276.0, + "step": 8956 + }, + { + "epoch": 1.139422465335199, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.696371078491211, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8876183032989502, + "num_tokens": 347632665.0, + "step": 8957 + }, + { + "epoch": 1.1395496756137895, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.717535972595215, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.9083225727081299, + "num_tokens": 347667689.0, + "step": 8958 + }, + { + "epoch": 1.13967688589238, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.70788288116455, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8830561637878418, + "num_tokens": 347709253.0, + "step": 8959 + }, + { + "epoch": 1.1398040961709706, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.679067611694336, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8854382634162903, + "num_tokens": 347752130.0, + "step": 8960 + }, + { + "epoch": 1.139931306449561, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.730447769165039, + "learning_rate": 1e-06, + "loss": 0.3834, + "mean_token_accuracy": 0.8687688708305359, + "num_tokens": 347791882.0, + "step": 8961 + }, + { + "epoch": 1.1400585167281516, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.802746772766113, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8788015246391296, + "num_tokens": 347830436.0, + "step": 8962 + }, + { + "epoch": 1.1401857270067421, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.70711898803711, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8866398334503174, + "num_tokens": 347870909.0, + "step": 8963 + }, + { + "epoch": 1.1403129372853327, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.671653747558594, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8814183473587036, + "num_tokens": 347915625.0, + "step": 8964 + }, + { + "epoch": 1.1404401475639232, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.728800773620605, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8886253237724304, + "num_tokens": 347949799.0, + "step": 8965 + }, + { + "epoch": 1.1405673578425137, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.68436050415039, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.9005757570266724, + "num_tokens": 347985196.0, + "step": 8966 + }, + { + "epoch": 1.1406945681211043, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.703619003295898, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8910788297653198, + "num_tokens": 348026638.0, + "step": 8967 + }, + { + "epoch": 1.1408217783996948, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.675355911254883, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8906446695327759, + "num_tokens": 348065688.0, + "step": 8968 + }, + { + "epoch": 1.1409489886782853, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.756841659545898, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8828243613243103, + "num_tokens": 348105121.0, + "step": 8969 + }, + { + "epoch": 1.1410761989568756, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.124641418457031e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.765885353088379, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8847414255142212, + "num_tokens": 348141348.0, + "step": 8970 + }, + { + "epoch": 1.1412034092354661, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.800055503845215, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8868649005889893, + "num_tokens": 348176131.0, + "step": 8971 + }, + { + "epoch": 1.1413306195140567, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.699447631835938, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8879920244216919, + "num_tokens": 348211671.0, + "step": 8972 + }, + { + "epoch": 1.1414578297926472, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.606769561767578, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8918649554252625, + "num_tokens": 348257613.0, + "step": 8973 + }, + { + "epoch": 1.1415850400712377, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.682259559631348, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8885055780410767, + "num_tokens": 348299864.0, + "step": 8974 + }, + { + "epoch": 1.1417122503498283, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.676201820373535, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8885195255279541, + "num_tokens": 348337263.0, + "step": 8975 + }, + { + "epoch": 1.1418394606284188, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.711477279663086, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8963829278945923, + "num_tokens": 348374130.0, + "step": 8976 + }, + { + "epoch": 1.1419666709070093, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.640976905822754, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8954641819000244, + "num_tokens": 348415888.0, + "step": 8977 + }, + { + "epoch": 1.1420938811855998, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.68167781829834, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.909618616104126, + "num_tokens": 348452850.0, + "step": 8978 + }, + { + "epoch": 1.1422210914641904, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.848827362060547, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8761278390884399, + "num_tokens": 348486728.0, + "step": 8979 + }, + { + "epoch": 1.142348301742781, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.691838264465332, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8898725509643555, + "num_tokens": 348526591.0, + "step": 8980 + }, + { + "epoch": 1.1424755120213712, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.69999885559082, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8819298148155212, + "num_tokens": 348568572.0, + "step": 8981 + }, + { + "epoch": 1.1426027222999617, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.772052764892578, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8991701602935791, + "num_tokens": 348600902.0, + "step": 8982 + }, + { + "epoch": 1.1427299325785523, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.781937599182129, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8890568017959595, + "num_tokens": 348636162.0, + "step": 8983 + }, + { + "epoch": 1.1428571428571428, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.734745979309082, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.886654257774353, + "num_tokens": 348674000.0, + "step": 8984 + }, + { + "epoch": 1.1429843531357333, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.716845512390137, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8823460340499878, + "num_tokens": 348715524.0, + "step": 8985 + }, + { + "epoch": 1.1431115634143239, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.673426628112793, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8933175206184387, + "num_tokens": 348753071.0, + "step": 8986 + }, + { + "epoch": 1.1432387736929144, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.899995803833008, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8817432522773743, + "num_tokens": 348789299.0, + "step": 8987 + }, + { + "epoch": 1.143365983971505, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.829763412475586, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.8694645166397095, + "num_tokens": 348821974.0, + "step": 8988 + }, + { + "epoch": 1.1434931942500954, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.740941047668457, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8846619129180908, + "num_tokens": 348858004.0, + "step": 8989 + }, + { + "epoch": 1.143620404528686, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.686782836914062, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8795440196990967, + "num_tokens": 348899605.0, + "step": 8990 + }, + { + "epoch": 1.1437476148072765, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.768838882446289, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8826782703399658, + "num_tokens": 348936119.0, + "step": 8991 + }, + { + "epoch": 1.143874825085867, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.706698417663574, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.880155622959137, + "num_tokens": 348973607.0, + "step": 8992 + }, + { + "epoch": 1.1440020353644575, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.778468132019043, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8977759480476379, + "num_tokens": 349006480.0, + "step": 8993 + }, + { + "epoch": 1.144129245643048, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.717135429382324, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.899938702583313, + "num_tokens": 349039281.0, + "step": 8994 + }, + { + "epoch": 1.1442564559216384, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.68644905090332, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8854146003723145, + "num_tokens": 349082091.0, + "step": 8995 + }, + { + "epoch": 1.144383666200229, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.638368606567383, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8917030096054077, + "num_tokens": 349124281.0, + "step": 8996 + }, + { + "epoch": 1.1445108764788194, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.712849617004395, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8920186758041382, + "num_tokens": 349160814.0, + "step": 8997 + }, + { + "epoch": 1.14463808675741, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.772582054138184, + "learning_rate": 1e-06, + "loss": 0.3714, + "mean_token_accuracy": 0.8746241927146912, + "num_tokens": 349199943.0, + "step": 8998 + }, + { + "epoch": 1.1447652970360005, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.635868072509766, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.901294469833374, + "num_tokens": 349240787.0, + "step": 8999 + }, + { + "epoch": 1.144892507314591, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.64714241027832, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8819721341133118, + "num_tokens": 349286839.0, + "step": 9000 + }, + { + "epoch": 1.1450197175931816, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.741880416870117, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8808286190032959, + "num_tokens": 349323683.0, + "step": 9001 + }, + { + "epoch": 1.145146927871772, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.90035343170166, + "learning_rate": 1e-06, + "loss": 0.4395, + "mean_token_accuracy": 0.853775680065155, + "num_tokens": 349363028.0, + "step": 9002 + }, + { + "epoch": 1.1452741381503626, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.679023742675781, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8820894956588745, + "num_tokens": 349408957.0, + "step": 9003 + }, + { + "epoch": 1.1454013484289531, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.68025016784668, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8858605623245239, + "num_tokens": 349451579.0, + "step": 9004 + }, + { + "epoch": 1.1455285587075437, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.805571556091309, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8756133913993835, + "num_tokens": 349488658.0, + "step": 9005 + }, + { + "epoch": 1.145655768986134, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.637079238891602, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8968836069107056, + "num_tokens": 349529378.0, + "step": 9006 + }, + { + "epoch": 1.1457829792647245, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.764062881469727, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8938441276550293, + "num_tokens": 349563742.0, + "step": 9007 + }, + { + "epoch": 1.145910189543315, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.674938201904297, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.891814112663269, + "num_tokens": 349602831.0, + "step": 9008 + }, + { + "epoch": 1.1460373998219056, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.722566604614258, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8871662616729736, + "num_tokens": 349635430.0, + "step": 9009 + }, + { + "epoch": 1.146164610100496, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.713150978088379, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8919622898101807, + "num_tokens": 349674334.0, + "step": 9010 + }, + { + "epoch": 1.1462918203790866, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.73816967010498, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8978127241134644, + "num_tokens": 349707264.0, + "step": 9011 + }, + { + "epoch": 1.1464190306576771, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.683178901672363, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8969733119010925, + "num_tokens": 349748139.0, + "step": 9012 + }, + { + "epoch": 1.1465462409362677, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.704048156738281, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8845148682594299, + "num_tokens": 349791988.0, + "step": 9013 + }, + { + "epoch": 1.1466734512148582, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.706255912780762, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8873566389083862, + "num_tokens": 349829881.0, + "step": 9014 + }, + { + "epoch": 1.1468006614934487, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.716681480407715, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8885764479637146, + "num_tokens": 349869231.0, + "step": 9015 + }, + { + "epoch": 1.1469278717720393, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.755392074584961, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8862404227256775, + "num_tokens": 349907860.0, + "step": 9016 + }, + { + "epoch": 1.1470550820506298, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.642306327819824, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8940399885177612, + "num_tokens": 349946313.0, + "step": 9017 + }, + { + "epoch": 1.1471822923292203, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.735522270202637, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8933124542236328, + "num_tokens": 349981851.0, + "step": 9018 + }, + { + "epoch": 1.1473095026078106, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.740622520446777, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8739122152328491, + "num_tokens": 350024056.0, + "step": 9019 + }, + { + "epoch": 1.1474367128864011, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.71534252166748, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8865575194358826, + "num_tokens": 350064061.0, + "step": 9020 + }, + { + "epoch": 1.1475639231649917, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.689318656921387, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8889746069908142, + "num_tokens": 350106145.0, + "step": 9021 + }, + { + "epoch": 1.1476911334435822, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.7170991897583, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8965049386024475, + "num_tokens": 350144780.0, + "step": 9022 + }, + { + "epoch": 1.1478183437221727, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.686786651611328, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.893254280090332, + "num_tokens": 350179427.0, + "step": 9023 + }, + { + "epoch": 1.1479455540007633, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.7882080078125, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8786741495132446, + "num_tokens": 350215813.0, + "step": 9024 + }, + { + "epoch": 1.1480727642793538, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.757794380187988, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8969757556915283, + "num_tokens": 350248325.0, + "step": 9025 + }, + { + "epoch": 1.1481999745579443, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.640443801879883, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8981614112854004, + "num_tokens": 350286305.0, + "step": 9026 + }, + { + "epoch": 1.1483271848365348, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.772618293762207, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.888957679271698, + "num_tokens": 350326344.0, + "step": 9027 + }, + { + "epoch": 1.1484543951151254, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.66242790222168, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.889575719833374, + "num_tokens": 350372287.0, + "step": 9028 + }, + { + "epoch": 1.148581605393716, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.763300895690918, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.883599042892456, + "num_tokens": 350414868.0, + "step": 9029 + }, + { + "epoch": 1.1487088156723062, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.744563102722168, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8981938362121582, + "num_tokens": 350452201.0, + "step": 9030 + }, + { + "epoch": 1.1488360259508967, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.79284954071045, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8743242025375366, + "num_tokens": 350494383.0, + "step": 9031 + }, + { + "epoch": 1.1489632362294873, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.826679229736328, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8818169832229614, + "num_tokens": 350533049.0, + "step": 9032 + }, + { + "epoch": 1.1490904465080778, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.735340118408203, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8964658379554749, + "num_tokens": 350572837.0, + "step": 9033 + }, + { + "epoch": 1.1492176567866683, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.878215789794922, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8770826458930969, + "num_tokens": 350605284.0, + "step": 9034 + }, + { + "epoch": 1.1493448670652588, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.792170524597168, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8778916597366333, + "num_tokens": 350644029.0, + "step": 9035 + }, + { + "epoch": 1.1494720773438494, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.778829574584961, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8817597031593323, + "num_tokens": 350680676.0, + "step": 9036 + }, + { + "epoch": 1.14959928762244, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.645013809204102, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8894977569580078, + "num_tokens": 350723315.0, + "step": 9037 + }, + { + "epoch": 1.1497264979010304, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.746184349060059, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8787054419517517, + "num_tokens": 350760815.0, + "step": 9038 + }, + { + "epoch": 1.149853708179621, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.766541481018066, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8760496377944946, + "num_tokens": 350803088.0, + "step": 9039 + }, + { + "epoch": 1.1499809184582115, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.67611312866211, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8961262702941895, + "num_tokens": 350841778.0, + "step": 9040 + }, + { + "epoch": 1.150108128736802, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.722394943237305, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8981777429580688, + "num_tokens": 350881707.0, + "step": 9041 + }, + { + "epoch": 1.1502353390153925, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.585370063781738, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8928125500679016, + "num_tokens": 350925066.0, + "step": 9042 + }, + { + "epoch": 1.150362549293983, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.786396026611328, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8902642726898193, + "num_tokens": 350964852.0, + "step": 9043 + }, + { + "epoch": 1.1504897595725734, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.88772201538086, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8940587043762207, + "num_tokens": 350999211.0, + "step": 9044 + }, + { + "epoch": 1.150616969851164, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.653021812438965, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8926012516021729, + "num_tokens": 351039995.0, + "step": 9045 + }, + { + "epoch": 1.1507441801297544, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.741477012634277, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8968081474304199, + "num_tokens": 351077047.0, + "step": 9046 + }, + { + "epoch": 1.150871390408345, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.729846954345703, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8885506391525269, + "num_tokens": 351115313.0, + "step": 9047 + }, + { + "epoch": 1.1509986006869355, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.685212135314941, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9037107229232788, + "num_tokens": 351150746.0, + "step": 9048 + }, + { + "epoch": 1.151125810965526, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.783445358276367, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8840740919113159, + "num_tokens": 351187596.0, + "step": 9049 + }, + { + "epoch": 1.1512530212441165, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.695394515991211, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8876582384109497, + "num_tokens": 351224177.0, + "step": 9050 + }, + { + "epoch": 1.151380231522707, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.740626335144043, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.896256685256958, + "num_tokens": 351261712.0, + "step": 9051 + }, + { + "epoch": 1.1515074418012976, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.678589820861816, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.903325080871582, + "num_tokens": 351301532.0, + "step": 9052 + }, + { + "epoch": 1.1516346520798881, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.671792030334473, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.9006249308586121, + "num_tokens": 351344571.0, + "step": 9053 + }, + { + "epoch": 1.1517618623584787, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.764233589172363, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.874786376953125, + "num_tokens": 351390213.0, + "step": 9054 + }, + { + "epoch": 1.151889072637069, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.722050666809082, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8853719234466553, + "num_tokens": 351432720.0, + "step": 9055 + }, + { + "epoch": 1.1520162829156595, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.834463119506836, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8791009187698364, + "num_tokens": 351473090.0, + "step": 9056 + }, + { + "epoch": 1.15214349319425, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.69262981414795, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8923934102058411, + "num_tokens": 351515798.0, + "step": 9057 + }, + { + "epoch": 1.1522707034728406, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.76777458190918, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8835384845733643, + "num_tokens": 351553869.0, + "step": 9058 + }, + { + "epoch": 1.152397913751431, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.784411430358887, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8821389675140381, + "num_tokens": 351594067.0, + "step": 9059 + }, + { + "epoch": 1.1525251240300216, + "ewc_loss": 0.0137939453125, + "ewc_loss_diag": 4.1484832763671875e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.716134071350098, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8882671594619751, + "num_tokens": 351632268.0, + "step": 9060 + }, + { + "epoch": 1.1526523343086121, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.825640678405762, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8798586130142212, + "num_tokens": 351670356.0, + "step": 9061 + }, + { + "epoch": 1.1527795445872027, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.677427291870117, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.895601749420166, + "num_tokens": 351709976.0, + "step": 9062 + }, + { + "epoch": 1.1529067548657932, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.854443550109863, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8798285126686096, + "num_tokens": 351744371.0, + "step": 9063 + }, + { + "epoch": 1.1530339651443837, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.694580078125, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8763569593429565, + "num_tokens": 351786986.0, + "step": 9064 + }, + { + "epoch": 1.1531611754229742, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.640100479125977, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9041934013366699, + "num_tokens": 351827460.0, + "step": 9065 + }, + { + "epoch": 1.1532883857015648, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.881548881530762, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8819172382354736, + "num_tokens": 351863029.0, + "step": 9066 + }, + { + "epoch": 1.1534155959801553, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.68321418762207, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8852206468582153, + "num_tokens": 351904463.0, + "step": 9067 + }, + { + "epoch": 1.1535428062587456, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.747167587280273, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8855647444725037, + "num_tokens": 351942272.0, + "step": 9068 + }, + { + "epoch": 1.1536700165373361, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.741127967834473, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8996309041976929, + "num_tokens": 351979357.0, + "step": 9069 + }, + { + "epoch": 1.1537972268159267, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.807229042053223, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8866608142852783, + "num_tokens": 352014171.0, + "step": 9070 + }, + { + "epoch": 1.1539244370945172, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.771583557128906, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8907319903373718, + "num_tokens": 352048805.0, + "step": 9071 + }, + { + "epoch": 1.1540516473731077, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.70221996307373, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8717361688613892, + "num_tokens": 352088285.0, + "step": 9072 + }, + { + "epoch": 1.1541788576516983, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.939205169677734, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8776392340660095, + "num_tokens": 352118495.0, + "step": 9073 + }, + { + "epoch": 1.1543060679302888, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.725160598754883, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8841116428375244, + "num_tokens": 352154155.0, + "step": 9074 + }, + { + "epoch": 1.1544332782088793, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.648767471313477, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8937647938728333, + "num_tokens": 352194862.0, + "step": 9075 + }, + { + "epoch": 1.1545604884874698, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.725336074829102, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8941092491149902, + "num_tokens": 352229872.0, + "step": 9076 + }, + { + "epoch": 1.1546876987660604, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.692606925964355, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8783689737319946, + "num_tokens": 352271429.0, + "step": 9077 + }, + { + "epoch": 1.154814909044651, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.696817398071289, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8891735672950745, + "num_tokens": 352313388.0, + "step": 9078 + }, + { + "epoch": 1.1549421193232412, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.89102554321289, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8775615692138672, + "num_tokens": 352346601.0, + "step": 9079 + }, + { + "epoch": 1.1550693296018317, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.630425453186035, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8923923969268799, + "num_tokens": 352389709.0, + "step": 9080 + }, + { + "epoch": 1.1551965398804223, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.682878494262695, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8893064260482788, + "num_tokens": 352430246.0, + "step": 9081 + }, + { + "epoch": 1.1553237501590128, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.6839017868042, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.884608805179596, + "num_tokens": 352471092.0, + "step": 9082 + }, + { + "epoch": 1.1554509604376033, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.6666259765625, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8960481286048889, + "num_tokens": 352511859.0, + "step": 9083 + }, + { + "epoch": 1.1555781707161938, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.749516487121582, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8784130811691284, + "num_tokens": 352553065.0, + "step": 9084 + }, + { + "epoch": 1.1557053809947844, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.639217376708984, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8915731310844421, + "num_tokens": 352597760.0, + "step": 9085 + }, + { + "epoch": 1.155832591273375, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.733102798461914, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8948724269866943, + "num_tokens": 352637383.0, + "step": 9086 + }, + { + "epoch": 1.1559598015519654, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.7695951461792, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8769551515579224, + "num_tokens": 352682979.0, + "step": 9087 + }, + { + "epoch": 1.156087011830556, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.687987327575684, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.899998664855957, + "num_tokens": 352719042.0, + "step": 9088 + }, + { + "epoch": 1.1562142221091465, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.724449157714844, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8848986625671387, + "num_tokens": 352757591.0, + "step": 9089 + }, + { + "epoch": 1.156341432387737, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.695221900939941, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.90367591381073, + "num_tokens": 352794787.0, + "step": 9090 + }, + { + "epoch": 1.1564686426663275, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.782561302185059, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8943544626235962, + "num_tokens": 352827178.0, + "step": 9091 + }, + { + "epoch": 1.156595852944918, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.646068572998047, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8870657682418823, + "num_tokens": 352871690.0, + "step": 9092 + }, + { + "epoch": 1.1567230632235084, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.749062538146973, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.9031796455383301, + "num_tokens": 352903915.0, + "step": 9093 + }, + { + "epoch": 1.156850273502099, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.781086921691895, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8898587226867676, + "num_tokens": 352938841.0, + "step": 9094 + }, + { + "epoch": 1.1569774837806894, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.702245712280273, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8963853716850281, + "num_tokens": 352973575.0, + "step": 9095 + }, + { + "epoch": 1.15710469405928, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.693675994873047, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8844618201255798, + "num_tokens": 353013781.0, + "step": 9096 + }, + { + "epoch": 1.1572319043378705, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.809636116027832, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8755578398704529, + "num_tokens": 353047751.0, + "step": 9097 + }, + { + "epoch": 1.157359114616461, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.71532917022705, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8955393433570862, + "num_tokens": 353083991.0, + "step": 9098 + }, + { + "epoch": 1.1574863248950515, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.705748558044434, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8719228506088257, + "num_tokens": 353127372.0, + "step": 9099 + }, + { + "epoch": 1.157613535173642, + "ewc_loss": 0.013916015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.712181091308594, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8997235298156738, + "num_tokens": 353167208.0, + "step": 9100 + }, + { + "epoch": 1.1577407454522326, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.74514389038086, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8822178840637207, + "num_tokens": 353204440.0, + "step": 9101 + }, + { + "epoch": 1.1578679557308231, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.7558012008667, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8960061073303223, + "num_tokens": 353241036.0, + "step": 9102 + }, + { + "epoch": 1.1579951660094137, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.172325134277344e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.658123970031738, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8919732570648193, + "num_tokens": 353283045.0, + "step": 9103 + }, + { + "epoch": 1.158122376288004, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.775044441223145, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8974189758300781, + "num_tokens": 353320304.0, + "step": 9104 + }, + { + "epoch": 1.1582495865665945, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.692502975463867, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8968437910079956, + "num_tokens": 353363740.0, + "step": 9105 + }, + { + "epoch": 1.158376796845185, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.824118614196777, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8826169967651367, + "num_tokens": 353404100.0, + "step": 9106 + }, + { + "epoch": 1.1585040071237755, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.75122356414795, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8920843601226807, + "num_tokens": 353443248.0, + "step": 9107 + }, + { + "epoch": 1.158631217402366, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.716775894165039, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8823161125183105, + "num_tokens": 353481822.0, + "step": 9108 + }, + { + "epoch": 1.1587584276809566, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.75623893737793, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8948870897293091, + "num_tokens": 353522654.0, + "step": 9109 + }, + { + "epoch": 1.1588856379595471, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.675917625427246, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8910105228424072, + "num_tokens": 353564114.0, + "step": 9110 + }, + { + "epoch": 1.1590128482381377, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.644708633422852, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8966314792633057, + "num_tokens": 353609908.0, + "step": 9111 + }, + { + "epoch": 1.1591400585167282, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.87539005279541, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8762043714523315, + "num_tokens": 353648884.0, + "step": 9112 + }, + { + "epoch": 1.1592672687953187, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.804308891296387, + "learning_rate": 1e-06, + "loss": 0.4129, + "mean_token_accuracy": 0.8574580550193787, + "num_tokens": 353691442.0, + "step": 9113 + }, + { + "epoch": 1.1593944790739092, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.817607879638672, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8832852244377136, + "num_tokens": 353725322.0, + "step": 9114 + }, + { + "epoch": 1.1595216893524998, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.719280242919922, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8931563496589661, + "num_tokens": 353762559.0, + "step": 9115 + }, + { + "epoch": 1.1596488996310903, + "ewc_loss": 0.01385498046875, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.655952453613281e-06, + "grad_norm": 8.681038856506348, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.893661618232727, + "num_tokens": 353800686.0, + "step": 9116 + }, + { + "epoch": 1.1597761099096806, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.764565467834473, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8824329972267151, + "num_tokens": 353840159.0, + "step": 9117 + }, + { + "epoch": 1.1599033201882711, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.220008850097656e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.712198257446289, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8824688792228699, + "num_tokens": 353878347.0, + "step": 9118 + }, + { + "epoch": 1.1600305304668617, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.821885108947754, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8935122489929199, + "num_tokens": 353916369.0, + "step": 9119 + }, + { + "epoch": 1.1601577407454522, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.644270896911621, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.89699387550354, + "num_tokens": 353953274.0, + "step": 9120 + }, + { + "epoch": 1.1602849510240427, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.65657901763916, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8954734802246094, + "num_tokens": 353993423.0, + "step": 9121 + }, + { + "epoch": 1.1604121613026332, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.84571361541748, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.886127233505249, + "num_tokens": 354025845.0, + "step": 9122 + }, + { + "epoch": 1.1605393715812238, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.880276679992676, + "learning_rate": 1e-06, + "loss": 0.4213, + "mean_token_accuracy": 0.8656886219978333, + "num_tokens": 354057177.0, + "step": 9123 + }, + { + "epoch": 1.1606665818598143, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.728181838989258, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.888888955116272, + "num_tokens": 354091011.0, + "step": 9124 + }, + { + "epoch": 1.1607937921384048, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.701066017150879, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8989359736442566, + "num_tokens": 354126592.0, + "step": 9125 + }, + { + "epoch": 1.1609210024169954, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.667251586914062, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8750650882720947, + "num_tokens": 354170948.0, + "step": 9126 + }, + { + "epoch": 1.161048212695586, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.681051254272461, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8867188692092896, + "num_tokens": 354213174.0, + "step": 9127 + }, + { + "epoch": 1.1611754229741762, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.764577865600586, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.88722825050354, + "num_tokens": 354250541.0, + "step": 9128 + }, + { + "epoch": 1.1613026332527667, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.676077842712402, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8909885287284851, + "num_tokens": 354289605.0, + "step": 9129 + }, + { + "epoch": 1.1614298435313573, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.778303146362305, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8822983503341675, + "num_tokens": 354330269.0, + "step": 9130 + }, + { + "epoch": 1.1615570538099478, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.671399116516113, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8866437673568726, + "num_tokens": 354367740.0, + "step": 9131 + }, + { + "epoch": 1.1616842640885383, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.70240306854248, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8867892026901245, + "num_tokens": 354410063.0, + "step": 9132 + }, + { + "epoch": 1.1618114743671288, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.752717971801758, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.9011125564575195, + "num_tokens": 354448999.0, + "step": 9133 + }, + { + "epoch": 1.1619386846457194, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.698821067810059, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8966262340545654, + "num_tokens": 354489276.0, + "step": 9134 + }, + { + "epoch": 1.16206589492431, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.765279769897461, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8790246844291687, + "num_tokens": 354532687.0, + "step": 9135 + }, + { + "epoch": 1.1621931052029004, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.738615989685059, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8776497840881348, + "num_tokens": 354575905.0, + "step": 9136 + }, + { + "epoch": 1.162320315481491, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.776344299316406, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8758755922317505, + "num_tokens": 354619284.0, + "step": 9137 + }, + { + "epoch": 1.1624475257600815, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.732481002807617, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.888824462890625, + "num_tokens": 354658443.0, + "step": 9138 + }, + { + "epoch": 1.162574736038672, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.729244232177734, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9076704978942871, + "num_tokens": 354691404.0, + "step": 9139 + }, + { + "epoch": 1.1627019463172625, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.656632423400879, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8870278596878052, + "num_tokens": 354733153.0, + "step": 9140 + }, + { + "epoch": 1.162829156595853, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.755199432373047, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8864568471908569, + "num_tokens": 354770539.0, + "step": 9141 + }, + { + "epoch": 1.1629563668744434, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.740533828735352, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.8707006573677063, + "num_tokens": 354814419.0, + "step": 9142 + }, + { + "epoch": 1.163083577153034, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.72326374053955, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.9016971588134766, + "num_tokens": 354847519.0, + "step": 9143 + }, + { + "epoch": 1.1632107874316244, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.709016799926758, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8903058767318726, + "num_tokens": 354886907.0, + "step": 9144 + }, + { + "epoch": 1.163337997710215, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.736549377441406, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8869448304176331, + "num_tokens": 354924382.0, + "step": 9145 + }, + { + "epoch": 1.1634652079888055, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.810189247131348, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8766970038414001, + "num_tokens": 354961566.0, + "step": 9146 + }, + { + "epoch": 1.163592418267396, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.691146850585938, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8830212354660034, + "num_tokens": 355001189.0, + "step": 9147 + }, + { + "epoch": 1.1637196285459865, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.784377098083496, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8961273431777954, + "num_tokens": 355038336.0, + "step": 9148 + }, + { + "epoch": 1.163846838824577, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.806126594543457, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9041107296943665, + "num_tokens": 355070699.0, + "step": 9149 + }, + { + "epoch": 1.1639740491031676, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.683048248291016, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8943174481391907, + "num_tokens": 355114595.0, + "step": 9150 + }, + { + "epoch": 1.1641012593817581, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.736756324768066, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8791495561599731, + "num_tokens": 355156523.0, + "step": 9151 + }, + { + "epoch": 1.1642284696603487, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.749951362609863, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8849051594734192, + "num_tokens": 355195980.0, + "step": 9152 + }, + { + "epoch": 1.164355679938939, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.796086311340332, + "learning_rate": 1e-06, + "loss": 0.3831, + "mean_token_accuracy": 0.8722759485244751, + "num_tokens": 355232735.0, + "step": 9153 + }, + { + "epoch": 1.1644828902175295, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.770960807800293, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8973373174667358, + "num_tokens": 355267122.0, + "step": 9154 + }, + { + "epoch": 1.16461010049612, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.70357608795166, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8928735256195068, + "num_tokens": 355305000.0, + "step": 9155 + }, + { + "epoch": 1.1647373107747105, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.695499420166016, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8885329961776733, + "num_tokens": 355349168.0, + "step": 9156 + }, + { + "epoch": 1.164864521053301, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.267692565917969e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.759018898010254, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8858435750007629, + "num_tokens": 355386904.0, + "step": 9157 + }, + { + "epoch": 1.1649917313318916, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.726512908935547, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8886366486549377, + "num_tokens": 355428022.0, + "step": 9158 + }, + { + "epoch": 1.1651189416104821, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.742273330688477, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8892300128936768, + "num_tokens": 355470913.0, + "step": 9159 + }, + { + "epoch": 1.1652461518890727, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.702339172363281, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8877306580543518, + "num_tokens": 355513158.0, + "step": 9160 + }, + { + "epoch": 1.1653733621676632, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.770790100097656, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8849266767501831, + "num_tokens": 355551866.0, + "step": 9161 + }, + { + "epoch": 1.1655005724462537, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.666186332702637, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.894939661026001, + "num_tokens": 355598976.0, + "step": 9162 + }, + { + "epoch": 1.1656277827248442, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.70677661895752, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.884039044380188, + "num_tokens": 355641524.0, + "step": 9163 + }, + { + "epoch": 1.1657549930034348, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.79624080657959, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.9032989740371704, + "num_tokens": 355672897.0, + "step": 9164 + }, + { + "epoch": 1.1658822032820253, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.742868423461914, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8833163380622864, + "num_tokens": 355711055.0, + "step": 9165 + }, + { + "epoch": 1.1660094135606156, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.743621826171875, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8835742473602295, + "num_tokens": 355754342.0, + "step": 9166 + }, + { + "epoch": 1.1661366238392061, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.886248588562012, + "learning_rate": 1e-06, + "loss": 0.4219, + "mean_token_accuracy": 0.8553671836853027, + "num_tokens": 355790493.0, + "step": 9167 + }, + { + "epoch": 1.1662638341177967, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.732455253601074, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.900238037109375, + "num_tokens": 355829294.0, + "step": 9168 + }, + { + "epoch": 1.1663910443963872, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.750240325927734, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8803000450134277, + "num_tokens": 355866642.0, + "step": 9169 + }, + { + "epoch": 1.1665182546749777, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.7550687789917, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8821662664413452, + "num_tokens": 355906942.0, + "step": 9170 + }, + { + "epoch": 1.1666454649535682, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.77729320526123, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8815048933029175, + "num_tokens": 355944101.0, + "step": 9171 + }, + { + "epoch": 1.1667726752321588, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.70977783203125, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8863728046417236, + "num_tokens": 355984895.0, + "step": 9172 + }, + { + "epoch": 1.1668998855107493, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.716439247131348, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8926941752433777, + "num_tokens": 356025776.0, + "step": 9173 + }, + { + "epoch": 1.1670270957893398, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.721691131591797, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.883479654788971, + "num_tokens": 356066756.0, + "step": 9174 + }, + { + "epoch": 1.1671543060679304, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.77454662322998, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8865234851837158, + "num_tokens": 356107418.0, + "step": 9175 + }, + { + "epoch": 1.1672815163465209, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.753405570983887, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8846055269241333, + "num_tokens": 356144147.0, + "step": 9176 + }, + { + "epoch": 1.1674087266251112, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.71010971069336, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8961923122406006, + "num_tokens": 356182688.0, + "step": 9177 + }, + { + "epoch": 1.1675359369037017, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.671537399291992, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.9037898778915405, + "num_tokens": 356221417.0, + "step": 9178 + }, + { + "epoch": 1.1676631471822922, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.315376281738281e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.808488845825195, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8763070106506348, + "num_tokens": 356259903.0, + "step": 9179 + }, + { + "epoch": 1.1677903574608828, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.787273406982422, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8764963150024414, + "num_tokens": 356299927.0, + "step": 9180 + }, + { + "epoch": 1.1679175677394733, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.714488983154297, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8897674083709717, + "num_tokens": 356337267.0, + "step": 9181 + }, + { + "epoch": 1.1680447780180638, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.711711883544922, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8924096822738647, + "num_tokens": 356378792.0, + "step": 9182 + }, + { + "epoch": 1.1681719882966544, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.803465843200684, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8845032453536987, + "num_tokens": 356414981.0, + "step": 9183 + }, + { + "epoch": 1.168299198575245, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.662912368774414, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8872125148773193, + "num_tokens": 356461289.0, + "step": 9184 + }, + { + "epoch": 1.1684264088538354, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.705442428588867, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8852308988571167, + "num_tokens": 356504631.0, + "step": 9185 + }, + { + "epoch": 1.168553619132426, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.68594741821289, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8978873491287231, + "num_tokens": 356545463.0, + "step": 9186 + }, + { + "epoch": 1.1686808294110165, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.77866268157959, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.881546139717102, + "num_tokens": 356583361.0, + "step": 9187 + }, + { + "epoch": 1.168808039689607, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.631430625915527, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8892391920089722, + "num_tokens": 356627088.0, + "step": 9188 + }, + { + "epoch": 1.1689352499681975, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.766810417175293, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.9023915529251099, + "num_tokens": 356664225.0, + "step": 9189 + }, + { + "epoch": 1.169062460246788, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.826874732971191, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8867865800857544, + "num_tokens": 356699930.0, + "step": 9190 + }, + { + "epoch": 1.1691896705253784, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.705154418945312, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8907142877578735, + "num_tokens": 356739832.0, + "step": 9191 + }, + { + "epoch": 1.169316880803969, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.766109466552734, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8871637582778931, + "num_tokens": 356780259.0, + "step": 9192 + }, + { + "epoch": 1.1694440910825594, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.630465507507324, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8848280906677246, + "num_tokens": 356825711.0, + "step": 9193 + }, + { + "epoch": 1.16957130136115, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.719006538391113, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.896385908126831, + "num_tokens": 356867495.0, + "step": 9194 + }, + { + "epoch": 1.1696985116397405, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.684661865234375, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8992536067962646, + "num_tokens": 356907593.0, + "step": 9195 + }, + { + "epoch": 1.169825721918331, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.789052963256836, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8868066072463989, + "num_tokens": 356951965.0, + "step": 9196 + }, + { + "epoch": 1.1699529321969215, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.783994674682617, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8818630576133728, + "num_tokens": 356992417.0, + "step": 9197 + }, + { + "epoch": 1.170080142475512, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.668965339660645, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.9030499458312988, + "num_tokens": 357033120.0, + "step": 9198 + }, + { + "epoch": 1.1702073527541026, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.707536697387695, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8876796960830688, + "num_tokens": 357076828.0, + "step": 9199 + }, + { + "epoch": 1.1703345630326931, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.727215766906738, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8771660923957825, + "num_tokens": 357120020.0, + "step": 9200 + }, + { + "epoch": 1.1704617733112836, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.692627906799316, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8953230381011963, + "num_tokens": 357160587.0, + "step": 9201 + }, + { + "epoch": 1.170588983589874, + "ewc_loss": 0.01409912109375, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.784384727478027, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8854665756225586, + "num_tokens": 357201801.0, + "step": 9202 + }, + { + "epoch": 1.1707161938684645, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.3392181396484375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.771130561828613, + "learning_rate": 1e-06, + "loss": 0.3784, + "mean_token_accuracy": 0.8730317950248718, + "num_tokens": 357241695.0, + "step": 9203 + }, + { + "epoch": 1.170843404147055, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.757532119750977, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8844399452209473, + "num_tokens": 357276963.0, + "step": 9204 + }, + { + "epoch": 1.1709706144256455, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.79776668548584, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8746767044067383, + "num_tokens": 357314997.0, + "step": 9205 + }, + { + "epoch": 1.171097824704236, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.637709617614746, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8984661102294922, + "num_tokens": 357352132.0, + "step": 9206 + }, + { + "epoch": 1.1712250349828266, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.700411796569824, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8864307403564453, + "num_tokens": 357396003.0, + "step": 9207 + }, + { + "epoch": 1.1713522452614171, + "ewc_loss": 0.0140380859375, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.67056941986084, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8916136026382446, + "num_tokens": 357440251.0, + "step": 9208 + }, + { + "epoch": 1.1714794555400077, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.723895072937012, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9033394455909729, + "num_tokens": 357478650.0, + "step": 9209 + }, + { + "epoch": 1.1716066658185982, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.774738311767578, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8871855735778809, + "num_tokens": 357515532.0, + "step": 9210 + }, + { + "epoch": 1.1717338760971887, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.874547958374023, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8856123685836792, + "num_tokens": 357549549.0, + "step": 9211 + }, + { + "epoch": 1.1718610863757792, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.688465118408203, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.895479679107666, + "num_tokens": 357590433.0, + "step": 9212 + }, + { + "epoch": 1.1719882966543698, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.756141662597656, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8864842057228088, + "num_tokens": 357634421.0, + "step": 9213 + }, + { + "epoch": 1.1721155069329603, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.720919609069824, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8868321180343628, + "num_tokens": 357675189.0, + "step": 9214 + }, + { + "epoch": 1.1722427172115506, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.783927917480469, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8845077157020569, + "num_tokens": 357714426.0, + "step": 9215 + }, + { + "epoch": 1.1723699274901411, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.718616485595703, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8826696872711182, + "num_tokens": 357757989.0, + "step": 9216 + }, + { + "epoch": 1.1724971377687317, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.780104637145996, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8895423412322998, + "num_tokens": 357796889.0, + "step": 9217 + }, + { + "epoch": 1.1726243480473222, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.782374382019043, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8795627355575562, + "num_tokens": 357833895.0, + "step": 9218 + }, + { + "epoch": 1.1727515583259127, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.781232833862305, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8728517889976501, + "num_tokens": 357874426.0, + "step": 9219 + }, + { + "epoch": 1.1728787686045032, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.696477890014648, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8804589509963989, + "num_tokens": 357918428.0, + "step": 9220 + }, + { + "epoch": 1.1730059788830938, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.907395362854004, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8747091889381409, + "num_tokens": 357954131.0, + "step": 9221 + }, + { + "epoch": 1.1731331891616843, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.705183029174805, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8931788802146912, + "num_tokens": 357993559.0, + "step": 9222 + }, + { + "epoch": 1.1732603994402748, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.728334426879883, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.892021894454956, + "num_tokens": 358032962.0, + "step": 9223 + }, + { + "epoch": 1.1733876097188654, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.735756874084473, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8881754875183105, + "num_tokens": 358073863.0, + "step": 9224 + }, + { + "epoch": 1.1735148199974559, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.798080444335938, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.87803053855896, + "num_tokens": 358116212.0, + "step": 9225 + }, + { + "epoch": 1.1736420302760462, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.885747909545898, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8878604173660278, + "num_tokens": 358147572.0, + "step": 9226 + }, + { + "epoch": 1.1737692405546367, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.7931547164917, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8755075931549072, + "num_tokens": 358191715.0, + "step": 9227 + }, + { + "epoch": 1.1738964508332272, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.784072875976562, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8848840594291687, + "num_tokens": 358230663.0, + "step": 9228 + }, + { + "epoch": 1.1740236611118178, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.738741874694824, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8942393064498901, + "num_tokens": 358274131.0, + "step": 9229 + }, + { + "epoch": 1.1741508713904083, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.363059997558594e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.815105438232422, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8835580348968506, + "num_tokens": 358318036.0, + "step": 9230 + }, + { + "epoch": 1.1742780816689988, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.38690185546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.698078155517578, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8948420286178589, + "num_tokens": 358360429.0, + "step": 9231 + }, + { + "epoch": 1.1744052919475894, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.38690185546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.833441734313965, + "learning_rate": 1e-06, + "loss": 0.3908, + "mean_token_accuracy": 0.8701074123382568, + "num_tokens": 358397206.0, + "step": 9232 + }, + { + "epoch": 1.1745325022261799, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.38690185546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.818357467651367, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8969925045967102, + "num_tokens": 358432441.0, + "step": 9233 + }, + { + "epoch": 1.1746597125047704, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.38690185546875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 10.091209411621094, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8834192156791687, + "num_tokens": 358470606.0, + "step": 9234 + }, + { + "epoch": 1.174786922783361, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.38690185546875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.830787658691406, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8944135904312134, + "num_tokens": 358509220.0, + "step": 9235 + }, + { + "epoch": 1.1749141330619515, + "ewc_loss": 0.01397705078125, + "ewc_loss_diag": 4.38690185546875e-05, + "ewc_loss_parallel": 9.59634780883789e-06, + "grad_norm": 8.579082489013672, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8933799266815186, + "num_tokens": 358548765.0, + "step": 9236 + }, + { + "epoch": 1.175041343340542, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.38690185546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.92794418334961, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8746036887168884, + "num_tokens": 358585546.0, + "step": 9237 + }, + { + "epoch": 1.1751685536191325, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.38690185546875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.642218589782715, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8858257532119751, + "num_tokens": 358625887.0, + "step": 9238 + }, + { + "epoch": 1.175295763897723, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.38690185546875e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.845701217651367, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8860138058662415, + "num_tokens": 358661078.0, + "step": 9239 + }, + { + "epoch": 1.1754229741763134, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.863141059875488, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8859313130378723, + "num_tokens": 358697547.0, + "step": 9240 + }, + { + "epoch": 1.175550184454904, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.68337631225586, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.9036109447479248, + "num_tokens": 358732500.0, + "step": 9241 + }, + { + "epoch": 1.1756773947334944, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.682098388671875, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8904997110366821, + "num_tokens": 358776925.0, + "step": 9242 + }, + { + "epoch": 1.175804605012085, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.899739265441895, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8795082569122314, + "num_tokens": 358811137.0, + "step": 9243 + }, + { + "epoch": 1.1759318152906755, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.64327335357666, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8976464867591858, + "num_tokens": 358852629.0, + "step": 9244 + }, + { + "epoch": 1.176059025569266, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.730255126953125, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8836575746536255, + "num_tokens": 358892966.0, + "step": 9245 + }, + { + "epoch": 1.1761862358478565, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.798033714294434, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.896979033946991, + "num_tokens": 358924897.0, + "step": 9246 + }, + { + "epoch": 1.176313446126447, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.738574981689453, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8740041255950928, + "num_tokens": 358968335.0, + "step": 9247 + }, + { + "epoch": 1.1764406564050376, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.797829627990723, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8914839625358582, + "num_tokens": 359008155.0, + "step": 9248 + }, + { + "epoch": 1.1765678666836281, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.724862098693848, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8949034214019775, + "num_tokens": 359048470.0, + "step": 9249 + }, + { + "epoch": 1.1766950769622184, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.75304889678955, + "learning_rate": 1e-06, + "loss": 0.4016, + "mean_token_accuracy": 0.8689029216766357, + "num_tokens": 359091658.0, + "step": 9250 + }, + { + "epoch": 1.176822287240809, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.70171070098877, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8920634984970093, + "num_tokens": 359137917.0, + "step": 9251 + }, + { + "epoch": 1.1769494975193995, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.788040161132812, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8845742344856262, + "num_tokens": 359172732.0, + "step": 9252 + }, + { + "epoch": 1.17707670779799, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.746274948120117, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8786230683326721, + "num_tokens": 359218238.0, + "step": 9253 + }, + { + "epoch": 1.1772039180765805, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.79176139831543, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8848313093185425, + "num_tokens": 359252895.0, + "step": 9254 + }, + { + "epoch": 1.177331128355171, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.820570945739746, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8930357098579407, + "num_tokens": 359286612.0, + "step": 9255 + }, + { + "epoch": 1.1774583386337616, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.728450775146484, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8887792825698853, + "num_tokens": 359328011.0, + "step": 9256 + }, + { + "epoch": 1.1775855489123521, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.744824409484863, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8830505013465881, + "num_tokens": 359368772.0, + "step": 9257 + }, + { + "epoch": 1.1777127591909426, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.812443733215332, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8917866349220276, + "num_tokens": 359403102.0, + "step": 9258 + }, + { + "epoch": 1.1778399694695332, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.652328491210938, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8747531175613403, + "num_tokens": 359450071.0, + "step": 9259 + }, + { + "epoch": 1.1779671797481237, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.740392684936523, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.888538122177124, + "num_tokens": 359487197.0, + "step": 9260 + }, + { + "epoch": 1.1780943900267142, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.747533798217773, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8913559913635254, + "num_tokens": 359527944.0, + "step": 9261 + }, + { + "epoch": 1.1782216003053048, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.70816707611084, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8790816068649292, + "num_tokens": 359568326.0, + "step": 9262 + }, + { + "epoch": 1.1783488105838953, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.734138488769531, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8841092586517334, + "num_tokens": 359611930.0, + "step": 9263 + }, + { + "epoch": 1.1784760208624856, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.797667503356934, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8900505900382996, + "num_tokens": 359648591.0, + "step": 9264 + }, + { + "epoch": 1.1786032311410761, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.71601676940918, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8918025493621826, + "num_tokens": 359687439.0, + "step": 9265 + }, + { + "epoch": 1.1787304414196667, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.735490798950195, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8855875730514526, + "num_tokens": 359728296.0, + "step": 9266 + }, + { + "epoch": 1.1788576516982572, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.634259223937988, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8955034613609314, + "num_tokens": 359770298.0, + "step": 9267 + }, + { + "epoch": 1.1789848619768477, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.711861610412598, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8892945051193237, + "num_tokens": 359810808.0, + "step": 9268 + }, + { + "epoch": 1.1791120722554382, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.690356254577637, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.896897554397583, + "num_tokens": 359857022.0, + "step": 9269 + }, + { + "epoch": 1.1792392825340288, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.723587036132812, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8741204738616943, + "num_tokens": 359898309.0, + "step": 9270 + }, + { + "epoch": 1.1793664928126193, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.776183128356934, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8789021968841553, + "num_tokens": 359942909.0, + "step": 9271 + }, + { + "epoch": 1.1794937030912098, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.760682106018066, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8952381610870361, + "num_tokens": 359978616.0, + "step": 9272 + }, + { + "epoch": 1.1796209133698004, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.745245933532715, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8954935669898987, + "num_tokens": 360013543.0, + "step": 9273 + }, + { + "epoch": 1.1797481236483909, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.7506685256958, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8950228691101074, + "num_tokens": 360049272.0, + "step": 9274 + }, + { + "epoch": 1.1798753339269812, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.79360580444336, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8741788864135742, + "num_tokens": 360088979.0, + "step": 9275 + }, + { + "epoch": 1.1800025442055717, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.856193542480469, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.8701838850975037, + "num_tokens": 360124691.0, + "step": 9276 + }, + { + "epoch": 1.1801297544841622, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.805322647094727, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8768778443336487, + "num_tokens": 360162885.0, + "step": 9277 + }, + { + "epoch": 1.1802569647627528, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.791987419128418, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8717941045761108, + "num_tokens": 360201950.0, + "step": 9278 + }, + { + "epoch": 1.1803841750413433, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.708537101745605, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8869919180870056, + "num_tokens": 360241556.0, + "step": 9279 + }, + { + "epoch": 1.1805113853199338, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.77527904510498, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8842208981513977, + "num_tokens": 360278749.0, + "step": 9280 + }, + { + "epoch": 1.1806385955985244, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.768471717834473, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8972480297088623, + "num_tokens": 360313467.0, + "step": 9281 + }, + { + "epoch": 1.1807658058771149, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.701507568359375, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8829999566078186, + "num_tokens": 360354915.0, + "step": 9282 + }, + { + "epoch": 1.1808930161557054, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.79544734954834, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8722827434539795, + "num_tokens": 360392534.0, + "step": 9283 + }, + { + "epoch": 1.181020226434296, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.843932151794434, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.8695302605628967, + "num_tokens": 360426160.0, + "step": 9284 + }, + { + "epoch": 1.1811474367128865, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.769439697265625, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8779771327972412, + "num_tokens": 360464460.0, + "step": 9285 + }, + { + "epoch": 1.181274646991477, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.752609252929688, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8823084831237793, + "num_tokens": 360503538.0, + "step": 9286 + }, + { + "epoch": 1.1814018572700675, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.638116836547852, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8937716484069824, + "num_tokens": 360547643.0, + "step": 9287 + }, + { + "epoch": 1.181529067548658, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.737615585327148, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8793366551399231, + "num_tokens": 360588496.0, + "step": 9288 + }, + { + "epoch": 1.1816562778272484, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.752941131591797, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8839727640151978, + "num_tokens": 360629399.0, + "step": 9289 + }, + { + "epoch": 1.1817834881058389, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.711209297180176, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8828250765800476, + "num_tokens": 360673224.0, + "step": 9290 + }, + { + "epoch": 1.1819106983844294, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.756938934326172, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8876574039459229, + "num_tokens": 360712102.0, + "step": 9291 + }, + { + "epoch": 1.18203790866302, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.813499450683594, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8721060752868652, + "num_tokens": 360751369.0, + "step": 9292 + }, + { + "epoch": 1.1821651189416105, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.677539825439453, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9049407243728638, + "num_tokens": 360790388.0, + "step": 9293 + }, + { + "epoch": 1.182292329220201, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.720589637756348, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8889414072036743, + "num_tokens": 360833439.0, + "step": 9294 + }, + { + "epoch": 1.1824195394987915, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.757349014282227, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8932764530181885, + "num_tokens": 360871799.0, + "step": 9295 + }, + { + "epoch": 1.182546749777382, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.711121559143066, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8904553055763245, + "num_tokens": 360911936.0, + "step": 9296 + }, + { + "epoch": 1.1826739600559726, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.70324420928955, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8869187235832214, + "num_tokens": 360960516.0, + "step": 9297 + }, + { + "epoch": 1.1828011703345631, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.822403907775879, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8918519616127014, + "num_tokens": 360992647.0, + "step": 9298 + }, + { + "epoch": 1.1829283806131534, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.704166412353516, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8873854875564575, + "num_tokens": 361036227.0, + "step": 9299 + }, + { + "epoch": 1.183055590891744, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.780485153198242, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.8693736791610718, + "num_tokens": 361074160.0, + "step": 9300 + }, + { + "epoch": 1.1831828011703345, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.874436378479004, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8868148922920227, + "num_tokens": 361109628.0, + "step": 9301 + }, + { + "epoch": 1.183310011448925, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.642895698547363, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8847676515579224, + "num_tokens": 361154041.0, + "step": 9302 + }, + { + "epoch": 1.1834372217275155, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.739144325256348, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8732181787490845, + "num_tokens": 361197416.0, + "step": 9303 + }, + { + "epoch": 1.183564432006106, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.86484146118164, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8724490404129028, + "num_tokens": 361233346.0, + "step": 9304 + }, + { + "epoch": 1.1836916422846966, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.770819664001465, + "learning_rate": 1e-06, + "loss": 0.395, + "mean_token_accuracy": 0.865679919719696, + "num_tokens": 361273857.0, + "step": 9305 + }, + { + "epoch": 1.1838188525632871, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.749091148376465, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8860071897506714, + "num_tokens": 361313476.0, + "step": 9306 + }, + { + "epoch": 1.1839460628418776, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.660184860229492, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9073094129562378, + "num_tokens": 361349176.0, + "step": 9307 + }, + { + "epoch": 1.1840732731204682, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.74476146697998, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8762779235839844, + "num_tokens": 361391692.0, + "step": 9308 + }, + { + "epoch": 1.1842004833990587, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.722122192382812, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8835405111312866, + "num_tokens": 361439403.0, + "step": 9309 + }, + { + "epoch": 1.1843276936776492, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.81749153137207, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8789018392562866, + "num_tokens": 361476904.0, + "step": 9310 + }, + { + "epoch": 1.1844549039562398, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.746479034423828, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8966295719146729, + "num_tokens": 361520505.0, + "step": 9311 + }, + { + "epoch": 1.1845821142348303, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.718522071838379, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8869690895080566, + "num_tokens": 361562679.0, + "step": 9312 + }, + { + "epoch": 1.1847093245134206, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.76019287109375, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8932470679283142, + "num_tokens": 361604101.0, + "step": 9313 + }, + { + "epoch": 1.1848365347920111, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.738818168640137, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8776353597640991, + "num_tokens": 361647466.0, + "step": 9314 + }, + { + "epoch": 1.1849637450706016, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.780924797058105, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.9000431895256042, + "num_tokens": 361683743.0, + "step": 9315 + }, + { + "epoch": 1.1850909553491922, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.726587295532227, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8852123618125916, + "num_tokens": 361727327.0, + "step": 9316 + }, + { + "epoch": 1.1852181656277827, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.782914161682129, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8873396515846252, + "num_tokens": 361765781.0, + "step": 9317 + }, + { + "epoch": 1.1853453759063732, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.783012390136719, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8863235116004944, + "num_tokens": 361804272.0, + "step": 9318 + }, + { + "epoch": 1.1854725861849638, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.706931114196777, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8938760757446289, + "num_tokens": 361848465.0, + "step": 9319 + }, + { + "epoch": 1.1855997964635543, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.807168960571289, + "learning_rate": 1e-06, + "loss": 0.3947, + "mean_token_accuracy": 0.8669982552528381, + "num_tokens": 361890868.0, + "step": 9320 + }, + { + "epoch": 1.1857270067421448, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.773881912231445, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8952410221099854, + "num_tokens": 361924344.0, + "step": 9321 + }, + { + "epoch": 1.1858542170207353, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.73179817199707, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.890534520149231, + "num_tokens": 361964469.0, + "step": 9322 + }, + { + "epoch": 1.1859814272993259, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.765504837036133, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8781322240829468, + "num_tokens": 362006903.0, + "step": 9323 + }, + { + "epoch": 1.1861086375779162, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.719267845153809, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8851667046546936, + "num_tokens": 362046788.0, + "step": 9324 + }, + { + "epoch": 1.1862358478565067, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.723078727722168, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8908647298812866, + "num_tokens": 362086932.0, + "step": 9325 + }, + { + "epoch": 1.1863630581350972, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.708948135375977, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9025640487670898, + "num_tokens": 362127904.0, + "step": 9326 + }, + { + "epoch": 1.1864902684136878, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.751187324523926, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8794620037078857, + "num_tokens": 362165897.0, + "step": 9327 + }, + { + "epoch": 1.1866174786922783, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.737656593322754, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8862576484680176, + "num_tokens": 362204649.0, + "step": 9328 + }, + { + "epoch": 1.1867446889708688, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.704896926879883, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8905071020126343, + "num_tokens": 362244852.0, + "step": 9329 + }, + { + "epoch": 1.1868718992494594, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.797372817993164, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8892949819564819, + "num_tokens": 362284498.0, + "step": 9330 + }, + { + "epoch": 1.1869991095280499, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.702539443969727, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8965895771980286, + "num_tokens": 362325640.0, + "step": 9331 + }, + { + "epoch": 1.1871263198066404, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.410743713378906e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.811612129211426, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8973855972290039, + "num_tokens": 362362158.0, + "step": 9332 + }, + { + "epoch": 1.187253530085231, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.743202209472656, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8955477476119995, + "num_tokens": 362402174.0, + "step": 9333 + }, + { + "epoch": 1.1873807403638215, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.872457504272461, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8689173460006714, + "num_tokens": 362439555.0, + "step": 9334 + }, + { + "epoch": 1.187507950642412, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.861348152160645, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8818836212158203, + "num_tokens": 362473457.0, + "step": 9335 + }, + { + "epoch": 1.1876351609210025, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.810775756835938, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8926924467086792, + "num_tokens": 362510588.0, + "step": 9336 + }, + { + "epoch": 1.187762371199593, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.6861572265625, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8856906890869141, + "num_tokens": 362551530.0, + "step": 9337 + }, + { + "epoch": 1.1878895814781834, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.797863006591797, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8951059579849243, + "num_tokens": 362589587.0, + "step": 9338 + }, + { + "epoch": 1.1880167917567739, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.733049392700195, + "learning_rate": 1e-06, + "loss": 0.4022, + "mean_token_accuracy": 0.8644285798072815, + "num_tokens": 362638080.0, + "step": 9339 + }, + { + "epoch": 1.1881440020353644, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.706743240356445, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8819347620010376, + "num_tokens": 362684625.0, + "step": 9340 + }, + { + "epoch": 1.188271212313955, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.830815315246582, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8763150572776794, + "num_tokens": 362723466.0, + "step": 9341 + }, + { + "epoch": 1.1883984225925455, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.713007926940918, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8934857249259949, + "num_tokens": 362762396.0, + "step": 9342 + }, + { + "epoch": 1.188525632871136, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.846282958984375, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8900101780891418, + "num_tokens": 362798359.0, + "step": 9343 + }, + { + "epoch": 1.1886528431497265, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.778182029724121, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8920949697494507, + "num_tokens": 362835365.0, + "step": 9344 + }, + { + "epoch": 1.188780053428317, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.8256254196167, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8816577792167664, + "num_tokens": 362868395.0, + "step": 9345 + }, + { + "epoch": 1.1889072637069076, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.751092910766602, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.8659262657165527, + "num_tokens": 362915639.0, + "step": 9346 + }, + { + "epoch": 1.189034473985498, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.63888931274414, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8847013711929321, + "num_tokens": 362957849.0, + "step": 9347 + }, + { + "epoch": 1.1891616842640884, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.744861602783203, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8996676802635193, + "num_tokens": 362994948.0, + "step": 9348 + }, + { + "epoch": 1.189288894542679, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.70968246459961, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8847795724868774, + "num_tokens": 363038864.0, + "step": 9349 + }, + { + "epoch": 1.1894161048212695, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.815438270568848, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8819636106491089, + "num_tokens": 363073357.0, + "step": 9350 + }, + { + "epoch": 1.18954331509986, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.754384994506836, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8810807466506958, + "num_tokens": 363113582.0, + "step": 9351 + }, + { + "epoch": 1.1896705253784505, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.791003227233887, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8838257193565369, + "num_tokens": 363151217.0, + "step": 9352 + }, + { + "epoch": 1.189797735657041, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.685585021972656, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8945375680923462, + "num_tokens": 363190933.0, + "step": 9353 + }, + { + "epoch": 1.1899249459356316, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.75786304473877, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8970761299133301, + "num_tokens": 363227454.0, + "step": 9354 + }, + { + "epoch": 1.1900521562142221, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.762516975402832, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8898031115531921, + "num_tokens": 363267729.0, + "step": 9355 + }, + { + "epoch": 1.1901793664928126, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.764679908752441, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8876006603240967, + "num_tokens": 363309687.0, + "step": 9356 + }, + { + "epoch": 1.1903065767714032, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.761687278747559, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8870501518249512, + "num_tokens": 363348046.0, + "step": 9357 + }, + { + "epoch": 1.1904337870499937, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.81483268737793, + "learning_rate": 1e-06, + "loss": 0.3838, + "mean_token_accuracy": 0.8700793981552124, + "num_tokens": 363386343.0, + "step": 9358 + }, + { + "epoch": 1.1905609973285842, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.729279518127441, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9024239182472229, + "num_tokens": 363426395.0, + "step": 9359 + }, + { + "epoch": 1.1906882076071748, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.709871292114258, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8937076926231384, + "num_tokens": 363468017.0, + "step": 9360 + }, + { + "epoch": 1.1908154178857653, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.78844928741455, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8980348110198975, + "num_tokens": 363506399.0, + "step": 9361 + }, + { + "epoch": 1.1909426281643556, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.787826538085938, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8910346031188965, + "num_tokens": 363542455.0, + "step": 9362 + }, + { + "epoch": 1.1910698384429461, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.717554092407227, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8929152488708496, + "num_tokens": 363581802.0, + "step": 9363 + }, + { + "epoch": 1.1911970487215366, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.788369178771973, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.9000119566917419, + "num_tokens": 363615201.0, + "step": 9364 + }, + { + "epoch": 1.1913242590001272, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.82553482055664, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8823246955871582, + "num_tokens": 363649761.0, + "step": 9365 + }, + { + "epoch": 1.1914514692787177, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.680502891540527, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8874788284301758, + "num_tokens": 363696267.0, + "step": 9366 + }, + { + "epoch": 1.1915786795573082, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.859856605529785, + "learning_rate": 1e-06, + "loss": 0.3849, + "mean_token_accuracy": 0.8710423707962036, + "num_tokens": 363733152.0, + "step": 9367 + }, + { + "epoch": 1.1917058898358988, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.771688461303711, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8877023458480835, + "num_tokens": 363776834.0, + "step": 9368 + }, + { + "epoch": 1.1918331001144893, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.82405948638916, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8775743246078491, + "num_tokens": 363813445.0, + "step": 9369 + }, + { + "epoch": 1.1919603103930798, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.919522285461426, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8801406025886536, + "num_tokens": 363844364.0, + "step": 9370 + }, + { + "epoch": 1.1920875206716703, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.807684898376465, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8846235275268555, + "num_tokens": 363880226.0, + "step": 9371 + }, + { + "epoch": 1.1922147309502609, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.708974838256836, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8954817056655884, + "num_tokens": 363915423.0, + "step": 9372 + }, + { + "epoch": 1.1923419412288512, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.797469139099121, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.89415442943573, + "num_tokens": 363952040.0, + "step": 9373 + }, + { + "epoch": 1.1924691515074417, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.822150230407715, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8934193849563599, + "num_tokens": 363984800.0, + "step": 9374 + }, + { + "epoch": 1.1925963617860322, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.778373718261719, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8798325061798096, + "num_tokens": 364027315.0, + "step": 9375 + }, + { + "epoch": 1.1927235720646228, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.809219360351562, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8963794708251953, + "num_tokens": 364061040.0, + "step": 9376 + }, + { + "epoch": 1.1928507823432133, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.752565383911133, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8904027938842773, + "num_tokens": 364097646.0, + "step": 9377 + }, + { + "epoch": 1.1929779926218038, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.719705581665039, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8971238136291504, + "num_tokens": 364142279.0, + "step": 9378 + }, + { + "epoch": 1.1931052029003943, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.771858215332031, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8776482939720154, + "num_tokens": 364181943.0, + "step": 9379 + }, + { + "epoch": 1.1932324131789849, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.765144348144531, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8817833065986633, + "num_tokens": 364222370.0, + "step": 9380 + }, + { + "epoch": 1.1933596234575754, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.787303924560547, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8817453384399414, + "num_tokens": 364264127.0, + "step": 9381 + }, + { + "epoch": 1.193486833736166, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.760810852050781, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8832329511642456, + "num_tokens": 364302110.0, + "step": 9382 + }, + { + "epoch": 1.1936140440147565, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.778154373168945, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8830662965774536, + "num_tokens": 364340858.0, + "step": 9383 + }, + { + "epoch": 1.193741254293347, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.918810844421387, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8975095748901367, + "num_tokens": 364381799.0, + "step": 9384 + }, + { + "epoch": 1.1938684645719375, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.73717975616455, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8810555934906006, + "num_tokens": 364424652.0, + "step": 9385 + }, + { + "epoch": 1.193995674850528, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.782920837402344, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8929104208946228, + "num_tokens": 364459710.0, + "step": 9386 + }, + { + "epoch": 1.1941228851291183, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.812880516052246, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8925291299819946, + "num_tokens": 364495571.0, + "step": 9387 + }, + { + "epoch": 1.1942500954077089, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.922733306884766, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8918092250823975, + "num_tokens": 364532708.0, + "step": 9388 + }, + { + "epoch": 1.1943773056862994, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.750393867492676, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8819859027862549, + "num_tokens": 364569466.0, + "step": 9389 + }, + { + "epoch": 1.19450451596489, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.606659889221191, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8973608016967773, + "num_tokens": 364613153.0, + "step": 9390 + }, + { + "epoch": 1.1946317262434805, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.685456275939941, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8877695202827454, + "num_tokens": 364654113.0, + "step": 9391 + }, + { + "epoch": 1.194758936522071, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.667956352233887, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.9032063484191895, + "num_tokens": 364696126.0, + "step": 9392 + }, + { + "epoch": 1.1948861468006615, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.851356506347656, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8838554620742798, + "num_tokens": 364726983.0, + "step": 9393 + }, + { + "epoch": 1.195013357079252, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.869733810424805, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8742620944976807, + "num_tokens": 364765498.0, + "step": 9394 + }, + { + "epoch": 1.1951405673578426, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.4345855712890625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.641684532165527, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8973102569580078, + "num_tokens": 364811642.0, + "step": 9395 + }, + { + "epoch": 1.195267777636433, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.670408248901367, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.9070405960083008, + "num_tokens": 364850223.0, + "step": 9396 + }, + { + "epoch": 1.1953949879150234, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.720335006713867, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8968205451965332, + "num_tokens": 364890273.0, + "step": 9397 + }, + { + "epoch": 1.195522198193614, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.810216903686523, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.884567141532898, + "num_tokens": 364925675.0, + "step": 9398 + }, + { + "epoch": 1.1956494084722045, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.724455833435059, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8941208124160767, + "num_tokens": 364966493.0, + "step": 9399 + }, + { + "epoch": 1.195776618750795, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.747784614562988, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8909980654716492, + "num_tokens": 365004349.0, + "step": 9400 + }, + { + "epoch": 1.1959038290293855, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.812108993530273, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8743798732757568, + "num_tokens": 365044104.0, + "step": 9401 + }, + { + "epoch": 1.196031039307976, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.803256034851074, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.884211540222168, + "num_tokens": 365078227.0, + "step": 9402 + }, + { + "epoch": 1.1961582495865666, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.830648422241211, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8851178884506226, + "num_tokens": 365116620.0, + "step": 9403 + }, + { + "epoch": 1.196285459865157, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.855362892150879, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8936986923217773, + "num_tokens": 365150079.0, + "step": 9404 + }, + { + "epoch": 1.1964126701437476, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.713047981262207, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8891746401786804, + "num_tokens": 365189652.0, + "step": 9405 + }, + { + "epoch": 1.1965398804223382, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.798812866210938, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8940802812576294, + "num_tokens": 365228516.0, + "step": 9406 + }, + { + "epoch": 1.1966670907009287, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.79619312286377, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8974601030349731, + "num_tokens": 365265931.0, + "step": 9407 + }, + { + "epoch": 1.1967943009795192, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.708723068237305, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.9029240608215332, + "num_tokens": 365306655.0, + "step": 9408 + }, + { + "epoch": 1.1969215112581097, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.783517837524414, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8799853920936584, + "num_tokens": 365346419.0, + "step": 9409 + }, + { + "epoch": 1.1970487215367003, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.779963493347168, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.890800416469574, + "num_tokens": 365384884.0, + "step": 9410 + }, + { + "epoch": 1.1971759318152906, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.780569076538086, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8977822065353394, + "num_tokens": 365419582.0, + "step": 9411 + }, + { + "epoch": 1.1973031420938811, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.685247421264648, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8856633901596069, + "num_tokens": 365457815.0, + "step": 9412 + }, + { + "epoch": 1.1974303523724716, + "ewc_loss": 0.01416015625, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.715557098388672e-06, + "grad_norm": 8.82931900024414, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8804475665092468, + "num_tokens": 365493259.0, + "step": 9413 + }, + { + "epoch": 1.1975575626510622, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.71078872680664, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8836889266967773, + "num_tokens": 365538299.0, + "step": 9414 + }, + { + "epoch": 1.1976847729296527, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.817326545715332, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8902902603149414, + "num_tokens": 365572654.0, + "step": 9415 + }, + { + "epoch": 1.1978119832082432, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.745139122009277, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8948109149932861, + "num_tokens": 365611553.0, + "step": 9416 + }, + { + "epoch": 1.1979391934868338, + "ewc_loss": 0.01422119140625, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.775161743164062e-06, + "grad_norm": 8.773143768310547, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8956067562103271, + "num_tokens": 365647548.0, + "step": 9417 + }, + { + "epoch": 1.1980664037654243, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.810277938842773, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8943722248077393, + "num_tokens": 365681978.0, + "step": 9418 + }, + { + "epoch": 1.1981936140440148, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.807994842529297, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8938420414924622, + "num_tokens": 365718663.0, + "step": 9419 + }, + { + "epoch": 1.1983208243226053, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.760852813720703, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8690119981765747, + "num_tokens": 365761500.0, + "step": 9420 + }, + { + "epoch": 1.1984480346011959, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.97219181060791, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8888779282569885, + "num_tokens": 365800673.0, + "step": 9421 + }, + { + "epoch": 1.1985752448797862, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.772672653198242, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8995979428291321, + "num_tokens": 365836416.0, + "step": 9422 + }, + { + "epoch": 1.1987024551583767, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.458427429199219e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.784504890441895, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8846140503883362, + "num_tokens": 365875628.0, + "step": 9423 + }, + { + "epoch": 1.1988296654369672, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.780633926391602, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8803813457489014, + "num_tokens": 365918455.0, + "step": 9424 + }, + { + "epoch": 1.1989568757155578, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.860520362854004, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8870049118995667, + "num_tokens": 365954444.0, + "step": 9425 + }, + { + "epoch": 1.1990840859941483, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.860451698303223, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8923171162605286, + "num_tokens": 365989733.0, + "step": 9426 + }, + { + "epoch": 1.1992112962727388, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.87856674194336, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8917980194091797, + "num_tokens": 366026177.0, + "step": 9427 + }, + { + "epoch": 1.1993385065513293, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.738481521606445, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8923828601837158, + "num_tokens": 366065123.0, + "step": 9428 + }, + { + "epoch": 1.1994657168299199, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.837767601013184, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8856555819511414, + "num_tokens": 366104736.0, + "step": 9429 + }, + { + "epoch": 1.1995929271085104, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.786470413208008, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8854042887687683, + "num_tokens": 366141404.0, + "step": 9430 + }, + { + "epoch": 1.199720137387101, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.764386177062988, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8854855298995972, + "num_tokens": 366179798.0, + "step": 9431 + }, + { + "epoch": 1.1998473476656915, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.777952194213867, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8888118863105774, + "num_tokens": 366225937.0, + "step": 9432 + }, + { + "epoch": 1.199974557944282, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.889911651611328, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8893711566925049, + "num_tokens": 366261231.0, + "step": 9433 + }, + { + "epoch": 1.2001017682228725, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.76081371307373, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.9024347066879272, + "num_tokens": 366295718.0, + "step": 9434 + }, + { + "epoch": 1.200228978501463, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.765105247497559, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8995949625968933, + "num_tokens": 366333898.0, + "step": 9435 + }, + { + "epoch": 1.2003561887800533, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.83288860321045, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8933694362640381, + "num_tokens": 366371300.0, + "step": 9436 + }, + { + "epoch": 1.2004833990586439, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.71358871459961, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8912061452865601, + "num_tokens": 366411180.0, + "step": 9437 + }, + { + "epoch": 1.2006106093372344, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.784751892089844, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.897980272769928, + "num_tokens": 366448168.0, + "step": 9438 + }, + { + "epoch": 1.200737819615825, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.805702209472656, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8848751783370972, + "num_tokens": 366492323.0, + "step": 9439 + }, + { + "epoch": 1.2008650298944155, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.80482006072998, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.8994443416595459, + "num_tokens": 366525763.0, + "step": 9440 + }, + { + "epoch": 1.200992240173006, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.785222053527832, + "learning_rate": 1e-06, + "loss": 0.392, + "mean_token_accuracy": 0.8677623271942139, + "num_tokens": 366569021.0, + "step": 9441 + }, + { + "epoch": 1.2011194504515965, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.727834701538086, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8812498450279236, + "num_tokens": 366611935.0, + "step": 9442 + }, + { + "epoch": 1.201246660730187, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.829977035522461, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8951402306556702, + "num_tokens": 366644910.0, + "step": 9443 + }, + { + "epoch": 1.2013738710087776, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.822237968444824, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8900867104530334, + "num_tokens": 366680557.0, + "step": 9444 + }, + { + "epoch": 1.201501081287368, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.805058479309082, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8880534768104553, + "num_tokens": 366719381.0, + "step": 9445 + }, + { + "epoch": 1.2016282915659584, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.761228561401367, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8935652375221252, + "num_tokens": 366758969.0, + "step": 9446 + }, + { + "epoch": 1.201755501844549, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.844487190246582, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8909640908241272, + "num_tokens": 366792494.0, + "step": 9447 + }, + { + "epoch": 1.2018827121231395, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.748748779296875, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8812046051025391, + "num_tokens": 366834680.0, + "step": 9448 + }, + { + "epoch": 1.20200992240173, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.811701774597168, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.88280189037323, + "num_tokens": 366871329.0, + "step": 9449 + }, + { + "epoch": 1.2021371326803205, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.732007026672363, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8907245397567749, + "num_tokens": 366908940.0, + "step": 9450 + }, + { + "epoch": 1.202264342958911, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.819491386413574, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8824569582939148, + "num_tokens": 366950021.0, + "step": 9451 + }, + { + "epoch": 1.2023915532375016, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.794145584106445, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8839685916900635, + "num_tokens": 366987033.0, + "step": 9452 + }, + { + "epoch": 1.202518763516092, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.770268440246582, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8807466626167297, + "num_tokens": 367028670.0, + "step": 9453 + }, + { + "epoch": 1.2026459737946826, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.749914169311523, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8810280561447144, + "num_tokens": 367071835.0, + "step": 9454 + }, + { + "epoch": 1.2027731840732732, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.772517204284668, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.867688775062561, + "num_tokens": 367113768.0, + "step": 9455 + }, + { + "epoch": 1.2029003943518637, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.745613098144531, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8876075148582458, + "num_tokens": 367154466.0, + "step": 9456 + }, + { + "epoch": 1.2030276046304542, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.762039184570312, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9011505842208862, + "num_tokens": 367193684.0, + "step": 9457 + }, + { + "epoch": 1.2031548149090447, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.776763916015625, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8800809979438782, + "num_tokens": 367236670.0, + "step": 9458 + }, + { + "epoch": 1.2032820251876353, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.78311824798584, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8839496374130249, + "num_tokens": 367276676.0, + "step": 9459 + }, + { + "epoch": 1.2034092354662256, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.823442459106445, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8868128061294556, + "num_tokens": 367311028.0, + "step": 9460 + }, + { + "epoch": 1.203536445744816, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.818338394165039, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8726726770401001, + "num_tokens": 367351168.0, + "step": 9461 + }, + { + "epoch": 1.2036636560234066, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.697142601013184, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8861790895462036, + "num_tokens": 367392791.0, + "step": 9462 + }, + { + "epoch": 1.2037908663019972, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.809711456298828, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8980491757392883, + "num_tokens": 367429479.0, + "step": 9463 + }, + { + "epoch": 1.2039180765805877, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.87281322479248, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.8778239488601685, + "num_tokens": 367466385.0, + "step": 9464 + }, + { + "epoch": 1.2040452868591782, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.777426719665527, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8822802901268005, + "num_tokens": 367504706.0, + "step": 9465 + }, + { + "epoch": 1.2041724971377687, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.656176567077637, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.899142861366272, + "num_tokens": 367542130.0, + "step": 9466 + }, + { + "epoch": 1.2042997074163593, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.811060905456543, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8847818374633789, + "num_tokens": 367581900.0, + "step": 9467 + }, + { + "epoch": 1.2044269176949498, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.772636413574219, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8799743056297302, + "num_tokens": 367621151.0, + "step": 9468 + }, + { + "epoch": 1.2045541279735403, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.816560745239258, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8812931776046753, + "num_tokens": 367659893.0, + "step": 9469 + }, + { + "epoch": 1.2046813382521309, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.668411254882812, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8924785852432251, + "num_tokens": 367703115.0, + "step": 9470 + }, + { + "epoch": 1.2048085485307212, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.82723331451416, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8849085569381714, + "num_tokens": 367742373.0, + "step": 9471 + }, + { + "epoch": 1.2049357588093117, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.814750671386719, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8684611320495605, + "num_tokens": 367782644.0, + "step": 9472 + }, + { + "epoch": 1.2050629690879022, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.902078628540039, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.880097508430481, + "num_tokens": 367820120.0, + "step": 9473 + }, + { + "epoch": 1.2051901793664928, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.740381240844727, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8920456171035767, + "num_tokens": 367855652.0, + "step": 9474 + }, + { + "epoch": 1.2053173896450833, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.783761978149414, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8799360990524292, + "num_tokens": 367895383.0, + "step": 9475 + }, + { + "epoch": 1.2054445999236738, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.755794525146484, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8906065821647644, + "num_tokens": 367931463.0, + "step": 9476 + }, + { + "epoch": 1.2055718102022643, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.817384719848633, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8978398442268372, + "num_tokens": 367968371.0, + "step": 9477 + }, + { + "epoch": 1.2056990204808549, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.65918254852295, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.9037169218063354, + "num_tokens": 368011149.0, + "step": 9478 + }, + { + "epoch": 1.2058262307594454, + "ewc_loss": 0.0142822265625, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.83983325958252, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8962641358375549, + "num_tokens": 368041876.0, + "step": 9479 + }, + { + "epoch": 1.205953441038036, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.711071014404297, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8938984274864197, + "num_tokens": 368085839.0, + "step": 9480 + }, + { + "epoch": 1.2060806513166265, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.772937774658203, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8918764591217041, + "num_tokens": 368128558.0, + "step": 9481 + }, + { + "epoch": 1.206207861595217, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.774148941040039, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8786784410476685, + "num_tokens": 368167426.0, + "step": 9482 + }, + { + "epoch": 1.2063350718738075, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.845807075500488, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8854635953903198, + "num_tokens": 368198055.0, + "step": 9483 + }, + { + "epoch": 1.206462282152398, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.765018463134766, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8852362632751465, + "num_tokens": 368241785.0, + "step": 9484 + }, + { + "epoch": 1.2065894924309883, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.822933197021484, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8840564489364624, + "num_tokens": 368279505.0, + "step": 9485 + }, + { + "epoch": 1.2067167027095789, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.793210983276367, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8927962183952332, + "num_tokens": 368322675.0, + "step": 9486 + }, + { + "epoch": 1.2068439129881694, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.701932907104492, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8956059217453003, + "num_tokens": 368363607.0, + "step": 9487 + }, + { + "epoch": 1.20697112326676, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.88114070892334, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8907634019851685, + "num_tokens": 368397542.0, + "step": 9488 + }, + { + "epoch": 1.2070983335453505, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.725059509277344, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8907343149185181, + "num_tokens": 368435496.0, + "step": 9489 + }, + { + "epoch": 1.207225543823941, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.755411148071289, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8924771547317505, + "num_tokens": 368475702.0, + "step": 9490 + }, + { + "epoch": 1.2073527541025315, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.931414604187012, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.880874514579773, + "num_tokens": 368507649.0, + "step": 9491 + }, + { + "epoch": 1.207479964381122, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.841314315795898, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8940990567207336, + "num_tokens": 368538238.0, + "step": 9492 + }, + { + "epoch": 1.2076071746597126, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.871358871459961, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8866633176803589, + "num_tokens": 368571999.0, + "step": 9493 + }, + { + "epoch": 1.207734384938303, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.765793800354004, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.889987587928772, + "num_tokens": 368609876.0, + "step": 9494 + }, + { + "epoch": 1.2078615952168934, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.758773803710938, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8946636915206909, + "num_tokens": 368645336.0, + "step": 9495 + }, + { + "epoch": 1.207988805495484, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.942220687866211, + "learning_rate": 1e-06, + "loss": 0.3671, + "mean_token_accuracy": 0.8753795027732849, + "num_tokens": 368676053.0, + "step": 9496 + }, + { + "epoch": 1.2081160157740745, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.818001747131348, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.894357442855835, + "num_tokens": 368711918.0, + "step": 9497 + }, + { + "epoch": 1.208243226052665, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.779610633850098, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9074711799621582, + "num_tokens": 368746919.0, + "step": 9498 + }, + { + "epoch": 1.2083704363312555, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.751867294311523, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.9001806974411011, + "num_tokens": 368782155.0, + "step": 9499 + }, + { + "epoch": 1.208497646609846, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.769190788269043, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8803810477256775, + "num_tokens": 368826664.0, + "step": 9500 + }, + { + "epoch": 1.2086248568884366, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.74245834350586, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8902748823165894, + "num_tokens": 368871120.0, + "step": 9501 + }, + { + "epoch": 1.208752067167027, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.482269287109375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.79322338104248, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8853943943977356, + "num_tokens": 368915014.0, + "step": 9502 + }, + { + "epoch": 1.2088792774456176, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.797245025634766, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9041668772697449, + "num_tokens": 368948702.0, + "step": 9503 + }, + { + "epoch": 1.2090064877242082, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.784948348999023, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8890972137451172, + "num_tokens": 368989303.0, + "step": 9504 + }, + { + "epoch": 1.2091336980027987, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.80237102508545, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8877482414245605, + "num_tokens": 369030277.0, + "step": 9505 + }, + { + "epoch": 1.2092609082813892, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.856854438781738, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8906682729721069, + "num_tokens": 369068296.0, + "step": 9506 + }, + { + "epoch": 1.2093881185599797, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.771809577941895, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8915263414382935, + "num_tokens": 369103965.0, + "step": 9507 + }, + { + "epoch": 1.2095153288385703, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.826367378234863, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8933460712432861, + "num_tokens": 369142882.0, + "step": 9508 + }, + { + "epoch": 1.2096425391171606, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.94027042388916, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8729674816131592, + "num_tokens": 369174019.0, + "step": 9509 + }, + { + "epoch": 1.209769749395751, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.817699432373047, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8872314691543579, + "num_tokens": 369211775.0, + "step": 9510 + }, + { + "epoch": 1.2098969596743416, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.801043510437012, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8949397802352905, + "num_tokens": 369247366.0, + "step": 9511 + }, + { + "epoch": 1.2100241699529322, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.752129554748535, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.9054170846939087, + "num_tokens": 369284078.0, + "step": 9512 + }, + { + "epoch": 1.2101513802315227, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.753902435302734, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8803979754447937, + "num_tokens": 369326675.0, + "step": 9513 + }, + { + "epoch": 1.2102785905101132, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.841900825500488, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.870104968547821, + "num_tokens": 369367373.0, + "step": 9514 + }, + { + "epoch": 1.2104058007887037, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.766961097717285, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.894585132598877, + "num_tokens": 369405684.0, + "step": 9515 + }, + { + "epoch": 1.2105330110672943, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.760323524475098, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.873370885848999, + "num_tokens": 369449674.0, + "step": 9516 + }, + { + "epoch": 1.2106602213458848, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.806400299072266, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8875472545623779, + "num_tokens": 369495929.0, + "step": 9517 + }, + { + "epoch": 1.2107874316244753, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.832242965698242, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8891888856887817, + "num_tokens": 369541946.0, + "step": 9518 + }, + { + "epoch": 1.2109146419030659, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.765435218811035, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8944705128669739, + "num_tokens": 369579565.0, + "step": 9519 + }, + { + "epoch": 1.2110418521816562, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.952119827270508, + "learning_rate": 1e-06, + "loss": 0.3902, + "mean_token_accuracy": 0.8690282702445984, + "num_tokens": 369611350.0, + "step": 9520 + }, + { + "epoch": 1.2111690624602467, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.723301887512207, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8877639770507812, + "num_tokens": 369651738.0, + "step": 9521 + }, + { + "epoch": 1.2112962727388372, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.803250312805176, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8766629695892334, + "num_tokens": 369693314.0, + "step": 9522 + }, + { + "epoch": 1.2114234830174277, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.818464279174805, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8908432126045227, + "num_tokens": 369732617.0, + "step": 9523 + }, + { + "epoch": 1.2115506932960183, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.735746383666992, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8876014947891235, + "num_tokens": 369775216.0, + "step": 9524 + }, + { + "epoch": 1.2116779035746088, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.807470321655273, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8899973630905151, + "num_tokens": 369815787.0, + "step": 9525 + }, + { + "epoch": 1.2118051138531993, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.747215270996094, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8998850584030151, + "num_tokens": 369854676.0, + "step": 9526 + }, + { + "epoch": 1.2119323241317899, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.954972267150879, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8750497102737427, + "num_tokens": 369888504.0, + "step": 9527 + }, + { + "epoch": 1.2120595344103804, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.755857467651367, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8846321702003479, + "num_tokens": 369928255.0, + "step": 9528 + }, + { + "epoch": 1.212186744688971, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.866064071655273, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8731042742729187, + "num_tokens": 369967132.0, + "step": 9529 + }, + { + "epoch": 1.2123139549675614, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.810992240905762, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.892327070236206, + "num_tokens": 370001714.0, + "step": 9530 + }, + { + "epoch": 1.212441165246152, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.786566734313965, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8888258934020996, + "num_tokens": 370043381.0, + "step": 9531 + }, + { + "epoch": 1.2125683755247425, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.72659683227539, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8955657482147217, + "num_tokens": 370084131.0, + "step": 9532 + }, + { + "epoch": 1.212695585803333, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.82123851776123, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8816667795181274, + "num_tokens": 370124662.0, + "step": 9533 + }, + { + "epoch": 1.2128227960819233, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.83784008026123, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8924022912979126, + "num_tokens": 370161566.0, + "step": 9534 + }, + { + "epoch": 1.2129500063605139, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.90658950805664, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8785291910171509, + "num_tokens": 370196133.0, + "step": 9535 + }, + { + "epoch": 1.2130772166391044, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.88994026184082, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8764848709106445, + "num_tokens": 370235236.0, + "step": 9536 + }, + { + "epoch": 1.213204426917695, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.768074989318848, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.9033187627792358, + "num_tokens": 370271851.0, + "step": 9537 + }, + { + "epoch": 1.2133316371962855, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.7430419921875, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8843857049942017, + "num_tokens": 370315068.0, + "step": 9538 + }, + { + "epoch": 1.213458847474876, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.785442352294922, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8847557902336121, + "num_tokens": 370356996.0, + "step": 9539 + }, + { + "epoch": 1.2135860577534665, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.851336479187012, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9014594554901123, + "num_tokens": 370387008.0, + "step": 9540 + }, + { + "epoch": 1.213713268032057, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.766129493713379, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8882608413696289, + "num_tokens": 370426765.0, + "step": 9541 + }, + { + "epoch": 1.2138404783106476, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.732444763183594, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8869186639785767, + "num_tokens": 370471357.0, + "step": 9542 + }, + { + "epoch": 1.213967688589238, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.78732967376709, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8878142833709717, + "num_tokens": 370508056.0, + "step": 9543 + }, + { + "epoch": 1.2140948988678284, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.708617210388184, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8892809152603149, + "num_tokens": 370552599.0, + "step": 9544 + }, + { + "epoch": 1.214222109146419, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.78903865814209, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.887148380279541, + "num_tokens": 370593293.0, + "step": 9545 + }, + { + "epoch": 1.2143493194250095, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.798876762390137, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8978297710418701, + "num_tokens": 370632498.0, + "step": 9546 + }, + { + "epoch": 1.2144765297036, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.721112251281738, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8882189989089966, + "num_tokens": 370675384.0, + "step": 9547 + }, + { + "epoch": 1.2146037399821905, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.750364303588867, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8920971155166626, + "num_tokens": 370718080.0, + "step": 9548 + }, + { + "epoch": 1.214730950260781, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.734710693359375, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8985821008682251, + "num_tokens": 370754201.0, + "step": 9549 + }, + { + "epoch": 1.2148581605393716, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.875986099243164, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8817703723907471, + "num_tokens": 370792778.0, + "step": 9550 + }, + { + "epoch": 1.214985370817962, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.756614685058594, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8919652700424194, + "num_tokens": 370830377.0, + "step": 9551 + }, + { + "epoch": 1.2151125810965526, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.883222579956055, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8805726766586304, + "num_tokens": 370862135.0, + "step": 9552 + }, + { + "epoch": 1.2152397913751432, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.756181716918945, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8868959546089172, + "num_tokens": 370897556.0, + "step": 9553 + }, + { + "epoch": 1.2153670016537337, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.80709457397461, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8914414644241333, + "num_tokens": 370936903.0, + "step": 9554 + }, + { + "epoch": 1.2154942119323242, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.679991722106934, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8933048248291016, + "num_tokens": 370979751.0, + "step": 9555 + }, + { + "epoch": 1.2156214222109147, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.869538307189941, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8976154327392578, + "num_tokens": 371016343.0, + "step": 9556 + }, + { + "epoch": 1.2157486324895053, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.772689819335938, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8888958096504211, + "num_tokens": 371055675.0, + "step": 9557 + }, + { + "epoch": 1.2158758427680956, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.506111145019531e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.853710174560547, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8838640451431274, + "num_tokens": 371092096.0, + "step": 9558 + }, + { + "epoch": 1.216003053046686, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.708282470703125, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8994483351707458, + "num_tokens": 371133193.0, + "step": 9559 + }, + { + "epoch": 1.2161302633252766, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.785663604736328, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8990456461906433, + "num_tokens": 371170367.0, + "step": 9560 + }, + { + "epoch": 1.2162574736038672, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.77493953704834, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.889401912689209, + "num_tokens": 371212059.0, + "step": 9561 + }, + { + "epoch": 1.2163846838824577, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.88337516784668, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8931890726089478, + "num_tokens": 371248188.0, + "step": 9562 + }, + { + "epoch": 1.2165118941610482, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.784992218017578, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8871088027954102, + "num_tokens": 371290081.0, + "step": 9563 + }, + { + "epoch": 1.2166391044396387, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.78287410736084, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8876048922538757, + "num_tokens": 371329081.0, + "step": 9564 + }, + { + "epoch": 1.2167663147182293, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.863641738891602, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8703329563140869, + "num_tokens": 371366122.0, + "step": 9565 + }, + { + "epoch": 1.2168935249968198, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.794825553894043, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8706119656562805, + "num_tokens": 371405958.0, + "step": 9566 + }, + { + "epoch": 1.2170207352754103, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.829566955566406, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.890364944934845, + "num_tokens": 371444868.0, + "step": 9567 + }, + { + "epoch": 1.2171479455540009, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.869315147399902, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8900575041770935, + "num_tokens": 371477976.0, + "step": 9568 + }, + { + "epoch": 1.2172751558325912, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.884456634521484, + "learning_rate": 1e-06, + "loss": 0.4081, + "mean_token_accuracy": 0.8607625961303711, + "num_tokens": 371516950.0, + "step": 9569 + }, + { + "epoch": 1.2174023661111817, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.715666770935059, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.890324592590332, + "num_tokens": 371561018.0, + "step": 9570 + }, + { + "epoch": 1.2175295763897722, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.951977729797363, + "learning_rate": 1e-06, + "loss": 0.3969, + "mean_token_accuracy": 0.8685411214828491, + "num_tokens": 371595145.0, + "step": 9571 + }, + { + "epoch": 1.2176567866683627, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.878178596496582, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8863017559051514, + "num_tokens": 371628426.0, + "step": 9572 + }, + { + "epoch": 1.2177839969469533, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.740884780883789, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8913587927818298, + "num_tokens": 371670991.0, + "step": 9573 + }, + { + "epoch": 1.2179112072255438, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.795296669006348, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8900797367095947, + "num_tokens": 371705959.0, + "step": 9574 + }, + { + "epoch": 1.2180384175041343, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 9.15256404876709, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8842403888702393, + "num_tokens": 371747903.0, + "step": 9575 + }, + { + "epoch": 1.2181656277827249, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.703107833862305, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8816762566566467, + "num_tokens": 371790310.0, + "step": 9576 + }, + { + "epoch": 1.2182928380613154, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.830957412719727, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8836895227432251, + "num_tokens": 371824404.0, + "step": 9577 + }, + { + "epoch": 1.218420048339906, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.798578262329102, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8916430473327637, + "num_tokens": 371866520.0, + "step": 9578 + }, + { + "epoch": 1.2185472586184964, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.723830223083496, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8948559761047363, + "num_tokens": 371902492.0, + "step": 9579 + }, + { + "epoch": 1.218674468897087, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.814380645751953, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8749160170555115, + "num_tokens": 371946009.0, + "step": 9580 + }, + { + "epoch": 1.2188016791756775, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.98758316040039, + "learning_rate": 1e-06, + "loss": 0.393, + "mean_token_accuracy": 0.8703258633613586, + "num_tokens": 371980899.0, + "step": 9581 + }, + { + "epoch": 1.218928889454268, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.768965721130371, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8968902826309204, + "num_tokens": 372021865.0, + "step": 9582 + }, + { + "epoch": 1.2190560997328583, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.760496139526367, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8896890878677368, + "num_tokens": 372063161.0, + "step": 9583 + }, + { + "epoch": 1.2191833100114489, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.69921588897705, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9015868306159973, + "num_tokens": 372105634.0, + "step": 9584 + }, + { + "epoch": 1.2193105202900394, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.876687049865723, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8935030698776245, + "num_tokens": 372140510.0, + "step": 9585 + }, + { + "epoch": 1.21943773056863, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.855164527893066, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8846641182899475, + "num_tokens": 372175123.0, + "step": 9586 + }, + { + "epoch": 1.2195649408472204, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.890931129455566, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8979746103286743, + "num_tokens": 372210279.0, + "step": 9587 + }, + { + "epoch": 1.219692151125811, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.82208251953125, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8871967792510986, + "num_tokens": 372246535.0, + "step": 9588 + }, + { + "epoch": 1.2198193614044015, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.809288024902344, + "learning_rate": 1e-06, + "loss": 0.4007, + "mean_token_accuracy": 0.8636504411697388, + "num_tokens": 372290159.0, + "step": 9589 + }, + { + "epoch": 1.219946571682992, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.90116024017334, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8797582983970642, + "num_tokens": 372325391.0, + "step": 9590 + }, + { + "epoch": 1.2200737819615826, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.737447738647461, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8786801695823669, + "num_tokens": 372371767.0, + "step": 9591 + }, + { + "epoch": 1.220200992240173, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.889827728271484, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8831477761268616, + "num_tokens": 372405567.0, + "step": 9592 + }, + { + "epoch": 1.2203282025187634, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.730688095092773, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8946599960327148, + "num_tokens": 372447926.0, + "step": 9593 + }, + { + "epoch": 1.220455412797354, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.825976371765137, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.894664466381073, + "num_tokens": 372487344.0, + "step": 9594 + }, + { + "epoch": 1.2205826230759445, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.804863929748535, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8858294486999512, + "num_tokens": 372529621.0, + "step": 9595 + }, + { + "epoch": 1.220709833354535, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.789005279541016, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8890653252601624, + "num_tokens": 372569289.0, + "step": 9596 + }, + { + "epoch": 1.2208370436331255, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.855642318725586, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8810266256332397, + "num_tokens": 372608825.0, + "step": 9597 + }, + { + "epoch": 1.220964253911716, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.742578506469727, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8736613392829895, + "num_tokens": 372651053.0, + "step": 9598 + }, + { + "epoch": 1.2210914641903066, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.816177368164062, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8805716037750244, + "num_tokens": 372694626.0, + "step": 9599 + }, + { + "epoch": 1.221218674468897, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.807287216186523, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8881149291992188, + "num_tokens": 372727725.0, + "step": 9600 + }, + { + "epoch": 1.2213458847474876, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.736055374145508, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.9003549218177795, + "num_tokens": 372768456.0, + "step": 9601 + }, + { + "epoch": 1.2214730950260781, + "ewc_loss": 0.01434326171875, + "ewc_loss_diag": 4.5299530029296875e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.73279857635498, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8841614723205566, + "num_tokens": 372813090.0, + "step": 9602 + }, + { + "epoch": 1.2216003053046687, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.731264114379883, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8934110403060913, + "num_tokens": 372854976.0, + "step": 9603 + }, + { + "epoch": 1.2217275155832592, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.792227745056152, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8974428176879883, + "num_tokens": 372891626.0, + "step": 9604 + }, + { + "epoch": 1.2218547258618497, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.839261054992676, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8896395564079285, + "num_tokens": 372928325.0, + "step": 9605 + }, + { + "epoch": 1.2219819361404403, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.699820518493652, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8915371894836426, + "num_tokens": 372972156.0, + "step": 9606 + }, + { + "epoch": 1.2221091464190306, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.88004207611084, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8904102444648743, + "num_tokens": 373006990.0, + "step": 9607 + }, + { + "epoch": 1.222236356697621, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.755230903625488, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8951207995414734, + "num_tokens": 373047379.0, + "step": 9608 + }, + { + "epoch": 1.2223635669762116, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.773964881896973, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8932851552963257, + "num_tokens": 373087137.0, + "step": 9609 + }, + { + "epoch": 1.2224907772548022, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.883201599121094, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.877005398273468, + "num_tokens": 373130105.0, + "step": 9610 + }, + { + "epoch": 1.2226179875333927, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.86196231842041, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8861786723136902, + "num_tokens": 373164223.0, + "step": 9611 + }, + { + "epoch": 1.2227451978119832, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.688077926635742, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8855452537536621, + "num_tokens": 373205223.0, + "step": 9612 + }, + { + "epoch": 1.2228724080905737, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.797942161560059, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8965859413146973, + "num_tokens": 373246672.0, + "step": 9613 + }, + { + "epoch": 1.2229996183691643, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.687389373779297, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8871427774429321, + "num_tokens": 373291831.0, + "step": 9614 + }, + { + "epoch": 1.2231268286477548, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.788678169250488, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.903148889541626, + "num_tokens": 373332598.0, + "step": 9615 + }, + { + "epoch": 1.2232540389263453, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.786865234375, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.89339280128479, + "num_tokens": 373370728.0, + "step": 9616 + }, + { + "epoch": 1.2233812492049359, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.817211151123047, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8893214464187622, + "num_tokens": 373405160.0, + "step": 9617 + }, + { + "epoch": 1.2235084594835262, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.775667190551758, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8769033551216125, + "num_tokens": 373442952.0, + "step": 9618 + }, + { + "epoch": 1.2236356697621167, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.820417404174805, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8913434743881226, + "num_tokens": 373484530.0, + "step": 9619 + }, + { + "epoch": 1.2237628800407072, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.741278648376465, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8831897974014282, + "num_tokens": 373528384.0, + "step": 9620 + }, + { + "epoch": 1.2238900903192977, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.943119049072266, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8835821151733398, + "num_tokens": 373560955.0, + "step": 9621 + }, + { + "epoch": 1.2240173005978883, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.803025245666504, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8997652530670166, + "num_tokens": 373597240.0, + "step": 9622 + }, + { + "epoch": 1.2241445108764788, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.802116394042969, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.882339596748352, + "num_tokens": 373635586.0, + "step": 9623 + }, + { + "epoch": 1.2242717211550693, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.791604995727539, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8872439861297607, + "num_tokens": 373683193.0, + "step": 9624 + }, + { + "epoch": 1.2243989314336599, + "ewc_loss": 0.014404296875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.774001121520996, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8958463668823242, + "num_tokens": 373715281.0, + "step": 9625 + }, + { + "epoch": 1.2245261417122504, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.553794860839844e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.836960792541504, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8866128921508789, + "num_tokens": 373753647.0, + "step": 9626 + }, + { + "epoch": 1.224653351990841, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.759929656982422, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8758105635643005, + "num_tokens": 373794911.0, + "step": 9627 + }, + { + "epoch": 1.2247805622694314, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.845486640930176, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8823156952857971, + "num_tokens": 373836457.0, + "step": 9628 + }, + { + "epoch": 1.224907772548022, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.947185516357422, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8821260333061218, + "num_tokens": 373867444.0, + "step": 9629 + }, + { + "epoch": 1.2250349828266125, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.795441627502441, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8898714780807495, + "num_tokens": 373905250.0, + "step": 9630 + }, + { + "epoch": 1.225162193105203, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.802464485168457, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8776369094848633, + "num_tokens": 373949114.0, + "step": 9631 + }, + { + "epoch": 1.2252894033837933, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.683773040771484, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.898419976234436, + "num_tokens": 373993392.0, + "step": 9632 + }, + { + "epoch": 1.2254166136623839, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.812216758728027, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8936183452606201, + "num_tokens": 374030708.0, + "step": 9633 + }, + { + "epoch": 1.2255438239409744, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.825430870056152, + "learning_rate": 1e-06, + "loss": 0.3838, + "mean_token_accuracy": 0.8717273473739624, + "num_tokens": 374073590.0, + "step": 9634 + }, + { + "epoch": 1.225671034219565, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.846179008483887, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8887871503829956, + "num_tokens": 374111487.0, + "step": 9635 + }, + { + "epoch": 1.2257982444981554, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.81624698638916, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8946102857589722, + "num_tokens": 374147277.0, + "step": 9636 + }, + { + "epoch": 1.225925454776746, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.806612014770508, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8927903175354004, + "num_tokens": 374184470.0, + "step": 9637 + }, + { + "epoch": 1.2260526650553365, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.864763259887695, + "learning_rate": 1e-06, + "loss": 0.3849, + "mean_token_accuracy": 0.8666460514068604, + "num_tokens": 374223765.0, + "step": 9638 + }, + { + "epoch": 1.226179875333927, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.788167953491211, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8823528289794922, + "num_tokens": 374266798.0, + "step": 9639 + }, + { + "epoch": 1.2263070856125176, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.829092979431152, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8923622369766235, + "num_tokens": 374303444.0, + "step": 9640 + }, + { + "epoch": 1.226434295891108, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.783248901367188, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8903396725654602, + "num_tokens": 374342688.0, + "step": 9641 + }, + { + "epoch": 1.2265615061696984, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.8233060836792, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8855866193771362, + "num_tokens": 374384015.0, + "step": 9642 + }, + { + "epoch": 1.226688716448289, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.897480964660645, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.882163405418396, + "num_tokens": 374417934.0, + "step": 9643 + }, + { + "epoch": 1.2268159267268794, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.794957160949707, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8881677389144897, + "num_tokens": 374451596.0, + "step": 9644 + }, + { + "epoch": 1.22694313700547, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.7799711227417, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8940380811691284, + "num_tokens": 374487151.0, + "step": 9645 + }, + { + "epoch": 1.2270703472840605, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.81533145904541, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8857807517051697, + "num_tokens": 374524090.0, + "step": 9646 + }, + { + "epoch": 1.227197557562651, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.754721641540527, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8962920904159546, + "num_tokens": 374563760.0, + "step": 9647 + }, + { + "epoch": 1.2273247678412416, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.75981616973877, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8765240907669067, + "num_tokens": 374605005.0, + "step": 9648 + }, + { + "epoch": 1.227451978119832, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.794327735900879, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8885481357574463, + "num_tokens": 374644631.0, + "step": 9649 + }, + { + "epoch": 1.2275791883984226, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.805221557617188, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8735377788543701, + "num_tokens": 374684946.0, + "step": 9650 + }, + { + "epoch": 1.2277063986770131, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.766217231750488, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8935631513595581, + "num_tokens": 374725814.0, + "step": 9651 + }, + { + "epoch": 1.2278336089556037, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.881936073303223, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8815058469772339, + "num_tokens": 374765948.0, + "step": 9652 + }, + { + "epoch": 1.2279608192341942, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.797078132629395, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8823930025100708, + "num_tokens": 374807530.0, + "step": 9653 + }, + { + "epoch": 1.2280880295127847, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.760499954223633, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8845393061637878, + "num_tokens": 374849589.0, + "step": 9654 + }, + { + "epoch": 1.2282152397913753, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.854512214660645, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8897117376327515, + "num_tokens": 374888238.0, + "step": 9655 + }, + { + "epoch": 1.2283424500699656, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.819146156311035, + "learning_rate": 1e-06, + "loss": 0.3771, + "mean_token_accuracy": 0.8659080266952515, + "num_tokens": 374925842.0, + "step": 9656 + }, + { + "epoch": 1.228469660348556, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.824098587036133, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.8731244206428528, + "num_tokens": 374967077.0, + "step": 9657 + }, + { + "epoch": 1.2285968706271466, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.786622047424316, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8999271988868713, + "num_tokens": 375000505.0, + "step": 9658 + }, + { + "epoch": 1.2287240809057371, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.850481986999512, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8890143036842346, + "num_tokens": 375034897.0, + "step": 9659 + }, + { + "epoch": 1.2288512911843277, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.795061111450195, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8864579796791077, + "num_tokens": 375072509.0, + "step": 9660 + }, + { + "epoch": 1.2289785014629182, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.70728874206543, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8778458833694458, + "num_tokens": 375119884.0, + "step": 9661 + }, + { + "epoch": 1.2291057117415087, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.752511978149414, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8992717266082764, + "num_tokens": 375159508.0, + "step": 9662 + }, + { + "epoch": 1.2292329220200993, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.733200073242188, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8996672034263611, + "num_tokens": 375201292.0, + "step": 9663 + }, + { + "epoch": 1.2293601322986898, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.819618225097656, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8775544762611389, + "num_tokens": 375242469.0, + "step": 9664 + }, + { + "epoch": 1.2294873425772803, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.83724308013916, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.8802379369735718, + "num_tokens": 375281941.0, + "step": 9665 + }, + { + "epoch": 1.2296145528558708, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.815138816833496, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8822900056838989, + "num_tokens": 375319253.0, + "step": 9666 + }, + { + "epoch": 1.2297417631344612, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.768240928649902, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8750331997871399, + "num_tokens": 375365418.0, + "step": 9667 + }, + { + "epoch": 1.2298689734130517, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.805587768554688, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.885933518409729, + "num_tokens": 375406044.0, + "step": 9668 + }, + { + "epoch": 1.2299961836916422, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.850375175476074, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8777036666870117, + "num_tokens": 375443620.0, + "step": 9669 + }, + { + "epoch": 1.2301233939702327, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.361556053161621, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.891462504863739, + "num_tokens": 375485184.0, + "step": 9670 + }, + { + "epoch": 1.2302506042488233, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.772984504699707, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.899329662322998, + "num_tokens": 375522054.0, + "step": 9671 + }, + { + "epoch": 1.2303778145274138, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.679644584655762, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8896317481994629, + "num_tokens": 375567658.0, + "step": 9672 + }, + { + "epoch": 1.2305050248060043, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.916523933410645, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8954747915267944, + "num_tokens": 375602805.0, + "step": 9673 + }, + { + "epoch": 1.2306322350845948, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.730034828186035, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8855874538421631, + "num_tokens": 375645601.0, + "step": 9674 + }, + { + "epoch": 1.2307594453631854, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.76915454864502, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.877427339553833, + "num_tokens": 375689957.0, + "step": 9675 + }, + { + "epoch": 1.230886655641776, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.57763671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.867191314697266, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8736962080001831, + "num_tokens": 375729925.0, + "step": 9676 + }, + { + "epoch": 1.2310138659203664, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.893977165222168, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8773329257965088, + "num_tokens": 375763730.0, + "step": 9677 + }, + { + "epoch": 1.231141076198957, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.76085376739502, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8777300715446472, + "num_tokens": 375804944.0, + "step": 9678 + }, + { + "epoch": 1.2312682864775475, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.950403213500977, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.87044757604599, + "num_tokens": 375839504.0, + "step": 9679 + }, + { + "epoch": 1.231395496756138, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.860225677490234, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.888386607170105, + "num_tokens": 375876141.0, + "step": 9680 + }, + { + "epoch": 1.2315227070347283, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.65469741821289, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8954498767852783, + "num_tokens": 375918347.0, + "step": 9681 + }, + { + "epoch": 1.2316499173133189, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.864067077636719, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8910850286483765, + "num_tokens": 375954003.0, + "step": 9682 + }, + { + "epoch": 1.2317771275919094, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.836819648742676, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8861583471298218, + "num_tokens": 375994668.0, + "step": 9683 + }, + { + "epoch": 1.2319043378705, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.650774955749512, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9103338122367859, + "num_tokens": 376036233.0, + "step": 9684 + }, + { + "epoch": 1.2320315481490904, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.85274600982666, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.884293794631958, + "num_tokens": 376077115.0, + "step": 9685 + }, + { + "epoch": 1.232158758427681, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.786846160888672, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8826296329498291, + "num_tokens": 376121684.0, + "step": 9686 + }, + { + "epoch": 1.2322859687062715, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.8748140335083, + "learning_rate": 1e-06, + "loss": 0.3776, + "mean_token_accuracy": 0.8727740049362183, + "num_tokens": 376156788.0, + "step": 9687 + }, + { + "epoch": 1.232413178984862, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.831650733947754, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8827997446060181, + "num_tokens": 376195279.0, + "step": 9688 + }, + { + "epoch": 1.2325403892634526, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.770788192749023, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8760921955108643, + "num_tokens": 376236395.0, + "step": 9689 + }, + { + "epoch": 1.232667599542043, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.914523124694824, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8869628310203552, + "num_tokens": 376269475.0, + "step": 9690 + }, + { + "epoch": 1.2327948098206334, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.692627906799316, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.9024654626846313, + "num_tokens": 376313547.0, + "step": 9691 + }, + { + "epoch": 1.232922020099224, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.601478576660156e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.78791332244873, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8882691264152527, + "num_tokens": 376349498.0, + "step": 9692 + }, + { + "epoch": 1.2330492303778144, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.76043701171875, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8894994854927063, + "num_tokens": 376394525.0, + "step": 9693 + }, + { + "epoch": 1.233176440656405, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.754589080810547, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8952546119689941, + "num_tokens": 376432734.0, + "step": 9694 + }, + { + "epoch": 1.2333036509349955, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.760330200195312, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.9028816223144531, + "num_tokens": 376474400.0, + "step": 9695 + }, + { + "epoch": 1.233430861213586, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.823702812194824, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8851270079612732, + "num_tokens": 376516493.0, + "step": 9696 + }, + { + "epoch": 1.2335580714921766, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.002494812011719, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8701837658882141, + "num_tokens": 376547842.0, + "step": 9697 + }, + { + "epoch": 1.233685281770767, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.791716575622559, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8841582536697388, + "num_tokens": 376589672.0, + "step": 9698 + }, + { + "epoch": 1.2338124920493576, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.777591705322266, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8833930492401123, + "num_tokens": 376633067.0, + "step": 9699 + }, + { + "epoch": 1.2339397023279481, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.863067626953125, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8925426602363586, + "num_tokens": 376673157.0, + "step": 9700 + }, + { + "epoch": 1.2340669126065387, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.827712059020996, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8811503052711487, + "num_tokens": 376712359.0, + "step": 9701 + }, + { + "epoch": 1.2341941228851292, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.867629051208496, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8970162868499756, + "num_tokens": 376744355.0, + "step": 9702 + }, + { + "epoch": 1.2343213331637197, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.789915084838867, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8811787962913513, + "num_tokens": 376789332.0, + "step": 9703 + }, + { + "epoch": 1.2344485434423103, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.836695671081543, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8951216340065002, + "num_tokens": 376823509.0, + "step": 9704 + }, + { + "epoch": 1.2345757537209006, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.808358192443848, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.88543701171875, + "num_tokens": 376863783.0, + "step": 9705 + }, + { + "epoch": 1.234702963999491, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.764430046081543, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.9011287689208984, + "num_tokens": 376904438.0, + "step": 9706 + }, + { + "epoch": 1.2348301742780816, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.808297157287598, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8954203128814697, + "num_tokens": 376939806.0, + "step": 9707 + }, + { + "epoch": 1.2349573845566721, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.990032196044922, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.8679571151733398, + "num_tokens": 376977596.0, + "step": 9708 + }, + { + "epoch": 1.2350845948352627, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.80955696105957, + "learning_rate": 1e-06, + "loss": 0.3899, + "mean_token_accuracy": 0.8679524064064026, + "num_tokens": 377021527.0, + "step": 9709 + }, + { + "epoch": 1.2352118051138532, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.842817306518555, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8938022255897522, + "num_tokens": 377062751.0, + "step": 9710 + }, + { + "epoch": 1.2353390153924437, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.737335205078125, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8933587074279785, + "num_tokens": 377108718.0, + "step": 9711 + }, + { + "epoch": 1.2354662256710343, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.831985473632812, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8884900808334351, + "num_tokens": 377147544.0, + "step": 9712 + }, + { + "epoch": 1.2355934359496248, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.859238624572754, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8886160850524902, + "num_tokens": 377184936.0, + "step": 9713 + }, + { + "epoch": 1.2357206462282153, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.850261688232422, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8714872598648071, + "num_tokens": 377225330.0, + "step": 9714 + }, + { + "epoch": 1.2358478565068058, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.799479484558105, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8945205211639404, + "num_tokens": 377262662.0, + "step": 9715 + }, + { + "epoch": 1.2359750667853961, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.811246871948242, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8881962299346924, + "num_tokens": 377304702.0, + "step": 9716 + }, + { + "epoch": 1.2361022770639867, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.907218933105469, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8848152756690979, + "num_tokens": 377342649.0, + "step": 9717 + }, + { + "epoch": 1.2362294873425772, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.783998489379883, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8969041705131531, + "num_tokens": 377379990.0, + "step": 9718 + }, + { + "epoch": 1.2363566976211677, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.849166870117188, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8816512227058411, + "num_tokens": 377421548.0, + "step": 9719 + }, + { + "epoch": 1.2364839078997583, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.923343658447266, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8744131326675415, + "num_tokens": 377456157.0, + "step": 9720 + }, + { + "epoch": 1.2366111181783488, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.890271186828613, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8804702758789062, + "num_tokens": 377495947.0, + "step": 9721 + }, + { + "epoch": 1.2367383284569393, + "ewc_loss": 0.01446533203125, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.678135871887207, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8993246555328369, + "num_tokens": 377541240.0, + "step": 9722 + }, + { + "epoch": 1.2368655387355298, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.934968948364258, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.8733596205711365, + "num_tokens": 377579358.0, + "step": 9723 + }, + { + "epoch": 1.2369927490141204, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.86924934387207, + "learning_rate": 1e-06, + "loss": 0.3797, + "mean_token_accuracy": 0.8679367899894714, + "num_tokens": 377621907.0, + "step": 9724 + }, + { + "epoch": 1.237119959292711, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.807059288024902, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8900154232978821, + "num_tokens": 377661247.0, + "step": 9725 + }, + { + "epoch": 1.2372471695713014, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.732810020446777, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.8976438045501709, + "num_tokens": 377699978.0, + "step": 9726 + }, + { + "epoch": 1.237374379849892, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.833368301391602, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8878070116043091, + "num_tokens": 377740315.0, + "step": 9727 + }, + { + "epoch": 1.2375015901284825, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.812461853027344, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.884270966053009, + "num_tokens": 377777954.0, + "step": 9728 + }, + { + "epoch": 1.237628800407073, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.831624031066895, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8889959454536438, + "num_tokens": 377814556.0, + "step": 9729 + }, + { + "epoch": 1.2377560106856633, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.915658950805664, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.891898512840271, + "num_tokens": 377843812.0, + "step": 9730 + }, + { + "epoch": 1.2378832209642538, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.868657112121582, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8898682594299316, + "num_tokens": 377881920.0, + "step": 9731 + }, + { + "epoch": 1.2380104312428444, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.716506958007812, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8897219896316528, + "num_tokens": 377921228.0, + "step": 9732 + }, + { + "epoch": 1.238137641521435, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.75813102722168, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8863879442214966, + "num_tokens": 377964530.0, + "step": 9733 + }, + { + "epoch": 1.2382648518000254, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.947708129882812, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.873399019241333, + "num_tokens": 377999296.0, + "step": 9734 + }, + { + "epoch": 1.238392062078616, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.752716064453125, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.897454023361206, + "num_tokens": 378039231.0, + "step": 9735 + }, + { + "epoch": 1.2385192723572065, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.777419090270996, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.9010968804359436, + "num_tokens": 378078386.0, + "step": 9736 + }, + { + "epoch": 1.238646482635797, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.795368194580078, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8926489353179932, + "num_tokens": 378119378.0, + "step": 9737 + }, + { + "epoch": 1.2387736929143875, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.847752571105957, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8968521356582642, + "num_tokens": 378153692.0, + "step": 9738 + }, + { + "epoch": 1.238900903192978, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.7711181640625, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8881177306175232, + "num_tokens": 378198294.0, + "step": 9739 + }, + { + "epoch": 1.2390281134715684, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.902531623840332, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8851702809333801, + "num_tokens": 378232713.0, + "step": 9740 + }, + { + "epoch": 1.239155323750159, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.78870964050293, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.88636314868927, + "num_tokens": 378272811.0, + "step": 9741 + }, + { + "epoch": 1.2392825340287494, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.870039939880371, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.879677414894104, + "num_tokens": 378314114.0, + "step": 9742 + }, + { + "epoch": 1.23940974430734, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.807234764099121, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8819020986557007, + "num_tokens": 378356685.0, + "step": 9743 + }, + { + "epoch": 1.2395369545859305, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.939950942993164, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8855113387107849, + "num_tokens": 378386974.0, + "step": 9744 + }, + { + "epoch": 1.239664164864521, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.80324649810791, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8977671265602112, + "num_tokens": 378421627.0, + "step": 9745 + }, + { + "epoch": 1.2397913751431116, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.834450721740723, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8872725963592529, + "num_tokens": 378458481.0, + "step": 9746 + }, + { + "epoch": 1.239918585421702, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.736757278442383, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8945497870445251, + "num_tokens": 378498994.0, + "step": 9747 + }, + { + "epoch": 1.2400457957002926, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 9.024957656860352, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8827623724937439, + "num_tokens": 378529147.0, + "step": 9748 + }, + { + "epoch": 1.2401730059788831, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.81580638885498, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8875822424888611, + "num_tokens": 378570165.0, + "step": 9749 + }, + { + "epoch": 1.2403002162574737, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.802312850952148, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.890329122543335, + "num_tokens": 378605271.0, + "step": 9750 + }, + { + "epoch": 1.2404274265360642, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.788687705993652, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8933422565460205, + "num_tokens": 378643616.0, + "step": 9751 + }, + { + "epoch": 1.2405546368146547, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.829853057861328, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.888506293296814, + "num_tokens": 378678943.0, + "step": 9752 + }, + { + "epoch": 1.2406818470932452, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.988435745239258, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8948320150375366, + "num_tokens": 378709768.0, + "step": 9753 + }, + { + "epoch": 1.2408090573718356, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.75694465637207, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8895200490951538, + "num_tokens": 378750593.0, + "step": 9754 + }, + { + "epoch": 1.240936267650426, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.847393989562988, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8852287530899048, + "num_tokens": 378786703.0, + "step": 9755 + }, + { + "epoch": 1.2410634779290166, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.6253204345703125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.918456077575684, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8811670541763306, + "num_tokens": 378818334.0, + "step": 9756 + }, + { + "epoch": 1.2411906882076071, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.748188972473145, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8891795873641968, + "num_tokens": 378856781.0, + "step": 9757 + }, + { + "epoch": 1.2413178984861977, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.816879272460938, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8821976184844971, + "num_tokens": 378895159.0, + "step": 9758 + }, + { + "epoch": 1.2414451087647882, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.858728408813477, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.894952118396759, + "num_tokens": 378928884.0, + "step": 9759 + }, + { + "epoch": 1.2415723190433787, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.77841854095459, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8787338137626648, + "num_tokens": 378971599.0, + "step": 9760 + }, + { + "epoch": 1.2416995293219693, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.835976600646973, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.8759725093841553, + "num_tokens": 379008783.0, + "step": 9761 + }, + { + "epoch": 1.2418267396005598, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.83928108215332, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8914337158203125, + "num_tokens": 379046512.0, + "step": 9762 + }, + { + "epoch": 1.2419539498791503, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.745000839233398, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.9002097845077515, + "num_tokens": 379083299.0, + "step": 9763 + }, + { + "epoch": 1.2420811601577408, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.823382377624512, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8920488357543945, + "num_tokens": 379117120.0, + "step": 9764 + }, + { + "epoch": 1.2422083704363311, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.78409194946289, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8906471729278564, + "num_tokens": 379158245.0, + "step": 9765 + }, + { + "epoch": 1.2423355807149217, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.81620979309082, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8817465305328369, + "num_tokens": 379194825.0, + "step": 9766 + }, + { + "epoch": 1.2424627909935122, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.85129165649414, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8812682628631592, + "num_tokens": 379231196.0, + "step": 9767 + }, + { + "epoch": 1.2425900012721027, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.821414947509766, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8880724310874939, + "num_tokens": 379267201.0, + "step": 9768 + }, + { + "epoch": 1.2427172115506933, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.970029830932617, + "learning_rate": 1e-06, + "loss": 0.3844, + "mean_token_accuracy": 0.8700958490371704, + "num_tokens": 379300051.0, + "step": 9769 + }, + { + "epoch": 1.2428444218292838, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.804363250732422, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8955180644989014, + "num_tokens": 379331783.0, + "step": 9770 + }, + { + "epoch": 1.2429716321078743, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.785323143005371, + "learning_rate": 1e-06, + "loss": 0.3917, + "mean_token_accuracy": 0.8688175082206726, + "num_tokens": 379374915.0, + "step": 9771 + }, + { + "epoch": 1.2430988423864648, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.754353523254395, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8916023373603821, + "num_tokens": 379413126.0, + "step": 9772 + }, + { + "epoch": 1.2432260526650554, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.806639671325684, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8922257423400879, + "num_tokens": 379450137.0, + "step": 9773 + }, + { + "epoch": 1.243353262943646, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.70736312866211, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.897603452205658, + "num_tokens": 379489414.0, + "step": 9774 + }, + { + "epoch": 1.2434804732222364, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.863179206848145, + "learning_rate": 1e-06, + "loss": 0.3631, + "mean_token_accuracy": 0.8759704232215881, + "num_tokens": 379527996.0, + "step": 9775 + }, + { + "epoch": 1.243607683500827, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.779230117797852, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8873637914657593, + "num_tokens": 379566748.0, + "step": 9776 + }, + { + "epoch": 1.2437348937794175, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.841395378112793, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8882821798324585, + "num_tokens": 379604843.0, + "step": 9777 + }, + { + "epoch": 1.243862104058008, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.853705406188965, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8828150629997253, + "num_tokens": 379640586.0, + "step": 9778 + }, + { + "epoch": 1.2439893143365983, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.848341941833496, + "learning_rate": 1e-06, + "loss": 0.3885, + "mean_token_accuracy": 0.8660621643066406, + "num_tokens": 379679656.0, + "step": 9779 + }, + { + "epoch": 1.2441165246151888, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.825551986694336, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8920425772666931, + "num_tokens": 379715914.0, + "step": 9780 + }, + { + "epoch": 1.2442437348937794, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.901025772094727, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8892607092857361, + "num_tokens": 379748495.0, + "step": 9781 + }, + { + "epoch": 1.24437094517237, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.776966094970703, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8908021450042725, + "num_tokens": 379792814.0, + "step": 9782 + }, + { + "epoch": 1.2444981554509604, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.837682723999023, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8805445432662964, + "num_tokens": 379829900.0, + "step": 9783 + }, + { + "epoch": 1.244625365729551, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.922736167907715, + "learning_rate": 1e-06, + "loss": 0.3965, + "mean_token_accuracy": 0.8690872192382812, + "num_tokens": 379867867.0, + "step": 9784 + }, + { + "epoch": 1.2447525760081415, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.726363182067871, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8904780149459839, + "num_tokens": 379912680.0, + "step": 9785 + }, + { + "epoch": 1.244879786286732, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.852398872375488, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8933844566345215, + "num_tokens": 379950690.0, + "step": 9786 + }, + { + "epoch": 1.2450069965653225, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.862288475036621, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8939586877822876, + "num_tokens": 379985833.0, + "step": 9787 + }, + { + "epoch": 1.245134206843913, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.808141708374023, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8889644145965576, + "num_tokens": 380024502.0, + "step": 9788 + }, + { + "epoch": 1.2452614171225034, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.759758949279785, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8926339745521545, + "num_tokens": 380064873.0, + "step": 9789 + }, + { + "epoch": 1.245388627401094, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.815840721130371, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8918294906616211, + "num_tokens": 380102057.0, + "step": 9790 + }, + { + "epoch": 1.2455158376796844, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.867938995361328, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8789980411529541, + "num_tokens": 380138923.0, + "step": 9791 + }, + { + "epoch": 1.245643047958275, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.886887550354004, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.871963381767273, + "num_tokens": 380177927.0, + "step": 9792 + }, + { + "epoch": 1.2457702582368655, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.836345672607422, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.901049792766571, + "num_tokens": 380211274.0, + "step": 9793 + }, + { + "epoch": 1.245897468515456, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.812742233276367, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8814052939414978, + "num_tokens": 380254865.0, + "step": 9794 + }, + { + "epoch": 1.2460246787940465, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.76366138458252, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8882911205291748, + "num_tokens": 380299590.0, + "step": 9795 + }, + { + "epoch": 1.246151889072637, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.857656478881836, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.8694361448287964, + "num_tokens": 380340043.0, + "step": 9796 + }, + { + "epoch": 1.2462790993512276, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.820902824401855, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8856931328773499, + "num_tokens": 380377177.0, + "step": 9797 + }, + { + "epoch": 1.2464063096298181, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.817460060119629, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8826196789741516, + "num_tokens": 380417351.0, + "step": 9798 + }, + { + "epoch": 1.2465335199084087, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.777297019958496, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8805959224700928, + "num_tokens": 380461378.0, + "step": 9799 + }, + { + "epoch": 1.2466607301869992, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.796995162963867, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9022809863090515, + "num_tokens": 380498575.0, + "step": 9800 + }, + { + "epoch": 1.2467879404655897, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.781618118286133, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8892443776130676, + "num_tokens": 380538733.0, + "step": 9801 + }, + { + "epoch": 1.2469151507441802, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.789200782775879, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8993517160415649, + "num_tokens": 380579674.0, + "step": 9802 + }, + { + "epoch": 1.2470423610227706, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.847892761230469, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.87386155128479, + "num_tokens": 380618498.0, + "step": 9803 + }, + { + "epoch": 1.247169571301361, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.898496627807617, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8839178681373596, + "num_tokens": 380655467.0, + "step": 9804 + }, + { + "epoch": 1.2472967815799516, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.754621505737305, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8966985940933228, + "num_tokens": 380700166.0, + "step": 9805 + }, + { + "epoch": 1.2474239918585421, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.870796203613281, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8826780915260315, + "num_tokens": 380737807.0, + "step": 9806 + }, + { + "epoch": 1.2475512021371327, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.771856307983398, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8909366130828857, + "num_tokens": 380781601.0, + "step": 9807 + }, + { + "epoch": 1.2476784124157232, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.868059158325195, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8832825422286987, + "num_tokens": 380820381.0, + "step": 9808 + }, + { + "epoch": 1.2478056226943137, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.790970802307129, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8854057788848877, + "num_tokens": 380862968.0, + "step": 9809 + }, + { + "epoch": 1.2479328329729042, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.80361270904541, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8966096043586731, + "num_tokens": 380901432.0, + "step": 9810 + }, + { + "epoch": 1.2480600432514948, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.825380325317383, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8936936855316162, + "num_tokens": 380939251.0, + "step": 9811 + }, + { + "epoch": 1.2481872535300853, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.807866096496582, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.898043155670166, + "num_tokens": 380974863.0, + "step": 9812 + }, + { + "epoch": 1.2483144638086758, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.73971939086914, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8894354104995728, + "num_tokens": 381016450.0, + "step": 9813 + }, + { + "epoch": 1.2484416740872661, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.80833911895752, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8832679986953735, + "num_tokens": 381054613.0, + "step": 9814 + }, + { + "epoch": 1.2485688843658567, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.791211128234863, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8833544850349426, + "num_tokens": 381091043.0, + "step": 9815 + }, + { + "epoch": 1.2486960946444472, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.739434242248535, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.9010115265846252, + "num_tokens": 381130332.0, + "step": 9816 + }, + { + "epoch": 1.2488233049230377, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.807458877563477, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8840380907058716, + "num_tokens": 381169402.0, + "step": 9817 + }, + { + "epoch": 1.2489505152016283, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.810070037841797, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8810669183731079, + "num_tokens": 381208118.0, + "step": 9818 + }, + { + "epoch": 1.2490777254802188, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.933415412902832, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8902550935745239, + "num_tokens": 381239040.0, + "step": 9819 + }, + { + "epoch": 1.2492049357588093, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.851032257080078, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8880301713943481, + "num_tokens": 381275676.0, + "step": 9820 + }, + { + "epoch": 1.2493321460373998, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.793745994567871, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8881329894065857, + "num_tokens": 381317579.0, + "step": 9821 + }, + { + "epoch": 1.2494593563159904, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.824945449829102, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8803585767745972, + "num_tokens": 381357589.0, + "step": 9822 + }, + { + "epoch": 1.249586566594581, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.844018936157227, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8880360126495361, + "num_tokens": 381393353.0, + "step": 9823 + }, + { + "epoch": 1.2497137768731714, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.818975448608398, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8786374926567078, + "num_tokens": 381433348.0, + "step": 9824 + }, + { + "epoch": 1.249840987151762, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.707361221313477, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9005309343338013, + "num_tokens": 381480561.0, + "step": 9825 + }, + { + "epoch": 1.2499681974303525, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.853909492492676, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8800727128982544, + "num_tokens": 381519446.0, + "step": 9826 + }, + { + "epoch": 1.250095407708943, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.919684410095215, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8786377906799316, + "num_tokens": 381555125.0, + "step": 9827 + }, + { + "epoch": 1.2502226179875333, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.803462982177734, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8844869136810303, + "num_tokens": 381593684.0, + "step": 9828 + }, + { + "epoch": 1.2503498282661238, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.93891429901123, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8825434446334839, + "num_tokens": 381632591.0, + "step": 9829 + }, + { + "epoch": 1.2504770385447144, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.795572280883789, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8911000490188599, + "num_tokens": 381671513.0, + "step": 9830 + }, + { + "epoch": 1.250604248823305, + "ewc_loss": 0.0145263671875, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.825089454650879, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8811625242233276, + "num_tokens": 381708973.0, + "step": 9831 + }, + { + "epoch": 1.2507314591018954, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.919842720031738, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8682558536529541, + "num_tokens": 381744044.0, + "step": 9832 + }, + { + "epoch": 1.250858669380486, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.78916072845459, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8843041658401489, + "num_tokens": 381788432.0, + "step": 9833 + }, + { + "epoch": 1.2509858796590765, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.87563419342041, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8816981315612793, + "num_tokens": 381823712.0, + "step": 9834 + }, + { + "epoch": 1.251113089937667, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.828654289245605, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8878539800643921, + "num_tokens": 381862363.0, + "step": 9835 + }, + { + "epoch": 1.2512403002162575, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.878662109375, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8959517478942871, + "num_tokens": 381898532.0, + "step": 9836 + }, + { + "epoch": 1.2513675104948478, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.805000305175781, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8930732011795044, + "num_tokens": 381938420.0, + "step": 9837 + }, + { + "epoch": 1.2514947207734384, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.879925727844238, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8909376859664917, + "num_tokens": 381974017.0, + "step": 9838 + }, + { + "epoch": 1.251621931052029, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.87959098815918, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8822354078292847, + "num_tokens": 382008481.0, + "step": 9839 + }, + { + "epoch": 1.2517491413306194, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.673004150390625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.83752155303955, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8992114067077637, + "num_tokens": 382046929.0, + "step": 9840 + }, + { + "epoch": 1.25187635160921, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.808894157409668, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8817164301872253, + "num_tokens": 382087738.0, + "step": 9841 + }, + { + "epoch": 1.2520035618878005, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.86294937133789, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.890560507774353, + "num_tokens": 382132644.0, + "step": 9842 + }, + { + "epoch": 1.252130772166391, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.885576248168945, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8789915442466736, + "num_tokens": 382173419.0, + "step": 9843 + }, + { + "epoch": 1.2522579824449815, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.91732406616211, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8772622346878052, + "num_tokens": 382208721.0, + "step": 9844 + }, + { + "epoch": 1.252385192723572, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.833344459533691, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.897818386554718, + "num_tokens": 382243198.0, + "step": 9845 + }, + { + "epoch": 1.2525124030021626, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.794900894165039, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8923408389091492, + "num_tokens": 382285453.0, + "step": 9846 + }, + { + "epoch": 1.2526396132807531, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.758010864257812, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8974099159240723, + "num_tokens": 382326512.0, + "step": 9847 + }, + { + "epoch": 1.2527668235593437, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.908638954162598, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.877740204334259, + "num_tokens": 382363292.0, + "step": 9848 + }, + { + "epoch": 1.2528940338379342, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.910101890563965, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8699308037757874, + "num_tokens": 382398289.0, + "step": 9849 + }, + { + "epoch": 1.2530212441165247, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.858689308166504, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8882753252983093, + "num_tokens": 382437796.0, + "step": 9850 + }, + { + "epoch": 1.2531484543951152, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.757819175720215, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8841115236282349, + "num_tokens": 382481170.0, + "step": 9851 + }, + { + "epoch": 1.2532756646737058, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.855508804321289, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8874132633209229, + "num_tokens": 382518896.0, + "step": 9852 + }, + { + "epoch": 1.253402874952296, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.938880920410156, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8806814551353455, + "num_tokens": 382554449.0, + "step": 9853 + }, + { + "epoch": 1.2535300852308866, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.872225761413574, + "learning_rate": 1e-06, + "loss": 0.3967, + "mean_token_accuracy": 0.8650650978088379, + "num_tokens": 382592265.0, + "step": 9854 + }, + { + "epoch": 1.2536572955094771, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.852340698242188, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8765273094177246, + "num_tokens": 382634404.0, + "step": 9855 + }, + { + "epoch": 1.2537845057880677, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.877777099609375, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8908206224441528, + "num_tokens": 382666696.0, + "step": 9856 + }, + { + "epoch": 1.2539117160666582, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.81009578704834, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8805621862411499, + "num_tokens": 382706449.0, + "step": 9857 + }, + { + "epoch": 1.2540389263452487, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.990180969238281, + "learning_rate": 1e-06, + "loss": 0.3965, + "mean_token_accuracy": 0.8663214445114136, + "num_tokens": 382743326.0, + "step": 9858 + }, + { + "epoch": 1.2541661366238392, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.94682502746582, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.879139244556427, + "num_tokens": 382777020.0, + "step": 9859 + }, + { + "epoch": 1.2542933469024298, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.731325149536133, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8813996315002441, + "num_tokens": 382819008.0, + "step": 9860 + }, + { + "epoch": 1.2544205571810203, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.766255378723145, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8793895244598389, + "num_tokens": 382863249.0, + "step": 9861 + }, + { + "epoch": 1.2545477674596106, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.811219215393066, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.890083372592926, + "num_tokens": 382906783.0, + "step": 9862 + }, + { + "epoch": 1.2546749777382011, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.83010196685791, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8908178806304932, + "num_tokens": 382944286.0, + "step": 9863 + }, + { + "epoch": 1.2548021880167917, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.83586597442627, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8809490203857422, + "num_tokens": 382984855.0, + "step": 9864 + }, + { + "epoch": 1.2549293982953822, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.877195358276367, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.892615556716919, + "num_tokens": 383022491.0, + "step": 9865 + }, + { + "epoch": 1.2550566085739727, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.937154769897461, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8814951181411743, + "num_tokens": 383057947.0, + "step": 9866 + }, + { + "epoch": 1.2551838188525632, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.817145347595215, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8897020220756531, + "num_tokens": 383099496.0, + "step": 9867 + }, + { + "epoch": 1.2553110291311538, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.803956031799316, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8893519639968872, + "num_tokens": 383137991.0, + "step": 9868 + }, + { + "epoch": 1.2554382394097443, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.848145484924316, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8839344382286072, + "num_tokens": 383178965.0, + "step": 9869 + }, + { + "epoch": 1.2555654496883348, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.827014923095703, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8922045230865479, + "num_tokens": 383219888.0, + "step": 9870 + }, + { + "epoch": 1.2556926599669254, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.847929000854492, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8918972015380859, + "num_tokens": 383256994.0, + "step": 9871 + }, + { + "epoch": 1.255819870245516, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.820477485656738, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.9024515748023987, + "num_tokens": 383295281.0, + "step": 9872 + }, + { + "epoch": 1.2559470805241064, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.753801345825195, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.9016374349594116, + "num_tokens": 383338187.0, + "step": 9873 + }, + { + "epoch": 1.256074290802697, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.795361518859863, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8825559616088867, + "num_tokens": 383380548.0, + "step": 9874 + }, + { + "epoch": 1.2562015010812875, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.83950424194336, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8866899013519287, + "num_tokens": 383422838.0, + "step": 9875 + }, + { + "epoch": 1.256328711359878, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.87010383605957, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8902118802070618, + "num_tokens": 383454713.0, + "step": 9876 + }, + { + "epoch": 1.2564559216384683, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.80057430267334, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.890272855758667, + "num_tokens": 383496827.0, + "step": 9877 + }, + { + "epoch": 1.2565831319170588, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.79128360748291, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8834587335586548, + "num_tokens": 383537348.0, + "step": 9878 + }, + { + "epoch": 1.2567103421956494, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.800958633422852, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8968064785003662, + "num_tokens": 383578342.0, + "step": 9879 + }, + { + "epoch": 1.25683755247424, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.983708381652832, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8752669095993042, + "num_tokens": 383613880.0, + "step": 9880 + }, + { + "epoch": 1.2569647627528304, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.771656036376953, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8920378684997559, + "num_tokens": 383656481.0, + "step": 9881 + }, + { + "epoch": 1.257091973031421, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.911130905151367, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8908018469810486, + "num_tokens": 383692896.0, + "step": 9882 + }, + { + "epoch": 1.2572191833100115, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.696846008300781e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.793559074401855, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.894506573677063, + "num_tokens": 383734866.0, + "step": 9883 + }, + { + "epoch": 1.257346393588602, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.856054306030273, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8888310194015503, + "num_tokens": 383771210.0, + "step": 9884 + }, + { + "epoch": 1.2574736038671925, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.881953239440918, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8861242532730103, + "num_tokens": 383806273.0, + "step": 9885 + }, + { + "epoch": 1.2576008141457828, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.836746215820312, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8825802206993103, + "num_tokens": 383847799.0, + "step": 9886 + }, + { + "epoch": 1.2577280244243734, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.778963088989258, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9005632400512695, + "num_tokens": 383888732.0, + "step": 9887 + }, + { + "epoch": 1.257855234702964, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.831160545349121, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8998278379440308, + "num_tokens": 383926971.0, + "step": 9888 + }, + { + "epoch": 1.2579824449815544, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.895825386047363, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8819060325622559, + "num_tokens": 383965875.0, + "step": 9889 + }, + { + "epoch": 1.258109655260145, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.809619903564453, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8944538235664368, + "num_tokens": 384008281.0, + "step": 9890 + }, + { + "epoch": 1.2582368655387355, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.852523803710938, + "learning_rate": 1e-06, + "loss": 0.3954, + "mean_token_accuracy": 0.865588903427124, + "num_tokens": 384053188.0, + "step": 9891 + }, + { + "epoch": 1.258364075817326, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.897147178649902, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8860175013542175, + "num_tokens": 384090509.0, + "step": 9892 + }, + { + "epoch": 1.2584912860959165, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.83725643157959, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8776817321777344, + "num_tokens": 384132524.0, + "step": 9893 + }, + { + "epoch": 1.258618496374507, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.762845039367676, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.9006646871566772, + "num_tokens": 384169021.0, + "step": 9894 + }, + { + "epoch": 1.2587457066530976, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.891621589660645, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8867498636245728, + "num_tokens": 384211608.0, + "step": 9895 + }, + { + "epoch": 1.2588729169316881, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.81499195098877, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.9004578590393066, + "num_tokens": 384244860.0, + "step": 9896 + }, + { + "epoch": 1.2590001272102787, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 9.020898818969727, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8848313093185425, + "num_tokens": 384275421.0, + "step": 9897 + }, + { + "epoch": 1.2591273374888692, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.886802673339844, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8938134908676147, + "num_tokens": 384310559.0, + "step": 9898 + }, + { + "epoch": 1.2592545477674597, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.947799682617188, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8892978429794312, + "num_tokens": 384346248.0, + "step": 9899 + }, + { + "epoch": 1.2593817580460502, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.926990509033203, + "learning_rate": 1e-06, + "loss": 0.3774, + "mean_token_accuracy": 0.8715353608131409, + "num_tokens": 384383139.0, + "step": 9900 + }, + { + "epoch": 1.2595089683246408, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.77252197265625, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8933936357498169, + "num_tokens": 384421630.0, + "step": 9901 + }, + { + "epoch": 1.259636178603231, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.928143501281738, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8785643577575684, + "num_tokens": 384461239.0, + "step": 9902 + }, + { + "epoch": 1.2597633888818216, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.846240997314453, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8769861459732056, + "num_tokens": 384506861.0, + "step": 9903 + }, + { + "epoch": 1.2598905991604121, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.899640083312988, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8814778327941895, + "num_tokens": 384546171.0, + "step": 9904 + }, + { + "epoch": 1.2600178094390027, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.816056251525879, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8882716298103333, + "num_tokens": 384586729.0, + "step": 9905 + }, + { + "epoch": 1.2601450197175932, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.885710716247559, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8706831336021423, + "num_tokens": 384633155.0, + "step": 9906 + }, + { + "epoch": 1.2602722299961837, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.85661792755127, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8897339105606079, + "num_tokens": 384671331.0, + "step": 9907 + }, + { + "epoch": 1.2603994402747742, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.920524597167969, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8762156963348389, + "num_tokens": 384708546.0, + "step": 9908 + }, + { + "epoch": 1.2605266505533648, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.92459487915039, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8855785131454468, + "num_tokens": 384746190.0, + "step": 9909 + }, + { + "epoch": 1.2606538608319553, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.864166259765625, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8873260021209717, + "num_tokens": 384784307.0, + "step": 9910 + }, + { + "epoch": 1.2607810711105456, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.84518814086914, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8960263133049011, + "num_tokens": 384818805.0, + "step": 9911 + }, + { + "epoch": 1.2609082813891361, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.92326831817627, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8793066740036011, + "num_tokens": 384851473.0, + "step": 9912 + }, + { + "epoch": 1.2610354916677267, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.870152473449707, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8952776193618774, + "num_tokens": 384884463.0, + "step": 9913 + }, + { + "epoch": 1.2611627019463172, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.862061500549316, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8873996734619141, + "num_tokens": 384921317.0, + "step": 9914 + }, + { + "epoch": 1.2612899122249077, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.828568458557129, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8859221935272217, + "num_tokens": 384961164.0, + "step": 9915 + }, + { + "epoch": 1.2614171225034982, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.820355415344238, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.8736171722412109, + "num_tokens": 385007039.0, + "step": 9916 + }, + { + "epoch": 1.2615443327820888, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.895500183105469, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8967061042785645, + "num_tokens": 385039866.0, + "step": 9917 + }, + { + "epoch": 1.2616715430606793, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.810620307922363, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8957628011703491, + "num_tokens": 385078011.0, + "step": 9918 + }, + { + "epoch": 1.2617987533392698, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.854249954223633, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8924202919006348, + "num_tokens": 385115217.0, + "step": 9919 + }, + { + "epoch": 1.2619259636178604, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.784465789794922, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8839195966720581, + "num_tokens": 385156312.0, + "step": 9920 + }, + { + "epoch": 1.2620531738964509, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.7206878662109375e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.841107368469238, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8896005153656006, + "num_tokens": 385200301.0, + "step": 9921 + }, + { + "epoch": 1.2621803841750414, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.826714515686035, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8960872888565063, + "num_tokens": 385239260.0, + "step": 9922 + }, + { + "epoch": 1.262307594453632, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.758793830871582, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.897399365901947, + "num_tokens": 385287336.0, + "step": 9923 + }, + { + "epoch": 1.2624348047322225, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.930094718933105, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8842076063156128, + "num_tokens": 385325157.0, + "step": 9924 + }, + { + "epoch": 1.262562015010813, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.90754222869873, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8796443939208984, + "num_tokens": 385361597.0, + "step": 9925 + }, + { + "epoch": 1.2626892252894033, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.752982139587402, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8808894753456116, + "num_tokens": 385410334.0, + "step": 9926 + }, + { + "epoch": 1.2628164355679938, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 9.069361686706543, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.884646475315094, + "num_tokens": 385451293.0, + "step": 9927 + }, + { + "epoch": 1.2629436458465844, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.767719268798828, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.899556577205658, + "num_tokens": 385493943.0, + "step": 9928 + }, + { + "epoch": 1.263070856125175, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.75584888458252, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8878092169761658, + "num_tokens": 385541617.0, + "step": 9929 + }, + { + "epoch": 1.2631980664037654, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.855198860168457, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.877770721912384, + "num_tokens": 385581959.0, + "step": 9930 + }, + { + "epoch": 1.263325276682356, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.784016609191895, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8990111351013184, + "num_tokens": 385622716.0, + "step": 9931 + }, + { + "epoch": 1.2634524869609465, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.925200462341309, + "learning_rate": 1e-06, + "loss": 0.3887, + "mean_token_accuracy": 0.8692155480384827, + "num_tokens": 385663763.0, + "step": 9932 + }, + { + "epoch": 1.263579697239537, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.851034164428711, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8854912519454956, + "num_tokens": 385704391.0, + "step": 9933 + }, + { + "epoch": 1.2637069075181275, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.757697105407715, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8861665725708008, + "num_tokens": 385747373.0, + "step": 9934 + }, + { + "epoch": 1.2638341177967178, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.850255966186523, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.8744117021560669, + "num_tokens": 385792096.0, + "step": 9935 + }, + { + "epoch": 1.2639613280753084, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.788732528686523, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8850575685501099, + "num_tokens": 385834674.0, + "step": 9936 + }, + { + "epoch": 1.264088538353899, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.793702125549316, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8822739720344543, + "num_tokens": 385877464.0, + "step": 9937 + }, + { + "epoch": 1.2642157486324894, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.863720893859863, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8869630098342896, + "num_tokens": 385916682.0, + "step": 9938 + }, + { + "epoch": 1.26434295891108, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.901395797729492, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8841681480407715, + "num_tokens": 385951509.0, + "step": 9939 + }, + { + "epoch": 1.2644701691896705, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.800958633422852, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8929622173309326, + "num_tokens": 385990468.0, + "step": 9940 + }, + { + "epoch": 1.264597379468261, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.988059997558594, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8731575012207031, + "num_tokens": 386023939.0, + "step": 9941 + }, + { + "epoch": 1.2647245897468515, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.829933166503906, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.883578896522522, + "num_tokens": 386065405.0, + "step": 9942 + }, + { + "epoch": 1.264851800025442, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.768573760986328, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8867130875587463, + "num_tokens": 386107678.0, + "step": 9943 + }, + { + "epoch": 1.2649790103040326, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.8460054397583, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.884323239326477, + "num_tokens": 386146097.0, + "step": 9944 + }, + { + "epoch": 1.2651062205826231, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.854524612426758, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8974646925926208, + "num_tokens": 386184721.0, + "step": 9945 + }, + { + "epoch": 1.2652334308612136, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.850237846374512, + "learning_rate": 1e-06, + "loss": 0.3576, + "mean_token_accuracy": 0.8764020204544067, + "num_tokens": 386226574.0, + "step": 9946 + }, + { + "epoch": 1.2653606411398042, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.950813293457031, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8725007772445679, + "num_tokens": 386263553.0, + "step": 9947 + }, + { + "epoch": 1.2654878514183947, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.844206809997559, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8771148324012756, + "num_tokens": 386300801.0, + "step": 9948 + }, + { + "epoch": 1.2656150616969852, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.951997756958008, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8756203651428223, + "num_tokens": 386338661.0, + "step": 9949 + }, + { + "epoch": 1.2657422719755758, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.763761520385742, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8927245140075684, + "num_tokens": 386381207.0, + "step": 9950 + }, + { + "epoch": 1.265869482254166, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.869439125061035, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8830108046531677, + "num_tokens": 386417736.0, + "step": 9951 + }, + { + "epoch": 1.2659966925327566, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.746549606323242, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8956542015075684, + "num_tokens": 386469260.0, + "step": 9952 + }, + { + "epoch": 1.2661239028113471, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.856046676635742, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8908071517944336, + "num_tokens": 386503532.0, + "step": 9953 + }, + { + "epoch": 1.2662511130899377, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.899020195007324, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8879363536834717, + "num_tokens": 386546530.0, + "step": 9954 + }, + { + "epoch": 1.2663783233685282, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.81424331665039, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8791055679321289, + "num_tokens": 386583610.0, + "step": 9955 + }, + { + "epoch": 1.2665055336471187, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.84759521484375, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8896020650863647, + "num_tokens": 386621172.0, + "step": 9956 + }, + { + "epoch": 1.2666327439257092, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.82634162902832, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8815740942955017, + "num_tokens": 386659011.0, + "step": 9957 + }, + { + "epoch": 1.2667599542042998, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.851311683654785, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.882235050201416, + "num_tokens": 386696533.0, + "step": 9958 + }, + { + "epoch": 1.2668871644828903, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.89890193939209, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.881151556968689, + "num_tokens": 386733518.0, + "step": 9959 + }, + { + "epoch": 1.2670143747614806, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.959638595581055, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8807020783424377, + "num_tokens": 386765896.0, + "step": 9960 + }, + { + "epoch": 1.2671415850400711, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.93709659576416, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8818578124046326, + "num_tokens": 386799854.0, + "step": 9961 + }, + { + "epoch": 1.2672687953186617, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.843865394592285, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8732311725616455, + "num_tokens": 386840276.0, + "step": 9962 + }, + { + "epoch": 1.2673960055972522, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.858686447143555, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8880178928375244, + "num_tokens": 386878323.0, + "step": 9963 + }, + { + "epoch": 1.2675232158758427, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.767284393310547, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8986773490905762, + "num_tokens": 386919620.0, + "step": 9964 + }, + { + "epoch": 1.2676504261544332, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.859785079956055, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8835203647613525, + "num_tokens": 386960472.0, + "step": 9965 + }, + { + "epoch": 1.2677776364330238, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.78175163269043, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8902833461761475, + "num_tokens": 387004396.0, + "step": 9966 + }, + { + "epoch": 1.2679048467116143, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.884072303771973, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8918672800064087, + "num_tokens": 387041790.0, + "step": 9967 + }, + { + "epoch": 1.2680320569902048, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.894892692565918, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.892612099647522, + "num_tokens": 387075883.0, + "step": 9968 + }, + { + "epoch": 1.2681592672687954, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.863736152648926, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8933203816413879, + "num_tokens": 387110574.0, + "step": 9969 + }, + { + "epoch": 1.2682864775473859, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.739428520202637, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8862050771713257, + "num_tokens": 387160142.0, + "step": 9970 + }, + { + "epoch": 1.2684136878259764, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.881702423095703, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8869954347610474, + "num_tokens": 387199725.0, + "step": 9971 + }, + { + "epoch": 1.268540898104567, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.815547943115234, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8871055245399475, + "num_tokens": 387246326.0, + "step": 9972 + }, + { + "epoch": 1.2686681083831575, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.872822761535645, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8960515856742859, + "num_tokens": 387282696.0, + "step": 9973 + }, + { + "epoch": 1.268795318661748, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.887969970703125, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8983131051063538, + "num_tokens": 387314838.0, + "step": 9974 + }, + { + "epoch": 1.2689225289403383, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.84671401977539, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8817025423049927, + "num_tokens": 387353917.0, + "step": 9975 + }, + { + "epoch": 1.2690497392189288, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.801127433776855, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8885200023651123, + "num_tokens": 387394240.0, + "step": 9976 + }, + { + "epoch": 1.2691769494975194, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.825596809387207, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8993393182754517, + "num_tokens": 387434338.0, + "step": 9977 + }, + { + "epoch": 1.2693041597761099, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.852049827575684, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8827859163284302, + "num_tokens": 387475710.0, + "step": 9978 + }, + { + "epoch": 1.2694313700547004, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.897477149963379, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8795077800750732, + "num_tokens": 387513779.0, + "step": 9979 + }, + { + "epoch": 1.269558580333291, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.792978286743164, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.891493022441864, + "num_tokens": 387556690.0, + "step": 9980 + }, + { + "epoch": 1.2696857906118815, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.823448181152344, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8947640657424927, + "num_tokens": 387597510.0, + "step": 9981 + }, + { + "epoch": 1.269813000890472, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.851425170898438, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.88582444190979, + "num_tokens": 387636892.0, + "step": 9982 + }, + { + "epoch": 1.2699402111690625, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.926250457763672, + "learning_rate": 1e-06, + "loss": 0.3943, + "mean_token_accuracy": 0.865439772605896, + "num_tokens": 387680109.0, + "step": 9983 + }, + { + "epoch": 1.2700674214476528, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.921754837036133, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8940186500549316, + "num_tokens": 387714137.0, + "step": 9984 + }, + { + "epoch": 1.2701946317262434, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.855567932128906, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8955097198486328, + "num_tokens": 387755009.0, + "step": 9985 + }, + { + "epoch": 1.270321842004834, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.817444801330566, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8950671553611755, + "num_tokens": 387792135.0, + "step": 9986 + }, + { + "epoch": 1.2704490522834244, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.938462257385254, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8914826512336731, + "num_tokens": 387828606.0, + "step": 9987 + }, + { + "epoch": 1.270576262562015, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.883516311645508, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8836321830749512, + "num_tokens": 387863222.0, + "step": 9988 + }, + { + "epoch": 1.2707034728406055, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.930462837219238, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8760513067245483, + "num_tokens": 387900741.0, + "step": 9989 + }, + { + "epoch": 1.270830683119196, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.770936012268066, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.884642481803894, + "num_tokens": 387944825.0, + "step": 9990 + }, + { + "epoch": 1.2709578933977865, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.910148620605469, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8974806070327759, + "num_tokens": 387977201.0, + "step": 9991 + }, + { + "epoch": 1.271085103676377, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.789133071899414, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8945877552032471, + "num_tokens": 388021012.0, + "step": 9992 + }, + { + "epoch": 1.2712123139549676, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.853096008300781, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8850184679031372, + "num_tokens": 388060477.0, + "step": 9993 + }, + { + "epoch": 1.2713395242335581, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.940832138061523, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8802340030670166, + "num_tokens": 388100266.0, + "step": 9994 + }, + { + "epoch": 1.2714667345121486, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.829007148742676, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8988347053527832, + "num_tokens": 388134386.0, + "step": 9995 + }, + { + "epoch": 1.2715939447907392, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.899280548095703, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8837774395942688, + "num_tokens": 388174270.0, + "step": 9996 + }, + { + "epoch": 1.2717211550693297, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.84259033203125, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8938623666763306, + "num_tokens": 388210138.0, + "step": 9997 + }, + { + "epoch": 1.2718483653479202, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.895341873168945, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8932777643203735, + "num_tokens": 388249492.0, + "step": 9998 + }, + { + "epoch": 1.2719755756265108, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.841604232788086, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8937913179397583, + "num_tokens": 388287176.0, + "step": 9999 + }, + { + "epoch": 1.272102785905101, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.928573608398438, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8761169910430908, + "num_tokens": 388323926.0, + "step": 10000 + }, + { + "epoch": 1.2722299961836916, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.97077751159668, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8832176923751831, + "num_tokens": 388358644.0, + "step": 10001 + }, + { + "epoch": 1.2723572064622821, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.74917221069336, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8924977779388428, + "num_tokens": 388405165.0, + "step": 10002 + }, + { + "epoch": 1.2724844167408726, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.801336288452148, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8915696144104004, + "num_tokens": 388445164.0, + "step": 10003 + }, + { + "epoch": 1.2726116270194632, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.957210540771484, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.888870120048523, + "num_tokens": 388478561.0, + "step": 10004 + }, + { + "epoch": 1.2727388372980537, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.90536117553711, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8813869953155518, + "num_tokens": 388515227.0, + "step": 10005 + }, + { + "epoch": 1.2728660475766442, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.792482376098633, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.884577214717865, + "num_tokens": 388555363.0, + "step": 10006 + }, + { + "epoch": 1.2729932578552348, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.990699768066406, + "learning_rate": 1e-06, + "loss": 0.3659, + "mean_token_accuracy": 0.8771331906318665, + "num_tokens": 388591059.0, + "step": 10007 + }, + { + "epoch": 1.2731204681338253, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.936443328857422, + "learning_rate": 1e-06, + "loss": 0.3926, + "mean_token_accuracy": 0.8664786219596863, + "num_tokens": 388628299.0, + "step": 10008 + }, + { + "epoch": 1.2732476784124156, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.888579368591309, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8807879686355591, + "num_tokens": 388666672.0, + "step": 10009 + }, + { + "epoch": 1.2733748886910061, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.824358940124512, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8829822540283203, + "num_tokens": 388704342.0, + "step": 10010 + }, + { + "epoch": 1.2735020989695967, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.829346656799316, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8896675109863281, + "num_tokens": 388743467.0, + "step": 10011 + }, + { + "epoch": 1.2736293092481872, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.836352348327637, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8847203254699707, + "num_tokens": 388783459.0, + "step": 10012 + }, + { + "epoch": 1.2737565195267777, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.825414657592773, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8804539442062378, + "num_tokens": 388825974.0, + "step": 10013 + }, + { + "epoch": 1.2738837298053682, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.872493743896484, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8857388496398926, + "num_tokens": 388865171.0, + "step": 10014 + }, + { + "epoch": 1.2740109400839588, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.857304573059082, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8864524364471436, + "num_tokens": 388906136.0, + "step": 10015 + }, + { + "epoch": 1.2741381503625493, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.880358695983887, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8958672881126404, + "num_tokens": 388940524.0, + "step": 10016 + }, + { + "epoch": 1.2742653606411398, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.897140502929688, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8786982297897339, + "num_tokens": 388979595.0, + "step": 10017 + }, + { + "epoch": 1.2743925709197303, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.816685676574707, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8947497010231018, + "num_tokens": 389019563.0, + "step": 10018 + }, + { + "epoch": 1.2745197811983209, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.858353614807129, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8990198373794556, + "num_tokens": 389054619.0, + "step": 10019 + }, + { + "epoch": 1.2746469914769114, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.882384300231934, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8858428001403809, + "num_tokens": 389093497.0, + "step": 10020 + }, + { + "epoch": 1.274774201755502, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.900545120239258, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8868973255157471, + "num_tokens": 389132494.0, + "step": 10021 + }, + { + "epoch": 1.2749014120340925, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.900313377380371, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8757354617118835, + "num_tokens": 389173277.0, + "step": 10022 + }, + { + "epoch": 1.275028622312683, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.864752769470215, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8854294419288635, + "num_tokens": 389211499.0, + "step": 10023 + }, + { + "epoch": 1.2751558325912733, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.93786907196045, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8841341137886047, + "num_tokens": 389246280.0, + "step": 10024 + }, + { + "epoch": 1.2752830428698638, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.833783149719238, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8897185325622559, + "num_tokens": 389287101.0, + "step": 10025 + }, + { + "epoch": 1.2754102531484544, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.919427871704102, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8813884854316711, + "num_tokens": 389323804.0, + "step": 10026 + }, + { + "epoch": 1.2755374634270449, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.8641939163208, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.878902792930603, + "num_tokens": 389367502.0, + "step": 10027 + }, + { + "epoch": 1.2756646737056354, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.914896965026855, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.892348051071167, + "num_tokens": 389403873.0, + "step": 10028 + }, + { + "epoch": 1.275791883984226, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.868030548095703, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8931799530982971, + "num_tokens": 389441526.0, + "step": 10029 + }, + { + "epoch": 1.2759190942628165, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.87317180633545, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8882866501808167, + "num_tokens": 389479710.0, + "step": 10030 + }, + { + "epoch": 1.276046304541407, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.790722846984863, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8966131210327148, + "num_tokens": 389522058.0, + "step": 10031 + }, + { + "epoch": 1.2761735148199975, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.055001258850098, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.895194411277771, + "num_tokens": 389562921.0, + "step": 10032 + }, + { + "epoch": 1.2763007250985878, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.850003242492676, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8933365345001221, + "num_tokens": 389600196.0, + "step": 10033 + }, + { + "epoch": 1.2764279353771784, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.857037544250488, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8730282783508301, + "num_tokens": 389639917.0, + "step": 10034 + }, + { + "epoch": 1.2765551456557689, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.841577529907227, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8942731022834778, + "num_tokens": 389678854.0, + "step": 10035 + }, + { + "epoch": 1.2766823559343594, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.807967185974121, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8885350823402405, + "num_tokens": 389721732.0, + "step": 10036 + }, + { + "epoch": 1.27680956621295, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.744529724121094e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.83584213256836, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8836630582809448, + "num_tokens": 389765412.0, + "step": 10037 + }, + { + "epoch": 1.2769367764915405, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.80366039276123, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8884516954421997, + "num_tokens": 389809703.0, + "step": 10038 + }, + { + "epoch": 1.277063986770131, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.824150085449219, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.9009494781494141, + "num_tokens": 389845885.0, + "step": 10039 + }, + { + "epoch": 1.2771911970487215, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.800949096679688, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8879778385162354, + "num_tokens": 389890201.0, + "step": 10040 + }, + { + "epoch": 1.277318407327312, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.957999229431152, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8843696117401123, + "num_tokens": 389924983.0, + "step": 10041 + }, + { + "epoch": 1.2774456176059026, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.915464401245117, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8787338733673096, + "num_tokens": 389959913.0, + "step": 10042 + }, + { + "epoch": 1.2775728278844931, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.834752082824707, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8801875114440918, + "num_tokens": 390003300.0, + "step": 10043 + }, + { + "epoch": 1.2777000381630836, + "ewc_loss": 0.0146484375, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 9.894371032714844e-06, + "grad_norm": 8.86651611328125, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.896233081817627, + "num_tokens": 390042300.0, + "step": 10044 + }, + { + "epoch": 1.2778272484416742, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.886052131652832, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8873396515846252, + "num_tokens": 390079525.0, + "step": 10045 + }, + { + "epoch": 1.2779544587202647, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.760002136230469, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9083379507064819, + "num_tokens": 390120389.0, + "step": 10046 + }, + { + "epoch": 1.2780816689988552, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 9.307517051696777, + "learning_rate": 1e-06, + "loss": 0.3867, + "mean_token_accuracy": 0.8701491951942444, + "num_tokens": 390160575.0, + "step": 10047 + }, + { + "epoch": 1.2782088792774458, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.897721290588379, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8892327547073364, + "num_tokens": 390198999.0, + "step": 10048 + }, + { + "epoch": 1.278336089556036, + "ewc_loss": 0.01458740234375, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 9.834766387939453e-06, + "grad_norm": 8.851649284362793, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8774378895759583, + "num_tokens": 390240322.0, + "step": 10049 + }, + { + "epoch": 1.2784632998346266, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.842854499816895, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.9069679975509644, + "num_tokens": 390282383.0, + "step": 10050 + }, + { + "epoch": 1.2785905101132171, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.89375114440918, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8806788921356201, + "num_tokens": 390319788.0, + "step": 10051 + }, + { + "epoch": 1.2787177203918076, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.88840103149414, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8863139152526855, + "num_tokens": 390358199.0, + "step": 10052 + }, + { + "epoch": 1.2788449306703982, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.917407989501953, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8847900032997131, + "num_tokens": 390393532.0, + "step": 10053 + }, + { + "epoch": 1.2789721409489887, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.882919311523438, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8822096586227417, + "num_tokens": 390434851.0, + "step": 10054 + }, + { + "epoch": 1.2790993512275792, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.004534721374512, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8794002532958984, + "num_tokens": 390477286.0, + "step": 10055 + }, + { + "epoch": 1.2792265615061698, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.904043197631836, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8814651966094971, + "num_tokens": 390512103.0, + "step": 10056 + }, + { + "epoch": 1.2793537717847603, + "ewc_loss": 0.01470947265625, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.853984832763672, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8881756663322449, + "num_tokens": 390547727.0, + "step": 10057 + }, + { + "epoch": 1.2794809820633506, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.803095817565918, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.9000288248062134, + "num_tokens": 390587795.0, + "step": 10058 + }, + { + "epoch": 1.2796081923419411, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.897308349609375, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8970767259597778, + "num_tokens": 390625686.0, + "step": 10059 + }, + { + "epoch": 1.2797354026205316, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.781344413757324, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.888885498046875, + "num_tokens": 390665469.0, + "step": 10060 + }, + { + "epoch": 1.2798626128991222, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.916881561279297, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8784002065658569, + "num_tokens": 390700710.0, + "step": 10061 + }, + { + "epoch": 1.2799898231777127, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.84221076965332, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8827794790267944, + "num_tokens": 390741339.0, + "step": 10062 + }, + { + "epoch": 1.2801170334563032, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.906786918640137, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8762167692184448, + "num_tokens": 390780396.0, + "step": 10063 + }, + { + "epoch": 1.2802442437348938, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.002740859985352, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8774023652076721, + "num_tokens": 390813772.0, + "step": 10064 + }, + { + "epoch": 1.2803714540134843, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.749947547912598, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8932708501815796, + "num_tokens": 390857097.0, + "step": 10065 + }, + { + "epoch": 1.2804986642920748, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.84593677520752, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8931170701980591, + "num_tokens": 390894715.0, + "step": 10066 + }, + { + "epoch": 1.2806258745706653, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.888761520385742, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8873156309127808, + "num_tokens": 390933831.0, + "step": 10067 + }, + { + "epoch": 1.2807530848492559, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.979557991027832, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8884650468826294, + "num_tokens": 390961692.0, + "step": 10068 + }, + { + "epoch": 1.2808802951278464, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.873077392578125, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8926304578781128, + "num_tokens": 391000897.0, + "step": 10069 + }, + { + "epoch": 1.281007505406437, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.821884155273438, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8822183609008789, + "num_tokens": 391045336.0, + "step": 10070 + }, + { + "epoch": 1.2811347156850275, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.911338806152344, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8828608393669128, + "num_tokens": 391085018.0, + "step": 10071 + }, + { + "epoch": 1.281261925963618, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.874433517456055, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8880279064178467, + "num_tokens": 391124315.0, + "step": 10072 + }, + { + "epoch": 1.2813891362422083, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.803781509399414, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8959176540374756, + "num_tokens": 391163438.0, + "step": 10073 + }, + { + "epoch": 1.2815163465207988, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.929954528808594, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8834569454193115, + "num_tokens": 391201487.0, + "step": 10074 + }, + { + "epoch": 1.2816435567993893, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.874007225036621, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.884148120880127, + "num_tokens": 391236957.0, + "step": 10075 + }, + { + "epoch": 1.2817707670779799, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.882065773010254, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8768669366836548, + "num_tokens": 391280325.0, + "step": 10076 + }, + { + "epoch": 1.2818979773565704, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.853907585144043, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8795535564422607, + "num_tokens": 391318401.0, + "step": 10077 + }, + { + "epoch": 1.282025187635161, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.831838607788086, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8866660594940186, + "num_tokens": 391355206.0, + "step": 10078 + }, + { + "epoch": 1.2821523979137515, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.943337440490723, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8844736814498901, + "num_tokens": 391388664.0, + "step": 10079 + }, + { + "epoch": 1.282279608192342, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.859864234924316, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8863861560821533, + "num_tokens": 391428186.0, + "step": 10080 + }, + { + "epoch": 1.2824068184709325, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.832040786743164, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8773707151412964, + "num_tokens": 391468428.0, + "step": 10081 + }, + { + "epoch": 1.2825340287495228, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.950668334960938, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8833293318748474, + "num_tokens": 391503781.0, + "step": 10082 + }, + { + "epoch": 1.2826612390281134, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.806209564208984, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8908547163009644, + "num_tokens": 391540627.0, + "step": 10083 + }, + { + "epoch": 1.2827884493067039, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.84754467010498, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8984067440032959, + "num_tokens": 391579494.0, + "step": 10084 + }, + { + "epoch": 1.2829156595852944, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.95868968963623, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8800784945487976, + "num_tokens": 391613176.0, + "step": 10085 + }, + { + "epoch": 1.283042869863885, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.872750282287598, + "learning_rate": 1e-06, + "loss": 0.3647, + "mean_token_accuracy": 0.8795430660247803, + "num_tokens": 391652436.0, + "step": 10086 + }, + { + "epoch": 1.2831700801424755, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.776220321655273, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.890405535697937, + "num_tokens": 391696462.0, + "step": 10087 + }, + { + "epoch": 1.283297290421066, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.929110527038574, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8822019100189209, + "num_tokens": 391736526.0, + "step": 10088 + }, + { + "epoch": 1.2834245006996565, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.91052532196045, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8898935914039612, + "num_tokens": 391771401.0, + "step": 10089 + }, + { + "epoch": 1.283551710978247, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.76837158203125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.785079002380371, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8882372379302979, + "num_tokens": 391811983.0, + "step": 10090 + }, + { + "epoch": 1.2836789212568376, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.863085746765137, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8956530094146729, + "num_tokens": 391844050.0, + "step": 10091 + }, + { + "epoch": 1.283806131535428, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.818746566772461, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8900082111358643, + "num_tokens": 391882121.0, + "step": 10092 + }, + { + "epoch": 1.2839333418140186, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.838977813720703, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8946682810783386, + "num_tokens": 391918426.0, + "step": 10093 + }, + { + "epoch": 1.2840605520926092, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.88979434967041, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.870789647102356, + "num_tokens": 391956506.0, + "step": 10094 + }, + { + "epoch": 1.2841877623711997, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.95069408416748, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8770962953567505, + "num_tokens": 391994822.0, + "step": 10095 + }, + { + "epoch": 1.2843149726497902, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.831491470336914, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8875122666358948, + "num_tokens": 392036614.0, + "step": 10096 + }, + { + "epoch": 1.2844421829283807, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.826746940612793, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8767478466033936, + "num_tokens": 392084464.0, + "step": 10097 + }, + { + "epoch": 1.284569393206971, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.9478759765625, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8763240575790405, + "num_tokens": 392123431.0, + "step": 10098 + }, + { + "epoch": 1.2846966034855616, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.890917778015137, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8773958086967468, + "num_tokens": 392164971.0, + "step": 10099 + }, + { + "epoch": 1.2848238137641521, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.965828895568848, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.889758288860321, + "num_tokens": 392198416.0, + "step": 10100 + }, + { + "epoch": 1.2849510240427426, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.798821449279785, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8922653198242188, + "num_tokens": 392240603.0, + "step": 10101 + }, + { + "epoch": 1.2850782343213332, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.897971153259277, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8773694038391113, + "num_tokens": 392282559.0, + "step": 10102 + }, + { + "epoch": 1.2852054445999237, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.812382698059082, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8886592388153076, + "num_tokens": 392326617.0, + "step": 10103 + }, + { + "epoch": 1.2853326548785142, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.991037368774414, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8757069706916809, + "num_tokens": 392360943.0, + "step": 10104 + }, + { + "epoch": 1.2854598651571048, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.934675216674805, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.876419723033905, + "num_tokens": 392398227.0, + "step": 10105 + }, + { + "epoch": 1.2855870754356953, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.825194358825684, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8956303596496582, + "num_tokens": 392438590.0, + "step": 10106 + }, + { + "epoch": 1.2857142857142856, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.87436580657959, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8796576261520386, + "num_tokens": 392484349.0, + "step": 10107 + }, + { + "epoch": 1.2858414959928761, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.861177444458008, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8855140209197998, + "num_tokens": 392522375.0, + "step": 10108 + }, + { + "epoch": 1.2859687062714666, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.014298439025879, + "learning_rate": 1e-06, + "loss": 0.3796, + "mean_token_accuracy": 0.8731163740158081, + "num_tokens": 392556858.0, + "step": 10109 + }, + { + "epoch": 1.2860959165500572, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.855423927307129, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8851979970932007, + "num_tokens": 392598553.0, + "step": 10110 + }, + { + "epoch": 1.2862231268286477, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.848857879638672, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8876224756240845, + "num_tokens": 392639678.0, + "step": 10111 + }, + { + "epoch": 1.2863503371072382, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.106789588928223, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8777956962585449, + "num_tokens": 392670083.0, + "step": 10112 + }, + { + "epoch": 1.2864775473858288, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.964643478393555, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8846485614776611, + "num_tokens": 392706619.0, + "step": 10113 + }, + { + "epoch": 1.2866047576644193, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.841876029968262, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8931397199630737, + "num_tokens": 392743307.0, + "step": 10114 + }, + { + "epoch": 1.2867319679430098, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.848758697509766, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.896615743637085, + "num_tokens": 392784016.0, + "step": 10115 + }, + { + "epoch": 1.2868591782216003, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.852788925170898, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8824036121368408, + "num_tokens": 392824528.0, + "step": 10116 + }, + { + "epoch": 1.2869863885001909, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.87952995300293, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8929997086524963, + "num_tokens": 392863061.0, + "step": 10117 + }, + { + "epoch": 1.2871135987787814, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.798615455627441, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9041318893432617, + "num_tokens": 392899842.0, + "step": 10118 + }, + { + "epoch": 1.287240809057372, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.93928050994873, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8777469396591187, + "num_tokens": 392938599.0, + "step": 10119 + }, + { + "epoch": 1.2873680193359625, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.803447723388672, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8858165740966797, + "num_tokens": 392982311.0, + "step": 10120 + }, + { + "epoch": 1.287495229614553, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.954599380493164, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8824260234832764, + "num_tokens": 393014975.0, + "step": 10121 + }, + { + "epoch": 1.2876224398931433, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.919569969177246, + "learning_rate": 1e-06, + "loss": 0.3776, + "mean_token_accuracy": 0.871492862701416, + "num_tokens": 393053153.0, + "step": 10122 + }, + { + "epoch": 1.2877496501717338, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.792213439941406e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.844573020935059, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8878268003463745, + "num_tokens": 393090576.0, + "step": 10123 + }, + { + "epoch": 1.2878768604503243, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.88393497467041, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8849977254867554, + "num_tokens": 393126530.0, + "step": 10124 + }, + { + "epoch": 1.2880040707289149, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.868711471557617, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8816450238227844, + "num_tokens": 393166573.0, + "step": 10125 + }, + { + "epoch": 1.2881312810075054, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.898026466369629, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.887666642665863, + "num_tokens": 393207983.0, + "step": 10126 + }, + { + "epoch": 1.288258491286096, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.85333251953125, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.880111575126648, + "num_tokens": 393253556.0, + "step": 10127 + }, + { + "epoch": 1.2883857015646865, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.837850570678711, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8962510228157043, + "num_tokens": 393298306.0, + "step": 10128 + }, + { + "epoch": 1.288512911843277, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.907114028930664, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8759963512420654, + "num_tokens": 393338477.0, + "step": 10129 + }, + { + "epoch": 1.2886401221218675, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.888263702392578, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8717283606529236, + "num_tokens": 393378230.0, + "step": 10130 + }, + { + "epoch": 1.2887673324004578, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.999526977539062, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8823537826538086, + "num_tokens": 393412320.0, + "step": 10131 + }, + { + "epoch": 1.2888945426790483, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.818609237670898, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8861489295959473, + "num_tokens": 393452508.0, + "step": 10132 + }, + { + "epoch": 1.2890217529576389, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.872998237609863, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8843172788619995, + "num_tokens": 393492663.0, + "step": 10133 + }, + { + "epoch": 1.2891489632362294, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.760769844055176, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8992406725883484, + "num_tokens": 393530616.0, + "step": 10134 + }, + { + "epoch": 1.28927617351482, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.785211563110352, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8943328857421875, + "num_tokens": 393569832.0, + "step": 10135 + }, + { + "epoch": 1.2894033837934105, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.924166679382324, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8925679326057434, + "num_tokens": 393604413.0, + "step": 10136 + }, + { + "epoch": 1.289530594072001, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.834887504577637, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8900743722915649, + "num_tokens": 393648232.0, + "step": 10137 + }, + { + "epoch": 1.2896578043505915, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.911871910095215, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8798997402191162, + "num_tokens": 393686468.0, + "step": 10138 + }, + { + "epoch": 1.289785014629182, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.904555320739746, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8920113444328308, + "num_tokens": 393722931.0, + "step": 10139 + }, + { + "epoch": 1.2899122249077726, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.856945991516113, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8823896050453186, + "num_tokens": 393762442.0, + "step": 10140 + }, + { + "epoch": 1.290039435186363, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.956430435180664, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8817495703697205, + "num_tokens": 393799760.0, + "step": 10141 + }, + { + "epoch": 1.2901666454649536, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.924417495727539, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8798395395278931, + "num_tokens": 393833684.0, + "step": 10142 + }, + { + "epoch": 1.2902938557435442, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.981017112731934, + "learning_rate": 1e-06, + "loss": 0.4158, + "mean_token_accuracy": 0.8579884767532349, + "num_tokens": 393872592.0, + "step": 10143 + }, + { + "epoch": 1.2904210660221347, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.845420837402344, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8719805479049683, + "num_tokens": 393914730.0, + "step": 10144 + }, + { + "epoch": 1.2905482763007252, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.851594924926758, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8819281458854675, + "num_tokens": 393957514.0, + "step": 10145 + }, + { + "epoch": 1.2906754865793157, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.824060440063477, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8869551420211792, + "num_tokens": 393997697.0, + "step": 10146 + }, + { + "epoch": 1.290802696857906, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.89756965637207, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.891663134098053, + "num_tokens": 394038144.0, + "step": 10147 + }, + { + "epoch": 1.2909299071364966, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.8821439743042, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8936556577682495, + "num_tokens": 394072404.0, + "step": 10148 + }, + { + "epoch": 1.291057117415087, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.83052921295166, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8869685530662537, + "num_tokens": 394113172.0, + "step": 10149 + }, + { + "epoch": 1.2911843276936776, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.015754699707031, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8734689950942993, + "num_tokens": 394145887.0, + "step": 10150 + }, + { + "epoch": 1.2913115379722682, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.8285493850708, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8776463270187378, + "num_tokens": 394188382.0, + "step": 10151 + }, + { + "epoch": 1.2914387482508587, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.958640098571777, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8837085962295532, + "num_tokens": 394229596.0, + "step": 10152 + }, + { + "epoch": 1.2915659585294492, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.837485313415527, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8909662365913391, + "num_tokens": 394266244.0, + "step": 10153 + }, + { + "epoch": 1.2916931688080397, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.901705741882324, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8863582015037537, + "num_tokens": 394304002.0, + "step": 10154 + }, + { + "epoch": 1.2918203790866303, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.919726371765137, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8946000337600708, + "num_tokens": 394340182.0, + "step": 10155 + }, + { + "epoch": 1.2919475893652206, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.890664100646973, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8857866525650024, + "num_tokens": 394380013.0, + "step": 10156 + }, + { + "epoch": 1.2920747996438111, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.938507080078125, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8838417530059814, + "num_tokens": 394412891.0, + "step": 10157 + }, + { + "epoch": 1.2922020099224016, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.885912895202637, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8783104419708252, + "num_tokens": 394453064.0, + "step": 10158 + }, + { + "epoch": 1.2923292202009922, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.935418128967285, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8928756713867188, + "num_tokens": 394487783.0, + "step": 10159 + }, + { + "epoch": 1.2924564304795827, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.903075218200684, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.9014873504638672, + "num_tokens": 394522977.0, + "step": 10160 + }, + { + "epoch": 1.2925836407581732, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.82234001159668, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8865594267845154, + "num_tokens": 394564042.0, + "step": 10161 + }, + { + "epoch": 1.2927108510367638, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.932281494140625, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8932178616523743, + "num_tokens": 394602463.0, + "step": 10162 + }, + { + "epoch": 1.2928380613153543, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.874536514282227, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.884369969367981, + "num_tokens": 394644962.0, + "step": 10163 + }, + { + "epoch": 1.2929652715939448, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.975350379943848, + "learning_rate": 1e-06, + "loss": 0.3771, + "mean_token_accuracy": 0.8740488886833191, + "num_tokens": 394685286.0, + "step": 10164 + }, + { + "epoch": 1.2930924818725353, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.050043106079102, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8778030872344971, + "num_tokens": 394715806.0, + "step": 10165 + }, + { + "epoch": 1.2932196921511259, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.850150108337402, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8947380781173706, + "num_tokens": 394756798.0, + "step": 10166 + }, + { + "epoch": 1.2933469024297164, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.91423225402832, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.884326696395874, + "num_tokens": 394794250.0, + "step": 10167 + }, + { + "epoch": 1.293474112708307, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.899958610534668, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8890067934989929, + "num_tokens": 394830495.0, + "step": 10168 + }, + { + "epoch": 1.2936013229868975, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.840642929077148, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8825896382331848, + "num_tokens": 394870527.0, + "step": 10169 + }, + { + "epoch": 1.293728533265488, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.83611011505127, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.900427520275116, + "num_tokens": 394908498.0, + "step": 10170 + }, + { + "epoch": 1.2938557435440783, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.89271354675293, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8832522034645081, + "num_tokens": 394948610.0, + "step": 10171 + }, + { + "epoch": 1.2939829538226688, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.892364501953125, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.88034987449646, + "num_tokens": 394989830.0, + "step": 10172 + }, + { + "epoch": 1.2941101641012593, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.986953735351562, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.8704777956008911, + "num_tokens": 395022676.0, + "step": 10173 + }, + { + "epoch": 1.2942373743798499, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.854001998901367, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.899250864982605, + "num_tokens": 395060995.0, + "step": 10174 + }, + { + "epoch": 1.2943645846584404, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.902018547058105, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8807324171066284, + "num_tokens": 395101505.0, + "step": 10175 + }, + { + "epoch": 1.294491794937031, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.859892845153809, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8922141790390015, + "num_tokens": 395141872.0, + "step": 10176 + }, + { + "epoch": 1.2946190052156215, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.882721900939941, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8854503631591797, + "num_tokens": 395180865.0, + "step": 10177 + }, + { + "epoch": 1.294746215494212, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.846467971801758, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.9006112813949585, + "num_tokens": 395219639.0, + "step": 10178 + }, + { + "epoch": 1.2948734257728025, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.826620101928711, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8886151909828186, + "num_tokens": 395267049.0, + "step": 10179 + }, + { + "epoch": 1.2950006360513928, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.899614334106445, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.9005018472671509, + "num_tokens": 395302902.0, + "step": 10180 + }, + { + "epoch": 1.2951278463299833, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.905997276306152, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.888221800327301, + "num_tokens": 395339460.0, + "step": 10181 + }, + { + "epoch": 1.2952550566085739, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.809868812561035, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8910817503929138, + "num_tokens": 395386967.0, + "step": 10182 + }, + { + "epoch": 1.2953822668871644, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.87742805480957, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8871722221374512, + "num_tokens": 395426327.0, + "step": 10183 + }, + { + "epoch": 1.295509477165755, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.874567985534668, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8945850133895874, + "num_tokens": 395466910.0, + "step": 10184 + }, + { + "epoch": 1.2956366874443455, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.940109252929688, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8894907832145691, + "num_tokens": 395506740.0, + "step": 10185 + }, + { + "epoch": 1.295763897722936, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.878410339355469, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8928866982460022, + "num_tokens": 395545601.0, + "step": 10186 + }, + { + "epoch": 1.2958911080015265, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.92694091796875, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8791712522506714, + "num_tokens": 395585159.0, + "step": 10187 + }, + { + "epoch": 1.296018318280117, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.013931274414062, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8827061653137207, + "num_tokens": 395622096.0, + "step": 10188 + }, + { + "epoch": 1.2961455285587076, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.873224258422852, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8936806917190552, + "num_tokens": 395660723.0, + "step": 10189 + }, + { + "epoch": 1.296272738837298, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.90747356414795, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8873960971832275, + "num_tokens": 395698017.0, + "step": 10190 + }, + { + "epoch": 1.2963999491158886, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.85075569152832, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.89203941822052, + "num_tokens": 395742866.0, + "step": 10191 + }, + { + "epoch": 1.2965271593944792, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.8160552978515625e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.849390029907227, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8985635042190552, + "num_tokens": 395778868.0, + "step": 10192 + }, + { + "epoch": 1.2966543696730697, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.889511108398438, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8910378217697144, + "num_tokens": 395820225.0, + "step": 10193 + }, + { + "epoch": 1.2967815799516602, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.944315910339355, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.879336953163147, + "num_tokens": 395860789.0, + "step": 10194 + }, + { + "epoch": 1.2969087902302507, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.931066513061523, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8853775262832642, + "num_tokens": 395898244.0, + "step": 10195 + }, + { + "epoch": 1.297036000508841, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.930747032165527, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8843132257461548, + "num_tokens": 395935837.0, + "step": 10196 + }, + { + "epoch": 1.2971632107874316, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.940364837646484, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8888492584228516, + "num_tokens": 395973691.0, + "step": 10197 + }, + { + "epoch": 1.297290421066022, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.826234817504883, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8951249122619629, + "num_tokens": 396017862.0, + "step": 10198 + }, + { + "epoch": 1.2974176313446126, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.883525848388672, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.895332932472229, + "num_tokens": 396054740.0, + "step": 10199 + }, + { + "epoch": 1.2975448416232032, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.864201545715332, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8944272994995117, + "num_tokens": 396093541.0, + "step": 10200 + }, + { + "epoch": 1.2976720519017937, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.974358558654785, + "learning_rate": 1e-06, + "loss": 0.361, + "mean_token_accuracy": 0.8775720596313477, + "num_tokens": 396133392.0, + "step": 10201 + }, + { + "epoch": 1.2977992621803842, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.924944877624512, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8844342231750488, + "num_tokens": 396174416.0, + "step": 10202 + }, + { + "epoch": 1.2979264724589747, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.89545726776123, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8875944018363953, + "num_tokens": 396216334.0, + "step": 10203 + }, + { + "epoch": 1.2980536827375653, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.988844871520996, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8968002200126648, + "num_tokens": 396248845.0, + "step": 10204 + }, + { + "epoch": 1.2981808930161556, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.934576034545898, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8955919146537781, + "num_tokens": 396285529.0, + "step": 10205 + }, + { + "epoch": 1.298308103294746, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.951141357421875, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8826576471328735, + "num_tokens": 396321319.0, + "step": 10206 + }, + { + "epoch": 1.2984353135733366, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.950096130371094, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8785953521728516, + "num_tokens": 396359640.0, + "step": 10207 + }, + { + "epoch": 1.2985625238519272, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.910772323608398, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8844022750854492, + "num_tokens": 396397192.0, + "step": 10208 + }, + { + "epoch": 1.2986897341305177, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.880367279052734, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8897353410720825, + "num_tokens": 396437175.0, + "step": 10209 + }, + { + "epoch": 1.2988169444091082, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.897649765014648, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8876159191131592, + "num_tokens": 396477592.0, + "step": 10210 + }, + { + "epoch": 1.2989441546876987, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.0086088180542, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8761452436447144, + "num_tokens": 396511583.0, + "step": 10211 + }, + { + "epoch": 1.2990713649662893, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.912943840026855, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8887195587158203, + "num_tokens": 396549378.0, + "step": 10212 + }, + { + "epoch": 1.2991985752448798, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.895838737487793, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.889427661895752, + "num_tokens": 396590086.0, + "step": 10213 + }, + { + "epoch": 1.2993257855234703, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.894791603088379, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8840314745903015, + "num_tokens": 396632671.0, + "step": 10214 + }, + { + "epoch": 1.2994529958020609, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 9.029828071594238, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8892218470573425, + "num_tokens": 396666377.0, + "step": 10215 + }, + { + "epoch": 1.2995802060806514, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.914727210998535, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8990087509155273, + "num_tokens": 396704580.0, + "step": 10216 + }, + { + "epoch": 1.299707416359242, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.909846305847168, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8792194128036499, + "num_tokens": 396743023.0, + "step": 10217 + }, + { + "epoch": 1.2998346266378324, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.122176170349121, + "learning_rate": 1e-06, + "loss": 0.3849, + "mean_token_accuracy": 0.8648995161056519, + "num_tokens": 396773554.0, + "step": 10218 + }, + { + "epoch": 1.299961836916423, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.904518127441406, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8866636753082275, + "num_tokens": 396813625.0, + "step": 10219 + }, + { + "epoch": 1.3000890471950133, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.927688598632812, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8855120539665222, + "num_tokens": 396848934.0, + "step": 10220 + }, + { + "epoch": 1.3002162574736038, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.912697792053223, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.889326810836792, + "num_tokens": 396890332.0, + "step": 10221 + }, + { + "epoch": 1.3003434677521943, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.925202369689941, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8777210712432861, + "num_tokens": 396932880.0, + "step": 10222 + }, + { + "epoch": 1.3004706780307849, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.870939254760742, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8806429505348206, + "num_tokens": 396974729.0, + "step": 10223 + }, + { + "epoch": 1.3005978883093754, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.975713729858398, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.890038013458252, + "num_tokens": 397010268.0, + "step": 10224 + }, + { + "epoch": 1.300725098587966, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.844732284545898, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8964415788650513, + "num_tokens": 397052867.0, + "step": 10225 + }, + { + "epoch": 1.3008523088665565, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.852596282958984, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8988975286483765, + "num_tokens": 397093367.0, + "step": 10226 + }, + { + "epoch": 1.300979519145147, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.817490577697754, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8896468877792358, + "num_tokens": 397134160.0, + "step": 10227 + }, + { + "epoch": 1.3011067294237375, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.908447265625, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8880385756492615, + "num_tokens": 397173520.0, + "step": 10228 + }, + { + "epoch": 1.3012339397023278, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.996488571166992, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8775445222854614, + "num_tokens": 397208135.0, + "step": 10229 + }, + { + "epoch": 1.3013611499809183, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.939047813415527, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8849329352378845, + "num_tokens": 397242426.0, + "step": 10230 + }, + { + "epoch": 1.3014883602595089, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.872323036193848, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8951632976531982, + "num_tokens": 397279722.0, + "step": 10231 + }, + { + "epoch": 1.3016155705380994, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.884435653686523, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9024544358253479, + "num_tokens": 397316748.0, + "step": 10232 + }, + { + "epoch": 1.30174278081669, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.897762298583984, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8796392679214478, + "num_tokens": 397356677.0, + "step": 10233 + }, + { + "epoch": 1.3018699910952805, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.84744644165039, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8822565078735352, + "num_tokens": 397404050.0, + "step": 10234 + }, + { + "epoch": 1.301997201373871, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.966757774353027, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8756566643714905, + "num_tokens": 397439857.0, + "step": 10235 + }, + { + "epoch": 1.3021244116524615, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.808290481567383, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.904044508934021, + "num_tokens": 397479094.0, + "step": 10236 + }, + { + "epoch": 1.302251621931052, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.839897155761719e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.832959175109863, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8737735152244568, + "num_tokens": 397528756.0, + "step": 10237 + }, + { + "epoch": 1.3023788322096426, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.915629386901855, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.888956606388092, + "num_tokens": 397566433.0, + "step": 10238 + }, + { + "epoch": 1.302506042488233, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.780749320983887, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8944739103317261, + "num_tokens": 397609237.0, + "step": 10239 + }, + { + "epoch": 1.3026332527668236, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.830802917480469, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9009305238723755, + "num_tokens": 397651155.0, + "step": 10240 + }, + { + "epoch": 1.3027604630454142, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.932403564453125, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8817113041877747, + "num_tokens": 397687731.0, + "step": 10241 + }, + { + "epoch": 1.3028876733240047, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.894181251525879, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8875239491462708, + "num_tokens": 397725302.0, + "step": 10242 + }, + { + "epoch": 1.3030148836025952, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.820335388183594, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8981247544288635, + "num_tokens": 397767006.0, + "step": 10243 + }, + { + "epoch": 1.3031420938811857, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.955291748046875, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8728638887405396, + "num_tokens": 397803065.0, + "step": 10244 + }, + { + "epoch": 1.303269304159776, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.927165031433105, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.8793081045150757, + "num_tokens": 397843860.0, + "step": 10245 + }, + { + "epoch": 1.3033965144383666, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.927459716796875, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8912882804870605, + "num_tokens": 397883731.0, + "step": 10246 + }, + { + "epoch": 1.303523724716957, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.93423080444336, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.879709005355835, + "num_tokens": 397925415.0, + "step": 10247 + }, + { + "epoch": 1.3036509349955476, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.916740417480469, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8930946588516235, + "num_tokens": 397962376.0, + "step": 10248 + }, + { + "epoch": 1.3037781452741382, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.827902793884277, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.9020922780036926, + "num_tokens": 398001917.0, + "step": 10249 + }, + { + "epoch": 1.3039053555527287, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.984868049621582, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8932881951332092, + "num_tokens": 398044529.0, + "step": 10250 + }, + { + "epoch": 1.3040325658313192, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.873309135437012, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8765400648117065, + "num_tokens": 398087589.0, + "step": 10251 + }, + { + "epoch": 1.3041597761099097, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 9.032891273498535, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8900594711303711, + "num_tokens": 398124400.0, + "step": 10252 + }, + { + "epoch": 1.3042869863885003, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.00111198425293, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8792540431022644, + "num_tokens": 398159410.0, + "step": 10253 + }, + { + "epoch": 1.3044141966670906, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.86243724822998, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8869722485542297, + "num_tokens": 398202217.0, + "step": 10254 + }, + { + "epoch": 1.304541406945681, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.910201072692871, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8875917196273804, + "num_tokens": 398240439.0, + "step": 10255 + }, + { + "epoch": 1.3046686172242716, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.005485534667969, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8761358857154846, + "num_tokens": 398276457.0, + "step": 10256 + }, + { + "epoch": 1.3047958275028622, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.954462051391602, + "learning_rate": 1e-06, + "loss": 0.3733, + "mean_token_accuracy": 0.870193600654602, + "num_tokens": 398317585.0, + "step": 10257 + }, + { + "epoch": 1.3049230377814527, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.872394561767578, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8901095986366272, + "num_tokens": 398358831.0, + "step": 10258 + }, + { + "epoch": 1.3050502480600432, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.957767486572266, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8903601765632629, + "num_tokens": 398395433.0, + "step": 10259 + }, + { + "epoch": 1.3051774583386337, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.931520462036133, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8808706402778625, + "num_tokens": 398435244.0, + "step": 10260 + }, + { + "epoch": 1.3053046686172243, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.938913345336914, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8781077265739441, + "num_tokens": 398473995.0, + "step": 10261 + }, + { + "epoch": 1.3054318788958148, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.844657897949219, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8923647403717041, + "num_tokens": 398514614.0, + "step": 10262 + }, + { + "epoch": 1.3055590891744053, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.895012855529785, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8790368437767029, + "num_tokens": 398559315.0, + "step": 10263 + }, + { + "epoch": 1.3056862994529959, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.974027633666992, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8860356211662292, + "num_tokens": 398593388.0, + "step": 10264 + }, + { + "epoch": 1.3058135097315864, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.871874809265137, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8764468431472778, + "num_tokens": 398640628.0, + "step": 10265 + }, + { + "epoch": 1.305940720010177, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.798935890197754, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.9018296003341675, + "num_tokens": 398686165.0, + "step": 10266 + }, + { + "epoch": 1.3060679302887674, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.922202110290527, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8906871676445007, + "num_tokens": 398724366.0, + "step": 10267 + }, + { + "epoch": 1.306195140567358, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.991988182067871, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.9002390503883362, + "num_tokens": 398758646.0, + "step": 10268 + }, + { + "epoch": 1.3063223508459483, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.969917297363281, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8764706254005432, + "num_tokens": 398796083.0, + "step": 10269 + }, + { + "epoch": 1.3064495611245388, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.977676391601562, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8924264907836914, + "num_tokens": 398827219.0, + "step": 10270 + }, + { + "epoch": 1.3065767714031293, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.872008323669434, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8882851600646973, + "num_tokens": 398864916.0, + "step": 10271 + }, + { + "epoch": 1.3067039816817199, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.883106231689453, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8823473453521729, + "num_tokens": 398908336.0, + "step": 10272 + }, + { + "epoch": 1.3068311919603104, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.874792098999023, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8854489922523499, + "num_tokens": 398946680.0, + "step": 10273 + }, + { + "epoch": 1.306958402238901, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.867171287536621, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8865330219268799, + "num_tokens": 398989820.0, + "step": 10274 + }, + { + "epoch": 1.3070856125174914, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.857213020324707, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8899419903755188, + "num_tokens": 399037319.0, + "step": 10275 + }, + { + "epoch": 1.307212822796082, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.032519340515137, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.89061039686203, + "num_tokens": 399071324.0, + "step": 10276 + }, + { + "epoch": 1.3073400330746725, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.937625885009766, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8831548690795898, + "num_tokens": 399106425.0, + "step": 10277 + }, + { + "epoch": 1.3074672433532628, + "ewc_loss": 0.0147705078125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.8412504196167, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8787202835083008, + "num_tokens": 399147817.0, + "step": 10278 + }, + { + "epoch": 1.3075944536318533, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.92404842376709, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8832851052284241, + "num_tokens": 399190759.0, + "step": 10279 + }, + { + "epoch": 1.3077216639104439, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.044960021972656, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8878597021102905, + "num_tokens": 399223257.0, + "step": 10280 + }, + { + "epoch": 1.3078488741890344, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.909826278686523, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8821858167648315, + "num_tokens": 399264924.0, + "step": 10281 + }, + { + "epoch": 1.307976084467625, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.873600959777832, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8883913159370422, + "num_tokens": 399305695.0, + "step": 10282 + }, + { + "epoch": 1.3081032947462155, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.943439483642578, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8847663402557373, + "num_tokens": 399344660.0, + "step": 10283 + }, + { + "epoch": 1.308230505024806, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.871868133544922, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8970044851303101, + "num_tokens": 399383210.0, + "step": 10284 + }, + { + "epoch": 1.3083577153033965, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.918957710266113, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.882718563079834, + "num_tokens": 399421609.0, + "step": 10285 + }, + { + "epoch": 1.308484925581987, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.899328231811523, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8833151459693909, + "num_tokens": 399462523.0, + "step": 10286 + }, + { + "epoch": 1.3086121358605776, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.823112487792969, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8990350365638733, + "num_tokens": 399505615.0, + "step": 10287 + }, + { + "epoch": 1.308739346139168, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.99328899383545, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8813635110855103, + "num_tokens": 399542469.0, + "step": 10288 + }, + { + "epoch": 1.3088665564177586, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.907513618469238, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8810684680938721, + "num_tokens": 399578572.0, + "step": 10289 + }, + { + "epoch": 1.3089937666963491, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.866737365722656, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8879032135009766, + "num_tokens": 399619484.0, + "step": 10290 + }, + { + "epoch": 1.3091209769749397, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.856011390686035, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9011059999465942, + "num_tokens": 399658749.0, + "step": 10291 + }, + { + "epoch": 1.3092481872535302, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.858052253723145, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9031368494033813, + "num_tokens": 399694651.0, + "step": 10292 + }, + { + "epoch": 1.3093753975321207, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.857970237731934, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8982402086257935, + "num_tokens": 399738370.0, + "step": 10293 + }, + { + "epoch": 1.309502607810711, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.887938499450684, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8861137628555298, + "num_tokens": 399780352.0, + "step": 10294 + }, + { + "epoch": 1.3096298180893016, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.917263984680176, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8975836038589478, + "num_tokens": 399817727.0, + "step": 10295 + }, + { + "epoch": 1.309757028367892, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.852399826049805, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.900087833404541, + "num_tokens": 399858714.0, + "step": 10296 + }, + { + "epoch": 1.3098842386464826, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.842531204223633, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8888267278671265, + "num_tokens": 399900795.0, + "step": 10297 + }, + { + "epoch": 1.3100114489250732, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.925639152526855, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.9028828740119934, + "num_tokens": 399933312.0, + "step": 10298 + }, + { + "epoch": 1.3101386592036637, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.863739013671875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.965418815612793, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8883679509162903, + "num_tokens": 399969786.0, + "step": 10299 + }, + { + "epoch": 1.3102658694822542, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.92835521697998, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.9015530347824097, + "num_tokens": 400004606.0, + "step": 10300 + }, + { + "epoch": 1.3103930797608447, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.950993537902832, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8955729007720947, + "num_tokens": 400043269.0, + "step": 10301 + }, + { + "epoch": 1.3105202900394353, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.947876930236816, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8957066535949707, + "num_tokens": 400080598.0, + "step": 10302 + }, + { + "epoch": 1.3106475003180256, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.878676414489746, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8910166621208191, + "num_tokens": 400120702.0, + "step": 10303 + }, + { + "epoch": 1.310774710596616, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.86860179901123, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8976745009422302, + "num_tokens": 400162368.0, + "step": 10304 + }, + { + "epoch": 1.3109019208752066, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.844742774963379, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8848658800125122, + "num_tokens": 400207065.0, + "step": 10305 + }, + { + "epoch": 1.3110291311537972, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.929524421691895, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8772482872009277, + "num_tokens": 400251599.0, + "step": 10306 + }, + { + "epoch": 1.3111563414323877, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.903407096862793, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8909945487976074, + "num_tokens": 400289927.0, + "step": 10307 + }, + { + "epoch": 1.3112835517109782, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 9.009605407714844, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8844116926193237, + "num_tokens": 400324255.0, + "step": 10308 + }, + { + "epoch": 1.3114107619895687, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.94244384765625, + "learning_rate": 1e-06, + "loss": 0.3947, + "mean_token_accuracy": 0.8680554032325745, + "num_tokens": 400366914.0, + "step": 10309 + }, + { + "epoch": 1.3115379722681593, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.884711265563965, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8818005323410034, + "num_tokens": 400408485.0, + "step": 10310 + }, + { + "epoch": 1.3116651825467498, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.96435832977295, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8799169659614563, + "num_tokens": 400443588.0, + "step": 10311 + }, + { + "epoch": 1.3117923928253403, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.867761611938477, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8932269811630249, + "num_tokens": 400484034.0, + "step": 10312 + }, + { + "epoch": 1.3119196031039309, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.871594429016113, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8870183825492859, + "num_tokens": 400525597.0, + "step": 10313 + }, + { + "epoch": 1.3120468133825214, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.882743835449219, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8812888264656067, + "num_tokens": 400566406.0, + "step": 10314 + }, + { + "epoch": 1.312174023661112, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.859513282775879, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8882550597190857, + "num_tokens": 400608508.0, + "step": 10315 + }, + { + "epoch": 1.3123012339397024, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.868436813354492, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8816452026367188, + "num_tokens": 400649525.0, + "step": 10316 + }, + { + "epoch": 1.312428444218293, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.959517478942871, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8813037276268005, + "num_tokens": 400687486.0, + "step": 10317 + }, + { + "epoch": 1.3125556544968833, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.85123062133789, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8952299356460571, + "num_tokens": 400721056.0, + "step": 10318 + }, + { + "epoch": 1.3126828647754738, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.949711799621582, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.887927770614624, + "num_tokens": 400756127.0, + "step": 10319 + }, + { + "epoch": 1.3128100750540643, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.978450775146484, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8875771760940552, + "num_tokens": 400789549.0, + "step": 10320 + }, + { + "epoch": 1.3129372853326549, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.851825714111328, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8855434656143188, + "num_tokens": 400833191.0, + "step": 10321 + }, + { + "epoch": 1.3130644956112454, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.996313095092773, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8871533870697021, + "num_tokens": 400867965.0, + "step": 10322 + }, + { + "epoch": 1.313191705889836, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.871459007263184, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8943326473236084, + "num_tokens": 400907831.0, + "step": 10323 + }, + { + "epoch": 1.3133189161684264, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.93062686920166, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8872414231300354, + "num_tokens": 400947442.0, + "step": 10324 + }, + { + "epoch": 1.313446126447017, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.885530471801758, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8953337669372559, + "num_tokens": 400984176.0, + "step": 10325 + }, + { + "epoch": 1.3135733367256075, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.915196418762207, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8825245499610901, + "num_tokens": 401029627.0, + "step": 10326 + }, + { + "epoch": 1.3137005470041978, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.993132591247559, + "learning_rate": 1e-06, + "loss": 0.3769, + "mean_token_accuracy": 0.8745229840278625, + "num_tokens": 401066246.0, + "step": 10327 + }, + { + "epoch": 1.3138277572827883, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.884780883789062, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8913215398788452, + "num_tokens": 401105634.0, + "step": 10328 + }, + { + "epoch": 1.3139549675613789, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.879554748535156, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8826014995574951, + "num_tokens": 401147764.0, + "step": 10329 + }, + { + "epoch": 1.3140821778399694, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.959298133850098, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8844706416130066, + "num_tokens": 401185411.0, + "step": 10330 + }, + { + "epoch": 1.31420938811856, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.904589653015137, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8980332016944885, + "num_tokens": 401222700.0, + "step": 10331 + }, + { + "epoch": 1.3143365983971504, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.864606857299805, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8910108804702759, + "num_tokens": 401256353.0, + "step": 10332 + }, + { + "epoch": 1.314463808675741, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.932210922241211, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8777520656585693, + "num_tokens": 401301581.0, + "step": 10333 + }, + { + "epoch": 1.3145910189543315, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.886404991149902, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8911014795303345, + "num_tokens": 401341358.0, + "step": 10334 + }, + { + "epoch": 1.314718229232922, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.970462799072266, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8865379691123962, + "num_tokens": 401377155.0, + "step": 10335 + }, + { + "epoch": 1.3148454395115126, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.860525131225586, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8731414675712585, + "num_tokens": 401422765.0, + "step": 10336 + }, + { + "epoch": 1.314972649790103, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.815644264221191, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8961037397384644, + "num_tokens": 401465282.0, + "step": 10337 + }, + { + "epoch": 1.3150998600686936, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.893814086914062, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8872630596160889, + "num_tokens": 401503101.0, + "step": 10338 + }, + { + "epoch": 1.3152270703472841, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.04278564453125, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8839332461357117, + "num_tokens": 401533816.0, + "step": 10339 + }, + { + "epoch": 1.3153542806258747, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.763556480407715, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.892845630645752, + "num_tokens": 401582640.0, + "step": 10340 + }, + { + "epoch": 1.3154814909044652, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.952866554260254, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8882912993431091, + "num_tokens": 401616627.0, + "step": 10341 + }, + { + "epoch": 1.3156087011830557, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.762069702148438, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9020819067955017, + "num_tokens": 401667582.0, + "step": 10342 + }, + { + "epoch": 1.315735911461646, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.848237991333008, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.885178804397583, + "num_tokens": 401710867.0, + "step": 10343 + }, + { + "epoch": 1.3158631217402366, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.991988182067871, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8779189586639404, + "num_tokens": 401751353.0, + "step": 10344 + }, + { + "epoch": 1.315990332018827, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.932342529296875, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8823888897895813, + "num_tokens": 401787750.0, + "step": 10345 + }, + { + "epoch": 1.3161175422974176, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.901447296142578, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8843736052513123, + "num_tokens": 401830429.0, + "step": 10346 + }, + { + "epoch": 1.3162447525760081, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.892012596130371, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8917239904403687, + "num_tokens": 401869498.0, + "step": 10347 + }, + { + "epoch": 1.3163719628545987, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.95604419708252, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8963262438774109, + "num_tokens": 401905121.0, + "step": 10348 + }, + { + "epoch": 1.3164991731331892, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.919937133789062, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8761501312255859, + "num_tokens": 401945641.0, + "step": 10349 + }, + { + "epoch": 1.3166263834117797, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.918302536010742, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.881049633026123, + "num_tokens": 401983935.0, + "step": 10350 + }, + { + "epoch": 1.3167535936903703, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.962241172790527, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8904471397399902, + "num_tokens": 402019589.0, + "step": 10351 + }, + { + "epoch": 1.3168808039689606, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.818856239318848, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8865909576416016, + "num_tokens": 402062038.0, + "step": 10352 + }, + { + "epoch": 1.317008014247551, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.88595962524414, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8866406679153442, + "num_tokens": 402108386.0, + "step": 10353 + }, + { + "epoch": 1.3171352245261416, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.908354759216309, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8948289752006531, + "num_tokens": 402143689.0, + "step": 10354 + }, + { + "epoch": 1.3172624348047322, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.96019172668457, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8939296007156372, + "num_tokens": 402179018.0, + "step": 10355 + }, + { + "epoch": 1.3173896450833227, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.954187393188477, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.8695112466812134, + "num_tokens": 402217289.0, + "step": 10356 + }, + { + "epoch": 1.3175168553619132, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.98838996887207, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.8762806057929993, + "num_tokens": 402253837.0, + "step": 10357 + }, + { + "epoch": 1.3176440656405037, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.81645393371582, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8944336771965027, + "num_tokens": 402296640.0, + "step": 10358 + }, + { + "epoch": 1.3177712759190943, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.858285903930664, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8950924873352051, + "num_tokens": 402333036.0, + "step": 10359 + }, + { + "epoch": 1.3178984861976848, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.95882511138916, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8995257616043091, + "num_tokens": 402366995.0, + "step": 10360 + }, + { + "epoch": 1.3180256964762753, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.916006088256836, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8773483037948608, + "num_tokens": 402406176.0, + "step": 10361 + }, + { + "epoch": 1.3181529067548658, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.983237266540527, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8792101144790649, + "num_tokens": 402442178.0, + "step": 10362 + }, + { + "epoch": 1.3182801170334564, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.867743492126465, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8823715448379517, + "num_tokens": 402485968.0, + "step": 10363 + }, + { + "epoch": 1.318407327312047, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.877250671386719, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.892932653427124, + "num_tokens": 402524849.0, + "step": 10364 + }, + { + "epoch": 1.3185345375906374, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.852560997009277, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.9022202491760254, + "num_tokens": 402563697.0, + "step": 10365 + }, + { + "epoch": 1.318661747869228, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.99954891204834, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8688540458679199, + "num_tokens": 402605938.0, + "step": 10366 + }, + { + "epoch": 1.3187889581478183, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.924208641052246, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8813984394073486, + "num_tokens": 402647784.0, + "step": 10367 + }, + { + "epoch": 1.3189161684264088, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.842062950134277, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8770207166671753, + "num_tokens": 402688681.0, + "step": 10368 + }, + { + "epoch": 1.3190433787049993, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.832648277282715, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8909698724746704, + "num_tokens": 402731048.0, + "step": 10369 + }, + { + "epoch": 1.3191705889835899, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.90635871887207, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8706449866294861, + "num_tokens": 402770743.0, + "step": 10370 + }, + { + "epoch": 1.3192977992621804, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.960999488830566, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8848345279693604, + "num_tokens": 402810446.0, + "step": 10371 + }, + { + "epoch": 1.319425009540771, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.836812019348145, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8879209160804749, + "num_tokens": 402854363.0, + "step": 10372 + }, + { + "epoch": 1.3195522198193614, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.901546478271484, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8935114741325378, + "num_tokens": 402891537.0, + "step": 10373 + }, + { + "epoch": 1.319679430097952, + "ewc_loss": 0.01483154296875, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 9.953975677490234e-06, + "grad_norm": 8.857446670532227, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.9004871845245361, + "num_tokens": 402926836.0, + "step": 10374 + }, + { + "epoch": 1.3198066403765425, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.00385856628418, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8863937854766846, + "num_tokens": 402963379.0, + "step": 10375 + }, + { + "epoch": 1.3199338506551328, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.80963134765625, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8939040899276733, + "num_tokens": 403005212.0, + "step": 10376 + }, + { + "epoch": 1.3200610609337233, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.93855094909668, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8877677917480469, + "num_tokens": 403040954.0, + "step": 10377 + }, + { + "epoch": 1.3201882712123139, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.816412925720215, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.878879189491272, + "num_tokens": 403084412.0, + "step": 10378 + }, + { + "epoch": 1.3203154814909044, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.93246078491211, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8978253602981567, + "num_tokens": 403119525.0, + "step": 10379 + }, + { + "epoch": 1.320442691769495, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.938034057617188, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8778058290481567, + "num_tokens": 403155221.0, + "step": 10380 + }, + { + "epoch": 1.3205699020480854, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.892659187316895, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8870383501052856, + "num_tokens": 403194948.0, + "step": 10381 + }, + { + "epoch": 1.320697112326676, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.975353240966797, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8853787183761597, + "num_tokens": 403231646.0, + "step": 10382 + }, + { + "epoch": 1.3208243226052665, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.854668617248535, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8955448865890503, + "num_tokens": 403271151.0, + "step": 10383 + }, + { + "epoch": 1.320951532883857, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.83549976348877, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9049533605575562, + "num_tokens": 403313812.0, + "step": 10384 + }, + { + "epoch": 1.3210787431624476, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.834094047546387, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8954365253448486, + "num_tokens": 403355709.0, + "step": 10385 + }, + { + "epoch": 1.321205953441038, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.896812438964844, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8982436656951904, + "num_tokens": 403393482.0, + "step": 10386 + }, + { + "epoch": 1.3213331637196286, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.966468811035156, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8794708251953125, + "num_tokens": 403430650.0, + "step": 10387 + }, + { + "epoch": 1.3214603739982191, + "ewc_loss": 0.014892578125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.812522888183594, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8927964568138123, + "num_tokens": 403474651.0, + "step": 10388 + }, + { + "epoch": 1.3215875842768097, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.987533569335938, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8911529183387756, + "num_tokens": 403508276.0, + "step": 10389 + }, + { + "epoch": 1.3217147945554002, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.849410057067871, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8953317999839783, + "num_tokens": 403547031.0, + "step": 10390 + }, + { + "epoch": 1.3218420048339907, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.971376419067383, + "learning_rate": 1e-06, + "loss": 0.3774, + "mean_token_accuracy": 0.870915949344635, + "num_tokens": 403587002.0, + "step": 10391 + }, + { + "epoch": 1.321969215112581, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.888018608093262, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8984041810035706, + "num_tokens": 403622789.0, + "step": 10392 + }, + { + "epoch": 1.3220964253911716, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.914738655090332, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8765489459037781, + "num_tokens": 403665928.0, + "step": 10393 + }, + { + "epoch": 1.322223635669762, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.920734405517578, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8801273703575134, + "num_tokens": 403705051.0, + "step": 10394 + }, + { + "epoch": 1.3223508459483526, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.90340805053711, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.890973687171936, + "num_tokens": 403741238.0, + "step": 10395 + }, + { + "epoch": 1.3224780562269431, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.895730018615723, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8913784027099609, + "num_tokens": 403778628.0, + "step": 10396 + }, + { + "epoch": 1.3226052665055337, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.92396068572998, + "learning_rate": 1e-06, + "loss": 0.4185, + "mean_token_accuracy": 0.8603343367576599, + "num_tokens": 403825275.0, + "step": 10397 + }, + { + "epoch": 1.3227324767841242, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.891144752502441, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8806949853897095, + "num_tokens": 403863409.0, + "step": 10398 + }, + { + "epoch": 1.3228596870627147, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.983979225158691, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8812152743339539, + "num_tokens": 403900544.0, + "step": 10399 + }, + { + "epoch": 1.3229868973413053, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.864516258239746, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8877725005149841, + "num_tokens": 403940660.0, + "step": 10400 + }, + { + "epoch": 1.3231141076198956, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.870099067687988, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8944823741912842, + "num_tokens": 403977018.0, + "step": 10401 + }, + { + "epoch": 1.323241317898486, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.918097496032715, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.884891152381897, + "num_tokens": 404018935.0, + "step": 10402 + }, + { + "epoch": 1.3233685281770766, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.820622444152832, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.901477575302124, + "num_tokens": 404059648.0, + "step": 10403 + }, + { + "epoch": 1.3234957384556671, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.88546085357666, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8981442451477051, + "num_tokens": 404100741.0, + "step": 10404 + }, + { + "epoch": 1.3236229487342577, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.851330757141113, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8914049863815308, + "num_tokens": 404141539.0, + "step": 10405 + }, + { + "epoch": 1.3237501590128482, + "ewc_loss": 0.01495361328125, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.851949691772461, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8813600540161133, + "num_tokens": 404187560.0, + "step": 10406 + }, + { + "epoch": 1.3238773692914387, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.03585433959961, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8827000856399536, + "num_tokens": 404221357.0, + "step": 10407 + }, + { + "epoch": 1.3240045795700293, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.895886421203613, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8871067762374878, + "num_tokens": 404265232.0, + "step": 10408 + }, + { + "epoch": 1.3241317898486198, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.896788597106934, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8757388591766357, + "num_tokens": 404307432.0, + "step": 10409 + }, + { + "epoch": 1.3242590001272103, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.887580871582031e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.848847389221191, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8845339417457581, + "num_tokens": 404349925.0, + "step": 10410 + }, + { + "epoch": 1.3243862104058008, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.921152114868164, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8748739957809448, + "num_tokens": 404393279.0, + "step": 10411 + }, + { + "epoch": 1.3245134206843914, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.900246620178223, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8857414722442627, + "num_tokens": 404437686.0, + "step": 10412 + }, + { + "epoch": 1.324640630962982, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.906575202941895, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8879653215408325, + "num_tokens": 404477796.0, + "step": 10413 + }, + { + "epoch": 1.3247678412415724, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.880786895751953, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8872243165969849, + "num_tokens": 404517175.0, + "step": 10414 + }, + { + "epoch": 1.324895051520163, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.93864917755127, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8939007520675659, + "num_tokens": 404552361.0, + "step": 10415 + }, + { + "epoch": 1.3250222617987533, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.853082656860352, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8945726156234741, + "num_tokens": 404590254.0, + "step": 10416 + }, + { + "epoch": 1.3251494720773438, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.004349708557129, + "learning_rate": 1e-06, + "loss": 0.3835, + "mean_token_accuracy": 0.8684238195419312, + "num_tokens": 404628152.0, + "step": 10417 + }, + { + "epoch": 1.3252766823559343, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.861685752868652, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8794092535972595, + "num_tokens": 404671220.0, + "step": 10418 + }, + { + "epoch": 1.3254038926345248, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.917550086975098, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8935682773590088, + "num_tokens": 404716313.0, + "step": 10419 + }, + { + "epoch": 1.3255311029131154, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.903450965881348, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.884023904800415, + "num_tokens": 404758587.0, + "step": 10420 + }, + { + "epoch": 1.325658313191706, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.032310485839844, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8790067434310913, + "num_tokens": 404794026.0, + "step": 10421 + }, + { + "epoch": 1.3257855234702964, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.884919166564941, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8913685083389282, + "num_tokens": 404834673.0, + "step": 10422 + }, + { + "epoch": 1.325912733748887, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.964559555053711, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8879324197769165, + "num_tokens": 404869200.0, + "step": 10423 + }, + { + "epoch": 1.3260399440274775, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.91871452331543, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8994430899620056, + "num_tokens": 404903306.0, + "step": 10424 + }, + { + "epoch": 1.3261671543060678, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.020487785339355, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.873185396194458, + "num_tokens": 404942636.0, + "step": 10425 + }, + { + "epoch": 1.3262943645846583, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.936031341552734, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8979238271713257, + "num_tokens": 404976054.0, + "step": 10426 + }, + { + "epoch": 1.3264215748632489, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.012031555175781, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8821391463279724, + "num_tokens": 405016370.0, + "step": 10427 + }, + { + "epoch": 1.3265487851418394, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.951553344726562, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8828428983688354, + "num_tokens": 405053975.0, + "step": 10428 + }, + { + "epoch": 1.32667599542043, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.862239837646484, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9070854187011719, + "num_tokens": 405092168.0, + "step": 10429 + }, + { + "epoch": 1.3268032056990204, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.965883255004883, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8898921012878418, + "num_tokens": 405126570.0, + "step": 10430 + }, + { + "epoch": 1.326930415977611, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.92059326171875, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8844836354255676, + "num_tokens": 405169311.0, + "step": 10431 + }, + { + "epoch": 1.3270576262562015, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.014034271240234, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8878172039985657, + "num_tokens": 405203613.0, + "step": 10432 + }, + { + "epoch": 1.327184836534792, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.923389434814453, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8875719904899597, + "num_tokens": 405242700.0, + "step": 10433 + }, + { + "epoch": 1.3273120468133826, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.945975303649902, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8829132318496704, + "num_tokens": 405276143.0, + "step": 10434 + }, + { + "epoch": 1.327439257091973, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.901942253112793, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8861372470855713, + "num_tokens": 405312933.0, + "step": 10435 + }, + { + "epoch": 1.3275664673705636, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.839727401733398, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8770491480827332, + "num_tokens": 405355309.0, + "step": 10436 + }, + { + "epoch": 1.3276936776491541, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.8822660446167, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8909381031990051, + "num_tokens": 405393404.0, + "step": 10437 + }, + { + "epoch": 1.3278208879277447, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.844643592834473, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.905160665512085, + "num_tokens": 405431488.0, + "step": 10438 + }, + { + "epoch": 1.3279480982063352, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.901351928710938, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8822999596595764, + "num_tokens": 405471263.0, + "step": 10439 + }, + { + "epoch": 1.3280753084849257, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.027759552001953, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.8685582280158997, + "num_tokens": 405505891.0, + "step": 10440 + }, + { + "epoch": 1.328202518763516, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.884089469909668, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8810700178146362, + "num_tokens": 405543530.0, + "step": 10441 + }, + { + "epoch": 1.3283297290421066, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.961071014404297, + "learning_rate": 1e-06, + "loss": 0.384, + "mean_token_accuracy": 0.8691845536231995, + "num_tokens": 405580441.0, + "step": 10442 + }, + { + "epoch": 1.328456939320697, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.836880683898926, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8776834011077881, + "num_tokens": 405622612.0, + "step": 10443 + }, + { + "epoch": 1.3285841495992876, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.86371898651123, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8929046392440796, + "num_tokens": 405661533.0, + "step": 10444 + }, + { + "epoch": 1.3287113598778781, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.869342803955078, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8878970742225647, + "num_tokens": 405701661.0, + "step": 10445 + }, + { + "epoch": 1.3288385701564687, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.95186996459961, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8865980505943298, + "num_tokens": 405741375.0, + "step": 10446 + }, + { + "epoch": 1.3289657804350592, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.982754707336426, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8752307891845703, + "num_tokens": 405781468.0, + "step": 10447 + }, + { + "epoch": 1.3290929907136497, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.983701705932617, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8824616074562073, + "num_tokens": 405817706.0, + "step": 10448 + }, + { + "epoch": 1.3292202009922403, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.953878402709961, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8820189833641052, + "num_tokens": 405857070.0, + "step": 10449 + }, + { + "epoch": 1.3293474112708306, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.060663223266602, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8823962807655334, + "num_tokens": 405889009.0, + "step": 10450 + }, + { + "epoch": 1.329474621549421, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.804896354675293, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.9030113220214844, + "num_tokens": 405927829.0, + "step": 10451 + }, + { + "epoch": 1.3296018318280116, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.919754028320312, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8815494775772095, + "num_tokens": 405962702.0, + "step": 10452 + }, + { + "epoch": 1.3297290421066021, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.948545455932617, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8924053907394409, + "num_tokens": 406001396.0, + "step": 10453 + }, + { + "epoch": 1.3298562523851927, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.907626152038574, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8946573734283447, + "num_tokens": 406033588.0, + "step": 10454 + }, + { + "epoch": 1.3299834626637832, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.964252471923828, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8829741477966309, + "num_tokens": 406072255.0, + "step": 10455 + }, + { + "epoch": 1.3301106729423737, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.921148300170898, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8881042003631592, + "num_tokens": 406108606.0, + "step": 10456 + }, + { + "epoch": 1.3302378832209643, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.024868965148926, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8830560445785522, + "num_tokens": 406144646.0, + "step": 10457 + }, + { + "epoch": 1.3303650934995548, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.02644157409668, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8920918703079224, + "num_tokens": 406176780.0, + "step": 10458 + }, + { + "epoch": 1.3304923037781453, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.843987464904785, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.885934591293335, + "num_tokens": 406215989.0, + "step": 10459 + }, + { + "epoch": 1.3306195140567358, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.934595108032227, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.885167121887207, + "num_tokens": 406257227.0, + "step": 10460 + }, + { + "epoch": 1.3307467243353264, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.885350227355957, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8875058889389038, + "num_tokens": 406295328.0, + "step": 10461 + }, + { + "epoch": 1.330873934613917, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.841224670410156, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.9003228545188904, + "num_tokens": 406334183.0, + "step": 10462 + }, + { + "epoch": 1.3310011448925074, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.900033950805664, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8866918087005615, + "num_tokens": 406371292.0, + "step": 10463 + }, + { + "epoch": 1.331128355171098, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.9290189743042, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.879565417766571, + "num_tokens": 406408266.0, + "step": 10464 + }, + { + "epoch": 1.3312555654496883, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.023752212524414, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8953865766525269, + "num_tokens": 406440180.0, + "step": 10465 + }, + { + "epoch": 1.3313827757282788, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.898533821105957, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8921969532966614, + "num_tokens": 406478060.0, + "step": 10466 + }, + { + "epoch": 1.3315099860068693, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.86485767364502, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8999487161636353, + "num_tokens": 406519208.0, + "step": 10467 + }, + { + "epoch": 1.3316371962854598, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.985406875610352, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8839578032493591, + "num_tokens": 406555270.0, + "step": 10468 + }, + { + "epoch": 1.3317644065640504, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.943690299987793, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.887667715549469, + "num_tokens": 406594198.0, + "step": 10469 + }, + { + "epoch": 1.331891616842641, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.0055570602417, + "learning_rate": 1e-06, + "loss": 0.3869, + "mean_token_accuracy": 0.8702405691146851, + "num_tokens": 406633652.0, + "step": 10470 + }, + { + "epoch": 1.3320188271212314, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.910347938537598, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8916364312171936, + "num_tokens": 406672855.0, + "step": 10471 + }, + { + "epoch": 1.332146037399822, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.967062950134277, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8799402713775635, + "num_tokens": 406712209.0, + "step": 10472 + }, + { + "epoch": 1.3322732476784125, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.88918685913086, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8954120874404907, + "num_tokens": 406753942.0, + "step": 10473 + }, + { + "epoch": 1.3324004579570028, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.884068489074707, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8932116031646729, + "num_tokens": 406790662.0, + "step": 10474 + }, + { + "epoch": 1.3325276682355933, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.027619361877441, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.8720904588699341, + "num_tokens": 406829094.0, + "step": 10475 + }, + { + "epoch": 1.3326548785141838, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.061192512512207, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8928146362304688, + "num_tokens": 406858473.0, + "step": 10476 + }, + { + "epoch": 1.3327820887927744, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.863932609558105, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.9056025147438049, + "num_tokens": 406896335.0, + "step": 10477 + }, + { + "epoch": 1.332909299071365, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.857645988464355, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8895634412765503, + "num_tokens": 406939560.0, + "step": 10478 + }, + { + "epoch": 1.3330365093499554, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.919136047363281, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8988674283027649, + "num_tokens": 406978264.0, + "step": 10479 + }, + { + "epoch": 1.333163719628546, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.799697875976562, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.9008622169494629, + "num_tokens": 407022095.0, + "step": 10480 + }, + { + "epoch": 1.3332909299071365, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.956304550170898, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8916480541229248, + "num_tokens": 407060923.0, + "step": 10481 + }, + { + "epoch": 1.333418140185727, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.909070014953613, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8961344361305237, + "num_tokens": 407095098.0, + "step": 10482 + }, + { + "epoch": 1.3335453504643175, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.93140697479248, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8998596668243408, + "num_tokens": 407128736.0, + "step": 10483 + }, + { + "epoch": 1.333672560742908, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.905901908874512, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8927212357521057, + "num_tokens": 407166321.0, + "step": 10484 + }, + { + "epoch": 1.3337997710214986, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.937250137329102, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8871235847473145, + "num_tokens": 407200480.0, + "step": 10485 + }, + { + "epoch": 1.3339269813000891, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.995648384094238, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8859915733337402, + "num_tokens": 407236969.0, + "step": 10486 + }, + { + "epoch": 1.3340541915786797, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.946455001831055, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.8706302642822266, + "num_tokens": 407273481.0, + "step": 10487 + }, + { + "epoch": 1.3341814018572702, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.050312995910645, + "learning_rate": 1e-06, + "loss": 0.3905, + "mean_token_accuracy": 0.8692384362220764, + "num_tokens": 407308474.0, + "step": 10488 + }, + { + "epoch": 1.3343086121358605, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.937357902526855, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8858380317687988, + "num_tokens": 407347096.0, + "step": 10489 + }, + { + "epoch": 1.334435822414451, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.935803413391113, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8719353675842285, + "num_tokens": 407387992.0, + "step": 10490 + }, + { + "epoch": 1.3345630326930416, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.962213516235352, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8883612751960754, + "num_tokens": 407425230.0, + "step": 10491 + }, + { + "epoch": 1.334690242971632, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.864348411560059, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8917844295501709, + "num_tokens": 407468281.0, + "step": 10492 + }, + { + "epoch": 1.3348174532502226, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.939608573913574, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.8732956051826477, + "num_tokens": 407507385.0, + "step": 10493 + }, + { + "epoch": 1.3349446635288131, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.840764045715332, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9015988111495972, + "num_tokens": 407544991.0, + "step": 10494 + }, + { + "epoch": 1.3350718738074037, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.935876846313477, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8879275321960449, + "num_tokens": 407584325.0, + "step": 10495 + }, + { + "epoch": 1.3351990840859942, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.894064903259277, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8968720436096191, + "num_tokens": 407623596.0, + "step": 10496 + }, + { + "epoch": 1.3353262943645847, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.896282196044922, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8982653021812439, + "num_tokens": 407662658.0, + "step": 10497 + }, + { + "epoch": 1.3354535046431752, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.935264587402344e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.979105949401855, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8771209716796875, + "num_tokens": 407700526.0, + "step": 10498 + }, + { + "epoch": 1.3355807149217656, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.944767951965332, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8918750882148743, + "num_tokens": 407740663.0, + "step": 10499 + }, + { + "epoch": 1.335707925200356, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.93986701965332, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8824828863143921, + "num_tokens": 407781518.0, + "step": 10500 + }, + { + "epoch": 1.3358351354789466, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.94532585144043, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8946094512939453, + "num_tokens": 407817785.0, + "step": 10501 + }, + { + "epoch": 1.3359623457575371, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.998534202575684, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8870093822479248, + "num_tokens": 407852821.0, + "step": 10502 + }, + { + "epoch": 1.3360895560361277, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.971966743469238, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8789106607437134, + "num_tokens": 407890740.0, + "step": 10503 + }, + { + "epoch": 1.3362167663147182, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.840221405029297, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8907942175865173, + "num_tokens": 407931207.0, + "step": 10504 + }, + { + "epoch": 1.3363439765933087, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.817811965942383, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8903653025627136, + "num_tokens": 407979013.0, + "step": 10505 + }, + { + "epoch": 1.3364711868718993, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.909017562866211, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8953368663787842, + "num_tokens": 408018571.0, + "step": 10506 + }, + { + "epoch": 1.3365983971504898, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.885815620422363, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.894233226776123, + "num_tokens": 408053866.0, + "step": 10507 + }, + { + "epoch": 1.3367256074290803, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.073700904846191, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8899534940719604, + "num_tokens": 408084572.0, + "step": 10508 + }, + { + "epoch": 1.3368528177076708, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.861251831054688, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8952198028564453, + "num_tokens": 408122137.0, + "step": 10509 + }, + { + "epoch": 1.3369800279862614, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.939621925354004, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8869034051895142, + "num_tokens": 408162323.0, + "step": 10510 + }, + { + "epoch": 1.337107238264852, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.842005729675293, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8952977657318115, + "num_tokens": 408204112.0, + "step": 10511 + }, + { + "epoch": 1.3372344485434424, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.886649131774902, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8953114748001099, + "num_tokens": 408246404.0, + "step": 10512 + }, + { + "epoch": 1.337361658822033, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.013589859008789, + "learning_rate": 1e-06, + "loss": 0.4183, + "mean_token_accuracy": 0.8584730625152588, + "num_tokens": 408287960.0, + "step": 10513 + }, + { + "epoch": 1.3374888691006233, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.963932991027832, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8861232995986938, + "num_tokens": 408326390.0, + "step": 10514 + }, + { + "epoch": 1.3376160793792138, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.943614959716797, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8919323086738586, + "num_tokens": 408363708.0, + "step": 10515 + }, + { + "epoch": 1.3377432896578043, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.90294361114502, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8985342383384705, + "num_tokens": 408406633.0, + "step": 10516 + }, + { + "epoch": 1.3378704999363948, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.923686981201172, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8901294469833374, + "num_tokens": 408449248.0, + "step": 10517 + }, + { + "epoch": 1.3379977102149854, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.091421127319336, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.877845823764801, + "num_tokens": 408483566.0, + "step": 10518 + }, + { + "epoch": 1.338124920493576, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.920309066772461, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.875038206577301, + "num_tokens": 408530808.0, + "step": 10519 + }, + { + "epoch": 1.3382521307721664, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 4.9591064453125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.95948314666748, + "learning_rate": 1e-06, + "loss": 0.3838, + "mean_token_accuracy": 0.8720242977142334, + "num_tokens": 408567841.0, + "step": 10520 + }, + { + "epoch": 1.338379341050757, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.1345796585083, + "learning_rate": 1e-06, + "loss": 0.4219, + "mean_token_accuracy": 0.8600878715515137, + "num_tokens": 408603080.0, + "step": 10521 + }, + { + "epoch": 1.3385065513293475, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.984338760375977, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8894971609115601, + "num_tokens": 408634865.0, + "step": 10522 + }, + { + "epoch": 1.3386337616079378, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.866591453552246, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8974967002868652, + "num_tokens": 408676757.0, + "step": 10523 + }, + { + "epoch": 1.3387609718865283, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.872233390808105, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8815697431564331, + "num_tokens": 408719987.0, + "step": 10524 + }, + { + "epoch": 1.3388881821651188, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.960955619812012, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8794746994972229, + "num_tokens": 408759153.0, + "step": 10525 + }, + { + "epoch": 1.3390153924437094, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.010577201843262, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8817636966705322, + "num_tokens": 408792463.0, + "step": 10526 + }, + { + "epoch": 1.3391426027223, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.943156242370605, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8843543529510498, + "num_tokens": 408831867.0, + "step": 10527 + }, + { + "epoch": 1.3392698130008904, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.977800369262695, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8785609602928162, + "num_tokens": 408870988.0, + "step": 10528 + }, + { + "epoch": 1.339397023279481, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.943060874938965, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8787106871604919, + "num_tokens": 408907137.0, + "step": 10529 + }, + { + "epoch": 1.3395242335580715, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.91158390045166, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9081771969795227, + "num_tokens": 408945138.0, + "step": 10530 + }, + { + "epoch": 1.339651443836662, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.944175720214844, + "learning_rate": 1e-06, + "loss": 0.3654, + "mean_token_accuracy": 0.8750280141830444, + "num_tokens": 408984145.0, + "step": 10531 + }, + { + "epoch": 1.3397786541152525, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.872188568115234, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8800825476646423, + "num_tokens": 409025605.0, + "step": 10532 + }, + { + "epoch": 1.339905864393843, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.883637428283691, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8775795102119446, + "num_tokens": 409066631.0, + "step": 10533 + }, + { + "epoch": 1.3400330746724336, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.902880668640137, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8831247687339783, + "num_tokens": 409112017.0, + "step": 10534 + }, + { + "epoch": 1.3401602849510241, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.870448112487793, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8894424438476562, + "num_tokens": 409155315.0, + "step": 10535 + }, + { + "epoch": 1.3402874952296147, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.936835289001465, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9031773209571838, + "num_tokens": 409187347.0, + "step": 10536 + }, + { + "epoch": 1.3404147055082052, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.069934844970703, + "learning_rate": 1e-06, + "loss": 0.394, + "mean_token_accuracy": 0.8668835759162903, + "num_tokens": 409222145.0, + "step": 10537 + }, + { + "epoch": 1.3405419157867955, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.877325057983398, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.9008576273918152, + "num_tokens": 409259727.0, + "step": 10538 + }, + { + "epoch": 1.340669126065386, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.872608184814453, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8981769680976868, + "num_tokens": 409296611.0, + "step": 10539 + }, + { + "epoch": 1.3407963363439765, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.940119743347168, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8959827423095703, + "num_tokens": 409332694.0, + "step": 10540 + }, + { + "epoch": 1.340923546622567, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.928306579589844, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8898811340332031, + "num_tokens": 409371470.0, + "step": 10541 + }, + { + "epoch": 1.3410507569011576, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.875697135925293, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8858692646026611, + "num_tokens": 409411693.0, + "step": 10542 + }, + { + "epoch": 1.3411779671797481, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.005607604980469, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8784998655319214, + "num_tokens": 409450318.0, + "step": 10543 + }, + { + "epoch": 1.3413051774583387, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.030426979064941, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.8700469136238098, + "num_tokens": 409486854.0, + "step": 10544 + }, + { + "epoch": 1.3414323877369292, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.93303394317627, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8860911130905151, + "num_tokens": 409522267.0, + "step": 10545 + }, + { + "epoch": 1.3415595980155197, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.839353561401367, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8872953057289124, + "num_tokens": 409566321.0, + "step": 10546 + }, + { + "epoch": 1.3416868082941102, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.939281463623047, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8978168964385986, + "num_tokens": 409603855.0, + "step": 10547 + }, + { + "epoch": 1.3418140185727006, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.923565864562988, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.886061429977417, + "num_tokens": 409643016.0, + "step": 10548 + }, + { + "epoch": 1.341941228851291, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.989434242248535, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8899298906326294, + "num_tokens": 409678260.0, + "step": 10549 + }, + { + "epoch": 1.3420684391298816, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.936446189880371, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8971753120422363, + "num_tokens": 409715567.0, + "step": 10550 + }, + { + "epoch": 1.3421956494084721, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.924067497253418, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8921730518341064, + "num_tokens": 409750126.0, + "step": 10551 + }, + { + "epoch": 1.3423228596870627, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.845562934875488, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8832197189331055, + "num_tokens": 409800346.0, + "step": 10552 + }, + { + "epoch": 1.3424500699656532, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.91323471069336, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8913513422012329, + "num_tokens": 409838688.0, + "step": 10553 + }, + { + "epoch": 1.3425772802442437, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.968122482299805, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8814389705657959, + "num_tokens": 409874305.0, + "step": 10554 + }, + { + "epoch": 1.3427044905228342, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.979451179504395, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8962934017181396, + "num_tokens": 409911629.0, + "step": 10555 + }, + { + "epoch": 1.3428317008014248, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.936095237731934, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8913376331329346, + "num_tokens": 409952252.0, + "step": 10556 + }, + { + "epoch": 1.3429589110800153, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.847146034240723, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.9001584053039551, + "num_tokens": 409998049.0, + "step": 10557 + }, + { + "epoch": 1.3430861213586058, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.037833213806152, + "learning_rate": 1e-06, + "loss": 0.3745, + "mean_token_accuracy": 0.8725993633270264, + "num_tokens": 410035351.0, + "step": 10558 + }, + { + "epoch": 1.3432133316371964, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.939115524291992, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8844513893127441, + "num_tokens": 410081535.0, + "step": 10559 + }, + { + "epoch": 1.343340541915787, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.821885108947754, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8891584277153015, + "num_tokens": 410126032.0, + "step": 10560 + }, + { + "epoch": 1.3434677521943774, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.90615463256836, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8983714580535889, + "num_tokens": 410166107.0, + "step": 10561 + }, + { + "epoch": 1.343594962472968, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.849483489990234, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8958956003189087, + "num_tokens": 410202874.0, + "step": 10562 + }, + { + "epoch": 1.3437221727515583, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.907479286193848, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8904337286949158, + "num_tokens": 410241015.0, + "step": 10563 + }, + { + "epoch": 1.3438493830301488, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.935760498046875, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8827066421508789, + "num_tokens": 410275387.0, + "step": 10564 + }, + { + "epoch": 1.3439765933087393, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.913336753845215, + "learning_rate": 1e-06, + "loss": 0.3863, + "mean_token_accuracy": 0.8665333986282349, + "num_tokens": 410319412.0, + "step": 10565 + }, + { + "epoch": 1.3441038035873298, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.829968452453613, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8927690982818604, + "num_tokens": 410358070.0, + "step": 10566 + }, + { + "epoch": 1.3442310138659204, + "ewc_loss": 0.01507568359375, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.960103034973145, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8963901996612549, + "num_tokens": 410397871.0, + "step": 10567 + }, + { + "epoch": 1.344358224144511, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.92404556274414, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8909927010536194, + "num_tokens": 410436576.0, + "step": 10568 + }, + { + "epoch": 1.3444854344231014, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.86927318572998, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8979429006576538, + "num_tokens": 410474698.0, + "step": 10569 + }, + { + "epoch": 1.344612644701692, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.912930488586426, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8810591101646423, + "num_tokens": 410518476.0, + "step": 10570 + }, + { + "epoch": 1.3447398549802825, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.894464492797852, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8845056295394897, + "num_tokens": 410563959.0, + "step": 10571 + }, + { + "epoch": 1.3448670652588728, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.940241813659668, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8893369436264038, + "num_tokens": 410602570.0, + "step": 10572 + }, + { + "epoch": 1.3449942755374633, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.111132621765137, + "learning_rate": 1e-06, + "loss": 0.3783, + "mean_token_accuracy": 0.872555136680603, + "num_tokens": 410639180.0, + "step": 10573 + }, + { + "epoch": 1.3451214858160538, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.0067901611328125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.854403495788574, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9000169038772583, + "num_tokens": 410678376.0, + "step": 10574 + }, + { + "epoch": 1.3452486960946444, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.959370613098145, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8856586813926697, + "num_tokens": 410715008.0, + "step": 10575 + }, + { + "epoch": 1.345375906373235, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.983319282531738, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.887505054473877, + "num_tokens": 410754269.0, + "step": 10576 + }, + { + "epoch": 1.3455031166518254, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.883774757385254, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8924039602279663, + "num_tokens": 410794501.0, + "step": 10577 + }, + { + "epoch": 1.345630326930416, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.005366325378418, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8761235475540161, + "num_tokens": 410834412.0, + "step": 10578 + }, + { + "epoch": 1.3457575372090065, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.918743133544922, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8905390501022339, + "num_tokens": 410869979.0, + "step": 10579 + }, + { + "epoch": 1.345884747487597, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.882265090942383, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.881611168384552, + "num_tokens": 410911792.0, + "step": 10580 + }, + { + "epoch": 1.3460119577661875, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.909347534179688, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.885378360748291, + "num_tokens": 410952301.0, + "step": 10581 + }, + { + "epoch": 1.346139168044778, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.907183647155762, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8744411468505859, + "num_tokens": 410997500.0, + "step": 10582 + }, + { + "epoch": 1.3462663783233686, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.104192733764648, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8786801099777222, + "num_tokens": 411029591.0, + "step": 10583 + }, + { + "epoch": 1.3463935886019591, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.814779281616211, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8900136351585388, + "num_tokens": 411070700.0, + "step": 10584 + }, + { + "epoch": 1.3465207988805497, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.843886375427246, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8922632336616516, + "num_tokens": 411114426.0, + "step": 10585 + }, + { + "epoch": 1.3466480091591402, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.962969779968262, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.8806100487709045, + "num_tokens": 411157399.0, + "step": 10586 + }, + { + "epoch": 1.3467752194377305, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.885815620422363, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8868613839149475, + "num_tokens": 411195222.0, + "step": 10587 + }, + { + "epoch": 1.346902429716321, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.987310409545898, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8902486562728882, + "num_tokens": 411235022.0, + "step": 10588 + }, + { + "epoch": 1.3470296399949115, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.863432884216309, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8869220018386841, + "num_tokens": 411275384.0, + "step": 10589 + }, + { + "epoch": 1.347156850273502, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.973167419433594, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8987648487091064, + "num_tokens": 411309408.0, + "step": 10590 + }, + { + "epoch": 1.3472840605520926, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.911606788635254, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8921306729316711, + "num_tokens": 411348483.0, + "step": 10591 + }, + { + "epoch": 1.3474112708306831, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.893050193786621, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8946890234947205, + "num_tokens": 411389524.0, + "step": 10592 + }, + { + "epoch": 1.3475384811092737, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.973302841186523, + "learning_rate": 1e-06, + "loss": 0.3933, + "mean_token_accuracy": 0.8660714626312256, + "num_tokens": 411430494.0, + "step": 10593 + }, + { + "epoch": 1.3476656913878642, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.91226863861084, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8906692862510681, + "num_tokens": 411467397.0, + "step": 10594 + }, + { + "epoch": 1.3477929016664547, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.049725532531738, + "learning_rate": 1e-06, + "loss": 0.3643, + "mean_token_accuracy": 0.8754077553749084, + "num_tokens": 411503248.0, + "step": 10595 + }, + { + "epoch": 1.3479201119450452, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.012740135192871, + "learning_rate": 1e-06, + "loss": 0.396, + "mean_token_accuracy": 0.8709284663200378, + "num_tokens": 411540611.0, + "step": 10596 + }, + { + "epoch": 1.3480473222236355, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.964190483093262, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.8778318762779236, + "num_tokens": 411581200.0, + "step": 10597 + }, + { + "epoch": 1.348174532502226, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.908087730407715, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8756455183029175, + "num_tokens": 411625705.0, + "step": 10598 + }, + { + "epoch": 1.3483017427808166, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.86396598815918, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8871756792068481, + "num_tokens": 411667652.0, + "step": 10599 + }, + { + "epoch": 1.3484289530594071, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.039054870605469, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8809596300125122, + "num_tokens": 411703105.0, + "step": 10600 + }, + { + "epoch": 1.3485561633379977, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.817843437194824, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8908983469009399, + "num_tokens": 411748393.0, + "step": 10601 + }, + { + "epoch": 1.3486833736165882, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.916600227355957, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.89402174949646, + "num_tokens": 411786820.0, + "step": 10602 + }, + { + "epoch": 1.3488105838951787, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.149002075195312, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8749115467071533, + "num_tokens": 411819236.0, + "step": 10603 + }, + { + "epoch": 1.3489377941737692, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.991534233093262, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8837215900421143, + "num_tokens": 411852091.0, + "step": 10604 + }, + { + "epoch": 1.3490650044523598, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.019062995910645, + "learning_rate": 1e-06, + "loss": 0.3945, + "mean_token_accuracy": 0.8696197867393494, + "num_tokens": 411889022.0, + "step": 10605 + }, + { + "epoch": 1.3491922147309503, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.966303825378418, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.893128514289856, + "num_tokens": 411925429.0, + "step": 10606 + }, + { + "epoch": 1.3493194250095408, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.939949035644531, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8818174600601196, + "num_tokens": 411963755.0, + "step": 10607 + }, + { + "epoch": 1.3494466352881314, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.002096176147461, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8864429593086243, + "num_tokens": 411999672.0, + "step": 10608 + }, + { + "epoch": 1.3495738455667219, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.864933013916016, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.894065797328949, + "num_tokens": 412042808.0, + "step": 10609 + }, + { + "epoch": 1.3497010558453124, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.95435905456543, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.875211238861084, + "num_tokens": 412082084.0, + "step": 10610 + }, + { + "epoch": 1.349828266123903, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.000732421875, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.885788083076477, + "num_tokens": 412116904.0, + "step": 10611 + }, + { + "epoch": 1.3499554764024932, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.850373268127441, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8886340260505676, + "num_tokens": 412154059.0, + "step": 10612 + }, + { + "epoch": 1.3500826866810838, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.911490440368652, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8828235268592834, + "num_tokens": 412195660.0, + "step": 10613 + }, + { + "epoch": 1.3502098969596743, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.823895454406738, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8987188339233398, + "num_tokens": 412237370.0, + "step": 10614 + }, + { + "epoch": 1.3503371072382648, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.92054557800293, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.886085569858551, + "num_tokens": 412277608.0, + "step": 10615 + }, + { + "epoch": 1.3504643175168554, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.909039497375488, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8886107206344604, + "num_tokens": 412317580.0, + "step": 10616 + }, + { + "epoch": 1.350591527795446, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.903616905212402, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8888437747955322, + "num_tokens": 412358660.0, + "step": 10617 + }, + { + "epoch": 1.3507187380740364, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.923004150390625, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8994886875152588, + "num_tokens": 412401322.0, + "step": 10618 + }, + { + "epoch": 1.350845948352627, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.01615047454834, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8855161666870117, + "num_tokens": 412434787.0, + "step": 10619 + }, + { + "epoch": 1.3509731586312175, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.916725158691406, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8912837505340576, + "num_tokens": 412472433.0, + "step": 10620 + }, + { + "epoch": 1.3511003689098078, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.884228706359863, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.889313817024231, + "num_tokens": 412514362.0, + "step": 10621 + }, + { + "epoch": 1.3512275791883983, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.960899353027344, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.89350426197052, + "num_tokens": 412550722.0, + "step": 10622 + }, + { + "epoch": 1.3513547894669888, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.081761360168457, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8807920813560486, + "num_tokens": 412582037.0, + "step": 10623 + }, + { + "epoch": 1.3514819997455794, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.842418670654297, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.8764404058456421, + "num_tokens": 412627120.0, + "step": 10624 + }, + { + "epoch": 1.35160921002417, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.9671049118042, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.872721791267395, + "num_tokens": 412661949.0, + "step": 10625 + }, + { + "epoch": 1.3517364203027604, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.886180877685547, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8960922956466675, + "num_tokens": 412703565.0, + "step": 10626 + }, + { + "epoch": 1.351863630581351, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.915705680847168, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8773818016052246, + "num_tokens": 412743900.0, + "step": 10627 + }, + { + "epoch": 1.3519908408599415, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.881487846374512, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8896239995956421, + "num_tokens": 412786656.0, + "step": 10628 + }, + { + "epoch": 1.352118051138532, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.893957138061523, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8895137310028076, + "num_tokens": 412830152.0, + "step": 10629 + }, + { + "epoch": 1.3522452614171225, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.906359672546387, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8937652111053467, + "num_tokens": 412870609.0, + "step": 10630 + }, + { + "epoch": 1.352372471695713, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.828837394714355, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.889763355255127, + "num_tokens": 412913677.0, + "step": 10631 + }, + { + "epoch": 1.3524996819743036, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.953104019165039, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8838715553283691, + "num_tokens": 412948848.0, + "step": 10632 + }, + { + "epoch": 1.3526268922528941, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.881623268127441, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9003180265426636, + "num_tokens": 412987510.0, + "step": 10633 + }, + { + "epoch": 1.3527541025314846, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.86706829071045, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9045073986053467, + "num_tokens": 413027913.0, + "step": 10634 + }, + { + "epoch": 1.3528813128100752, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.860274314880371, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8991496562957764, + "num_tokens": 413065194.0, + "step": 10635 + }, + { + "epoch": 1.3530085230886655, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.965306282043457, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8966042399406433, + "num_tokens": 413100897.0, + "step": 10636 + }, + { + "epoch": 1.353135733367256, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.853133201599121, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8960778713226318, + "num_tokens": 413141256.0, + "step": 10637 + }, + { + "epoch": 1.3532629436458465, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.881885528564453, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8940191864967346, + "num_tokens": 413180604.0, + "step": 10638 + }, + { + "epoch": 1.353390153924437, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.870223045349121, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9063509106636047, + "num_tokens": 413218777.0, + "step": 10639 + }, + { + "epoch": 1.3535173642030276, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.892980575561523, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8877588510513306, + "num_tokens": 413258562.0, + "step": 10640 + }, + { + "epoch": 1.3536445744816181, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.878326416015625, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8945168256759644, + "num_tokens": 413298972.0, + "step": 10641 + }, + { + "epoch": 1.3537717847602087, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.975835800170898, + "learning_rate": 1e-06, + "loss": 0.3955, + "mean_token_accuracy": 0.8713932633399963, + "num_tokens": 413339907.0, + "step": 10642 + }, + { + "epoch": 1.3538989950387992, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.016319274902344, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8759628534317017, + "num_tokens": 413374537.0, + "step": 10643 + }, + { + "epoch": 1.3540262053173897, + "ewc_loss": 0.0150146484375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.894447326660156, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8838974237442017, + "num_tokens": 413411288.0, + "step": 10644 + }, + { + "epoch": 1.3541534155959802, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.95801067352295, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8927397727966309, + "num_tokens": 413449683.0, + "step": 10645 + }, + { + "epoch": 1.3542806258745705, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.973888397216797, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.889815092086792, + "num_tokens": 413487678.0, + "step": 10646 + }, + { + "epoch": 1.354407836153161, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.9500150680542, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8858479261398315, + "num_tokens": 413524818.0, + "step": 10647 + }, + { + "epoch": 1.3545350464317516, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.807409286499023, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8944158554077148, + "num_tokens": 413568677.0, + "step": 10648 + }, + { + "epoch": 1.3546622567103421, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.933989524841309, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8883569836616516, + "num_tokens": 413606711.0, + "step": 10649 + }, + { + "epoch": 1.3547894669889327, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.939008712768555, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8856803178787231, + "num_tokens": 413647174.0, + "step": 10650 + }, + { + "epoch": 1.3549166772675232, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.91055679321289, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8811962604522705, + "num_tokens": 413692524.0, + "step": 10651 + }, + { + "epoch": 1.3550438875461137, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.998766899108887, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.884548544883728, + "num_tokens": 413730282.0, + "step": 10652 + }, + { + "epoch": 1.3551710978247042, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.9326810836792, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8916715383529663, + "num_tokens": 413771150.0, + "step": 10653 + }, + { + "epoch": 1.3552983081032948, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.802740097045898, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9069957733154297, + "num_tokens": 413809298.0, + "step": 10654 + }, + { + "epoch": 1.3554255183818853, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.950763702392578, + "learning_rate": 1e-06, + "loss": 0.3783, + "mean_token_accuracy": 0.8694916367530823, + "num_tokens": 413855016.0, + "step": 10655 + }, + { + "epoch": 1.3555527286604758, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.950759887695312, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8840281963348389, + "num_tokens": 413893334.0, + "step": 10656 + }, + { + "epoch": 1.3556799389390664, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.942112922668457, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.902506947517395, + "num_tokens": 413928722.0, + "step": 10657 + }, + { + "epoch": 1.3558071492176569, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.95219612121582, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8854461312294006, + "num_tokens": 413972407.0, + "step": 10658 + }, + { + "epoch": 1.3559343594962474, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.954875946044922, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8799335956573486, + "num_tokens": 414014187.0, + "step": 10659 + }, + { + "epoch": 1.356061569774838, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.917623519897461, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8898559808731079, + "num_tokens": 414053124.0, + "step": 10660 + }, + { + "epoch": 1.3561887800534282, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.990019798278809, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8860405683517456, + "num_tokens": 414092067.0, + "step": 10661 + }, + { + "epoch": 1.3563159903320188, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.91482925415039, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8760945200920105, + "num_tokens": 414131428.0, + "step": 10662 + }, + { + "epoch": 1.3564432006106093, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.973129272460938, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.9037070274353027, + "num_tokens": 414168852.0, + "step": 10663 + }, + { + "epoch": 1.3565704108891998, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.915874481201172, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8803859949111938, + "num_tokens": 414213419.0, + "step": 10664 + }, + { + "epoch": 1.3566976211677904, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.910018920898438, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8857635259628296, + "num_tokens": 414258257.0, + "step": 10665 + }, + { + "epoch": 1.3568248314463809, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.971936225891113, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.883661687374115, + "num_tokens": 414294907.0, + "step": 10666 + }, + { + "epoch": 1.3569520417249714, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.117738723754883, + "learning_rate": 1e-06, + "loss": 0.3754, + "mean_token_accuracy": 0.8723137378692627, + "num_tokens": 414327750.0, + "step": 10667 + }, + { + "epoch": 1.357079252003562, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.883874893188477, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8918259739875793, + "num_tokens": 414368240.0, + "step": 10668 + }, + { + "epoch": 1.3572064622821525, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.874484062194824, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8920988440513611, + "num_tokens": 414410922.0, + "step": 10669 + }, + { + "epoch": 1.3573336725607428, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.940834045410156, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8958081603050232, + "num_tokens": 414445462.0, + "step": 10670 + }, + { + "epoch": 1.3574608828393333, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.922788619995117, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8951775431632996, + "num_tokens": 414481806.0, + "step": 10671 + }, + { + "epoch": 1.3575880931179238, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.92961597442627, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8889687061309814, + "num_tokens": 414523807.0, + "step": 10672 + }, + { + "epoch": 1.3577153033965144, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.9281644821167, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8809866905212402, + "num_tokens": 414563403.0, + "step": 10673 + }, + { + "epoch": 1.357842513675105, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.973217010498047, + "learning_rate": 1e-06, + "loss": 0.4178, + "mean_token_accuracy": 0.8605636358261108, + "num_tokens": 414604471.0, + "step": 10674 + }, + { + "epoch": 1.3579697239536954, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.901516914367676, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8810626268386841, + "num_tokens": 414645134.0, + "step": 10675 + }, + { + "epoch": 1.358096934232286, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.840779304504395, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8994847536087036, + "num_tokens": 414681984.0, + "step": 10676 + }, + { + "epoch": 1.3582241445108765, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.905109405517578, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8941845893859863, + "num_tokens": 414717763.0, + "step": 10677 + }, + { + "epoch": 1.358351354789467, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.913320541381836, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.9005982875823975, + "num_tokens": 414753855.0, + "step": 10678 + }, + { + "epoch": 1.3584785650680575, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.056380271911621, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.880695104598999, + "num_tokens": 414785318.0, + "step": 10679 + }, + { + "epoch": 1.358605775346648, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.917119026184082, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8858440518379211, + "num_tokens": 414823452.0, + "step": 10680 + }, + { + "epoch": 1.3587329856252386, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.898613929748535, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.881382405757904, + "num_tokens": 414870822.0, + "step": 10681 + }, + { + "epoch": 1.3588601959038291, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.926432609558105, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8933554887771606, + "num_tokens": 414904472.0, + "step": 10682 + }, + { + "epoch": 1.3589874061824196, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.009453773498535, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8841390013694763, + "num_tokens": 414941659.0, + "step": 10683 + }, + { + "epoch": 1.3591146164610102, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.048384666442871, + "learning_rate": 1e-06, + "loss": 0.3814, + "mean_token_accuracy": 0.8700948357582092, + "num_tokens": 414975359.0, + "step": 10684 + }, + { + "epoch": 1.3592418267396005, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.940627098083496, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8855594396591187, + "num_tokens": 415010406.0, + "step": 10685 + }, + { + "epoch": 1.359369037018191, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.84692668914795, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8900796175003052, + "num_tokens": 415049895.0, + "step": 10686 + }, + { + "epoch": 1.3594962472967815, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.984277725219727, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8825713396072388, + "num_tokens": 415088641.0, + "step": 10687 + }, + { + "epoch": 1.359623457575372, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.940443992614746, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8904428482055664, + "num_tokens": 415125320.0, + "step": 10688 + }, + { + "epoch": 1.3597506678539626, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.935901641845703, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8759416341781616, + "num_tokens": 415165415.0, + "step": 10689 + }, + { + "epoch": 1.3598778781325531, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.810213088989258, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8985445499420166, + "num_tokens": 415211041.0, + "step": 10690 + }, + { + "epoch": 1.3600050884111436, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.960987091064453, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8885292410850525, + "num_tokens": 415253199.0, + "step": 10691 + }, + { + "epoch": 1.3601322986897342, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.883646965026855, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8924583196640015, + "num_tokens": 415291049.0, + "step": 10692 + }, + { + "epoch": 1.3602595089683247, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.937646865844727, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8816344738006592, + "num_tokens": 415331118.0, + "step": 10693 + }, + { + "epoch": 1.3603867192469152, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.950383186340332, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8917003870010376, + "num_tokens": 415365644.0, + "step": 10694 + }, + { + "epoch": 1.3605139295255055, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.961112976074219, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8869103193283081, + "num_tokens": 415403120.0, + "step": 10695 + }, + { + "epoch": 1.360641139804096, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.855547904968262, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.895567774772644, + "num_tokens": 415448927.0, + "step": 10696 + }, + { + "epoch": 1.3607683500826866, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.863518714904785, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9019278883934021, + "num_tokens": 415489930.0, + "step": 10697 + }, + { + "epoch": 1.3608955603612771, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.090847969055176, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8909449577331543, + "num_tokens": 415522438.0, + "step": 10698 + }, + { + "epoch": 1.3610227706398677, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.90012264251709, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8990935683250427, + "num_tokens": 415559410.0, + "step": 10699 + }, + { + "epoch": 1.3611499809184582, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.9909029006958, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8755654692649841, + "num_tokens": 415596914.0, + "step": 10700 + }, + { + "epoch": 1.3612771911970487, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.921436309814453, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8923070430755615, + "num_tokens": 415633661.0, + "step": 10701 + }, + { + "epoch": 1.3614044014756392, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.810587882995605, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8957341313362122, + "num_tokens": 415678166.0, + "step": 10702 + }, + { + "epoch": 1.3615316117542298, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.960453987121582, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8901166915893555, + "num_tokens": 415721473.0, + "step": 10703 + }, + { + "epoch": 1.3616588220328203, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.977707862854004, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8760339021682739, + "num_tokens": 415760164.0, + "step": 10704 + }, + { + "epoch": 1.3617860323114108, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.919641494750977, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8811715841293335, + "num_tokens": 415801065.0, + "step": 10705 + }, + { + "epoch": 1.3619132425900013, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.991262435913086, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8776394128799438, + "num_tokens": 415837905.0, + "step": 10706 + }, + { + "epoch": 1.3620404528685919, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.914176940917969, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8925912380218506, + "num_tokens": 415877813.0, + "step": 10707 + }, + { + "epoch": 1.3621676631471824, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.993745803833008, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8825479745864868, + "num_tokens": 415913982.0, + "step": 10708 + }, + { + "epoch": 1.362294873425773, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.924480438232422, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8960338830947876, + "num_tokens": 415955108.0, + "step": 10709 + }, + { + "epoch": 1.3624220837043632, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.931760787963867, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8944271802902222, + "num_tokens": 415993591.0, + "step": 10710 + }, + { + "epoch": 1.3625492939829538, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.912275314331055, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8934302926063538, + "num_tokens": 416031195.0, + "step": 10711 + }, + { + "epoch": 1.3626765042615443, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.961806297302246, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8866485357284546, + "num_tokens": 416069899.0, + "step": 10712 + }, + { + "epoch": 1.3628037145401348, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.908049583435059, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8982464075088501, + "num_tokens": 416110000.0, + "step": 10713 + }, + { + "epoch": 1.3629309248187254, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.951722145080566, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8808931112289429, + "num_tokens": 416151972.0, + "step": 10714 + }, + { + "epoch": 1.3630581350973159, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.897899627685547, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.9038013219833374, + "num_tokens": 416195105.0, + "step": 10715 + }, + { + "epoch": 1.3631853453759064, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.916427612304688, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8818217515945435, + "num_tokens": 416238349.0, + "step": 10716 + }, + { + "epoch": 1.363312555654497, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.997790336608887, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8872988224029541, + "num_tokens": 416280576.0, + "step": 10717 + }, + { + "epoch": 1.3634397659330875, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.894829750061035, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.895802915096283, + "num_tokens": 416328574.0, + "step": 10718 + }, + { + "epoch": 1.3635669762116778, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.06231689453125, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8760315179824829, + "num_tokens": 416364020.0, + "step": 10719 + }, + { + "epoch": 1.3636941864902683, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.014161109924316, + "learning_rate": 1e-06, + "loss": 0.3889, + "mean_token_accuracy": 0.8653920888900757, + "num_tokens": 416403041.0, + "step": 10720 + }, + { + "epoch": 1.3638213967688588, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.988981246948242, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8882772922515869, + "num_tokens": 416440855.0, + "step": 10721 + }, + { + "epoch": 1.3639486070474494, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.8919677734375, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8946408033370972, + "num_tokens": 416476982.0, + "step": 10722 + }, + { + "epoch": 1.3640758173260399, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.020078659057617, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.875343918800354, + "num_tokens": 416516691.0, + "step": 10723 + }, + { + "epoch": 1.3642030276046304, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.91591739654541, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8925011157989502, + "num_tokens": 416554482.0, + "step": 10724 + }, + { + "epoch": 1.364330237883221, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.019920349121094, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8817123174667358, + "num_tokens": 416590016.0, + "step": 10725 + }, + { + "epoch": 1.3644574481618115, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.9071626663208, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8953663110733032, + "num_tokens": 416629366.0, + "step": 10726 + }, + { + "epoch": 1.364584658440402, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.932831764221191, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8837357759475708, + "num_tokens": 416668703.0, + "step": 10727 + }, + { + "epoch": 1.3647118687189925, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.979246139526367, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9021201133728027, + "num_tokens": 416698941.0, + "step": 10728 + }, + { + "epoch": 1.364839078997583, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.838346481323242, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8907337188720703, + "num_tokens": 416743287.0, + "step": 10729 + }, + { + "epoch": 1.3649662892761736, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.93844985961914, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8895861506462097, + "num_tokens": 416782810.0, + "step": 10730 + }, + { + "epoch": 1.3650934995547641, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.095088958740234, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8869493007659912, + "num_tokens": 416811507.0, + "step": 10731 + }, + { + "epoch": 1.3652207098333546, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.887391090393066, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8803558945655823, + "num_tokens": 416851191.0, + "step": 10732 + }, + { + "epoch": 1.3653479201119452, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.874063491821289, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.892918586730957, + "num_tokens": 416890175.0, + "step": 10733 + }, + { + "epoch": 1.3654751303905355, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.948089599609375, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8860681653022766, + "num_tokens": 416927600.0, + "step": 10734 + }, + { + "epoch": 1.365602340669126, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.942984580993652, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8845790028572083, + "num_tokens": 416965764.0, + "step": 10735 + }, + { + "epoch": 1.3657295509477165, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.89155387878418, + "learning_rate": 1e-06, + "loss": 0.3804, + "mean_token_accuracy": 0.8733741044998169, + "num_tokens": 417013306.0, + "step": 10736 + }, + { + "epoch": 1.365856761226307, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.9330415725708, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8858586549758911, + "num_tokens": 417051949.0, + "step": 10737 + }, + { + "epoch": 1.3659839715048976, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.969816207885742, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8840283751487732, + "num_tokens": 417088518.0, + "step": 10738 + }, + { + "epoch": 1.3661111817834881, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.030632019042969e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.979891777038574, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8884536623954773, + "num_tokens": 417127515.0, + "step": 10739 + }, + { + "epoch": 1.3662383920620786, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.913599967956543, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8950777053833008, + "num_tokens": 417168313.0, + "step": 10740 + }, + { + "epoch": 1.3663656023406692, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.03061580657959, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8935506939888, + "num_tokens": 417202086.0, + "step": 10741 + }, + { + "epoch": 1.3664928126192597, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.892629623413086, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8883594870567322, + "num_tokens": 417242657.0, + "step": 10742 + }, + { + "epoch": 1.3666200228978502, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.859094619750977, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8898255825042725, + "num_tokens": 417287702.0, + "step": 10743 + }, + { + "epoch": 1.3667472331764405, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.080780982971191, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8880940079689026, + "num_tokens": 417320787.0, + "step": 10744 + }, + { + "epoch": 1.366874443455031, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.858819961547852, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8973820209503174, + "num_tokens": 417362244.0, + "step": 10745 + }, + { + "epoch": 1.3670016537336216, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.976141929626465, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8858765363693237, + "num_tokens": 417400066.0, + "step": 10746 + }, + { + "epoch": 1.3671288640122121, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.055127143859863, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.8719111084938049, + "num_tokens": 417441677.0, + "step": 10747 + }, + { + "epoch": 1.3672560742908026, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.919931411743164, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8889124989509583, + "num_tokens": 417481830.0, + "step": 10748 + }, + { + "epoch": 1.3673832845693932, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.881786346435547, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.883030891418457, + "num_tokens": 417522499.0, + "step": 10749 + }, + { + "epoch": 1.3675104948479837, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.924060821533203, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8932753205299377, + "num_tokens": 417563503.0, + "step": 10750 + }, + { + "epoch": 1.3676377051265742, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.96817684173584, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8864808082580566, + "num_tokens": 417606438.0, + "step": 10751 + }, + { + "epoch": 1.3677649154051648, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.914653778076172, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8911566734313965, + "num_tokens": 417651582.0, + "step": 10752 + }, + { + "epoch": 1.3678921256837553, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.853293418884277, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.8709765672683716, + "num_tokens": 417699347.0, + "step": 10753 + }, + { + "epoch": 1.3680193359623458, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.99892807006836, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.8732120990753174, + "num_tokens": 417737552.0, + "step": 10754 + }, + { + "epoch": 1.3681465462409363, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.062394142150879, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8823180198669434, + "num_tokens": 417768071.0, + "step": 10755 + }, + { + "epoch": 1.3682737565195269, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.940057754516602, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8944987058639526, + "num_tokens": 417805170.0, + "step": 10756 + }, + { + "epoch": 1.3684009667981174, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.978649139404297, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8889436721801758, + "num_tokens": 417843522.0, + "step": 10757 + }, + { + "epoch": 1.368528177076708, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.974675178527832, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8796800374984741, + "num_tokens": 417879054.0, + "step": 10758 + }, + { + "epoch": 1.3686553873552982, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.958099365234375, + "learning_rate": 1e-06, + "loss": 0.3896, + "mean_token_accuracy": 0.8690078854560852, + "num_tokens": 417922676.0, + "step": 10759 + }, + { + "epoch": 1.3687825976338888, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.940653800964355, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8877015113830566, + "num_tokens": 417961072.0, + "step": 10760 + }, + { + "epoch": 1.3689098079124793, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.86913013458252, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8983503580093384, + "num_tokens": 418000242.0, + "step": 10761 + }, + { + "epoch": 1.3690370181910698, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.968439102172852, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8945925831794739, + "num_tokens": 418039779.0, + "step": 10762 + }, + { + "epoch": 1.3691642284696603, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.908062934875488, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8813345432281494, + "num_tokens": 418078762.0, + "step": 10763 + }, + { + "epoch": 1.3692914387482509, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.917013168334961, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8910428881645203, + "num_tokens": 418118742.0, + "step": 10764 + }, + { + "epoch": 1.3694186490268414, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.908387184143066, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8889145851135254, + "num_tokens": 418159672.0, + "step": 10765 + }, + { + "epoch": 1.369545859305432, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.960780143737793, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.894524097442627, + "num_tokens": 418192362.0, + "step": 10766 + }, + { + "epoch": 1.3696730695840225, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.022796630859375, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8730534315109253, + "num_tokens": 418232663.0, + "step": 10767 + }, + { + "epoch": 1.3698002798626128, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.95397663116455, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8984737396240234, + "num_tokens": 418267923.0, + "step": 10768 + }, + { + "epoch": 1.3699274901412033, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.992130279541016, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8815792798995972, + "num_tokens": 418304218.0, + "step": 10769 + }, + { + "epoch": 1.3700547004197938, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.058019638061523, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8912397623062134, + "num_tokens": 418343006.0, + "step": 10770 + }, + { + "epoch": 1.3701819106983844, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.027265548706055, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8830425143241882, + "num_tokens": 418380788.0, + "step": 10771 + }, + { + "epoch": 1.3703091209769749, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.87571907043457, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8901907205581665, + "num_tokens": 418422387.0, + "step": 10772 + }, + { + "epoch": 1.3704363312555654, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.007475852966309, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8727903366088867, + "num_tokens": 418460032.0, + "step": 10773 + }, + { + "epoch": 1.370563541534156, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.910873413085938, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8846009373664856, + "num_tokens": 418501372.0, + "step": 10774 + }, + { + "epoch": 1.3706907518127465, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.029932975769043, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8892590999603271, + "num_tokens": 418535203.0, + "step": 10775 + }, + { + "epoch": 1.370817962091337, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.926591873168945, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9088448286056519, + "num_tokens": 418572661.0, + "step": 10776 + }, + { + "epoch": 1.3709451723699275, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.975286483764648, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8861355185508728, + "num_tokens": 418614129.0, + "step": 10777 + }, + { + "epoch": 1.371072382648518, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.964369773864746, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8766578435897827, + "num_tokens": 418659681.0, + "step": 10778 + }, + { + "epoch": 1.3711995929271086, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.986590385437012, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8868503570556641, + "num_tokens": 418695067.0, + "step": 10779 + }, + { + "epoch": 1.371326803205699, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.942479133605957, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8917640447616577, + "num_tokens": 418734490.0, + "step": 10780 + }, + { + "epoch": 1.3714540134842896, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.984976768493652, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8869684934616089, + "num_tokens": 418774049.0, + "step": 10781 + }, + { + "epoch": 1.3715812237628802, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.91925048828125, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.90151447057724, + "num_tokens": 418809886.0, + "step": 10782 + }, + { + "epoch": 1.3717084340414705, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.964896202087402, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8932629227638245, + "num_tokens": 418849633.0, + "step": 10783 + }, + { + "epoch": 1.371835644320061, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.947948455810547, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8891041278839111, + "num_tokens": 418888157.0, + "step": 10784 + }, + { + "epoch": 1.3719628545986515, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.877686500549316, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8932616114616394, + "num_tokens": 418930911.0, + "step": 10785 + }, + { + "epoch": 1.372090064877242, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.970385551452637, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8815611600875854, + "num_tokens": 418969875.0, + "step": 10786 + }, + { + "epoch": 1.3722172751558326, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.96579360961914, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8983889818191528, + "num_tokens": 419005393.0, + "step": 10787 + }, + { + "epoch": 1.3723444854344231, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.982471466064453, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8871920108795166, + "num_tokens": 419047011.0, + "step": 10788 + }, + { + "epoch": 1.3724716957130136, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.008007049560547, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8889524340629578, + "num_tokens": 419083166.0, + "step": 10789 + }, + { + "epoch": 1.3725989059916042, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.958653450012207, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8960177898406982, + "num_tokens": 419117906.0, + "step": 10790 + }, + { + "epoch": 1.3727261162701947, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.940841674804688, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8870981931686401, + "num_tokens": 419158321.0, + "step": 10791 + }, + { + "epoch": 1.3728533265487852, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.954185485839844, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8887907266616821, + "num_tokens": 419196444.0, + "step": 10792 + }, + { + "epoch": 1.3729805368273755, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.940577507019043, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8827419281005859, + "num_tokens": 419234827.0, + "step": 10793 + }, + { + "epoch": 1.373107747105966, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.954835891723633, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8972090482711792, + "num_tokens": 419270828.0, + "step": 10794 + }, + { + "epoch": 1.3732349573845566, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.969228744506836, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.9001457691192627, + "num_tokens": 419305657.0, + "step": 10795 + }, + { + "epoch": 1.3733621676631471, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.838142395019531, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8944424390792847, + "num_tokens": 419346692.0, + "step": 10796 + }, + { + "epoch": 1.3734893779417376, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.90465259552002, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8924964666366577, + "num_tokens": 419392847.0, + "step": 10797 + }, + { + "epoch": 1.3736165882203282, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.945809364318848, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8902415037155151, + "num_tokens": 419435323.0, + "step": 10798 + }, + { + "epoch": 1.3737437984989187, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.936178207397461, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8787590861320496, + "num_tokens": 419475247.0, + "step": 10799 + }, + { + "epoch": 1.3738710087775092, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.972612380981445, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8921856880187988, + "num_tokens": 419510278.0, + "step": 10800 + }, + { + "epoch": 1.3739982190560998, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.054473876953125e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.98977279663086, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8822348713874817, + "num_tokens": 419548329.0, + "step": 10801 + }, + { + "epoch": 1.3741254293346903, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.931832313537598, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8892303705215454, + "num_tokens": 419589710.0, + "step": 10802 + }, + { + "epoch": 1.3742526396132808, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.977486610412598, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8907305002212524, + "num_tokens": 419629326.0, + "step": 10803 + }, + { + "epoch": 1.3743798498918713, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.009161949157715, + "learning_rate": 1e-06, + "loss": 0.3869, + "mean_token_accuracy": 0.8692299127578735, + "num_tokens": 419668671.0, + "step": 10804 + }, + { + "epoch": 1.3745070601704619, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.937061309814453, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8862867951393127, + "num_tokens": 419710261.0, + "step": 10805 + }, + { + "epoch": 1.3746342704490524, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.972370147705078, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8943515419960022, + "num_tokens": 419744136.0, + "step": 10806 + }, + { + "epoch": 1.374761480727643, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.000420570373535, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8840233087539673, + "num_tokens": 419781250.0, + "step": 10807 + }, + { + "epoch": 1.3748886910062332, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.883471488952637, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8955775499343872, + "num_tokens": 419823556.0, + "step": 10808 + }, + { + "epoch": 1.3750159012848238, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.012289047241211, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8976205587387085, + "num_tokens": 419855587.0, + "step": 10809 + }, + { + "epoch": 1.3751431115634143, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.013440132141113, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8874688148498535, + "num_tokens": 419889309.0, + "step": 10810 + }, + { + "epoch": 1.3752703218420048, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.874298095703125, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8863002061843872, + "num_tokens": 419934676.0, + "step": 10811 + }, + { + "epoch": 1.3753975321205953, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.928519248962402, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.899080753326416, + "num_tokens": 419972266.0, + "step": 10812 + }, + { + "epoch": 1.3755247423991859, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.961837768554688, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8849666118621826, + "num_tokens": 420013375.0, + "step": 10813 + }, + { + "epoch": 1.3756519526777764, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.912714004516602, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8835436701774597, + "num_tokens": 420059788.0, + "step": 10814 + }, + { + "epoch": 1.375779162956367, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.848875999450684, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9078966379165649, + "num_tokens": 420097782.0, + "step": 10815 + }, + { + "epoch": 1.3759063732349575, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.009797096252441, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8920767307281494, + "num_tokens": 420133726.0, + "step": 10816 + }, + { + "epoch": 1.3760335835135478, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.95593547821045, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8864963054656982, + "num_tokens": 420171932.0, + "step": 10817 + }, + { + "epoch": 1.3761607937921383, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.919452667236328, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.88883376121521, + "num_tokens": 420213124.0, + "step": 10818 + }, + { + "epoch": 1.3762880040707288, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.082494735717773, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8730690479278564, + "num_tokens": 420252287.0, + "step": 10819 + }, + { + "epoch": 1.3764152143493193, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.962118148803711, + "learning_rate": 1e-06, + "loss": 0.3804, + "mean_token_accuracy": 0.8743058443069458, + "num_tokens": 420292700.0, + "step": 10820 + }, + { + "epoch": 1.3765424246279099, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.965400695800781, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8813843727111816, + "num_tokens": 420332822.0, + "step": 10821 + }, + { + "epoch": 1.3766696349065004, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.006356239318848, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8843575716018677, + "num_tokens": 420367729.0, + "step": 10822 + }, + { + "epoch": 1.376796845185091, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.871025085449219, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8847804069519043, + "num_tokens": 420415367.0, + "step": 10823 + }, + { + "epoch": 1.3769240554636815, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.0419340133667, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.889667272567749, + "num_tokens": 420454443.0, + "step": 10824 + }, + { + "epoch": 1.377051265742272, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.991018295288086, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8847088813781738, + "num_tokens": 420495394.0, + "step": 10825 + }, + { + "epoch": 1.3771784760208625, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.811090469360352, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8869705200195312, + "num_tokens": 420539548.0, + "step": 10826 + }, + { + "epoch": 1.377305686299453, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.053138732910156, + "learning_rate": 1e-06, + "loss": 0.3744, + "mean_token_accuracy": 0.8712413311004639, + "num_tokens": 420576874.0, + "step": 10827 + }, + { + "epoch": 1.3774328965780436, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.951889991760254, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8915806412696838, + "num_tokens": 420613544.0, + "step": 10828 + }, + { + "epoch": 1.377560106856634, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.892094612121582, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.9035772085189819, + "num_tokens": 420651849.0, + "step": 10829 + }, + { + "epoch": 1.3776873171352246, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.976518630981445, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8899580240249634, + "num_tokens": 420695428.0, + "step": 10830 + }, + { + "epoch": 1.3778145274138152, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.27950382232666, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8976686596870422, + "num_tokens": 420729856.0, + "step": 10831 + }, + { + "epoch": 1.3779417376924055, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.923839569091797, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8781219720840454, + "num_tokens": 420764591.0, + "step": 10832 + }, + { + "epoch": 1.378068947970996, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.952980995178223, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8978252410888672, + "num_tokens": 420801483.0, + "step": 10833 + }, + { + "epoch": 1.3781961582495865, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.91103458404541, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8869724273681641, + "num_tokens": 420838363.0, + "step": 10834 + }, + { + "epoch": 1.378323368528177, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.09498119354248, + "learning_rate": 1e-06, + "loss": 0.3998, + "mean_token_accuracy": 0.8626490831375122, + "num_tokens": 420872276.0, + "step": 10835 + }, + { + "epoch": 1.3784505788067676, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.053997039794922, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.87373286485672, + "num_tokens": 420909138.0, + "step": 10836 + }, + { + "epoch": 1.378577789085358, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.853324890136719, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8796987533569336, + "num_tokens": 420949381.0, + "step": 10837 + }, + { + "epoch": 1.3787049993639486, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.869787216186523, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9001294374465942, + "num_tokens": 420991142.0, + "step": 10838 + }, + { + "epoch": 1.3788322096425392, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.865528106689453, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8946434855461121, + "num_tokens": 421033270.0, + "step": 10839 + }, + { + "epoch": 1.3789594199211297, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.952739715576172, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8918024897575378, + "num_tokens": 421069916.0, + "step": 10840 + }, + { + "epoch": 1.3790866301997202, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.919182777404785, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.904836118221283, + "num_tokens": 421110544.0, + "step": 10841 + }, + { + "epoch": 1.3792138404783105, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.966116905212402, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8956907391548157, + "num_tokens": 421146323.0, + "step": 10842 + }, + { + "epoch": 1.379341050756901, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.996172904968262, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8937680125236511, + "num_tokens": 421179381.0, + "step": 10843 + }, + { + "epoch": 1.3794682610354916, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.02778434753418, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8810446262359619, + "num_tokens": 421213029.0, + "step": 10844 + }, + { + "epoch": 1.3795954713140821, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.977106094360352, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8822905421257019, + "num_tokens": 421245746.0, + "step": 10845 + }, + { + "epoch": 1.3797226815926726, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.862037658691406, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8945253491401672, + "num_tokens": 421284909.0, + "step": 10846 + }, + { + "epoch": 1.3798498918712632, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.945557594299316, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8812352418899536, + "num_tokens": 421323316.0, + "step": 10847 + }, + { + "epoch": 1.3799771021498537, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.982213973999023, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8763322830200195, + "num_tokens": 421363096.0, + "step": 10848 + }, + { + "epoch": 1.3801043124284442, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.100834846496582, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8820051550865173, + "num_tokens": 421392380.0, + "step": 10849 + }, + { + "epoch": 1.3802315227070348, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.831788063049316, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8916988372802734, + "num_tokens": 421438443.0, + "step": 10850 + }, + { + "epoch": 1.3803587329856253, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.952970504760742, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8767348527908325, + "num_tokens": 421479096.0, + "step": 10851 + }, + { + "epoch": 1.3804859432642158, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.93453311920166, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8906475901603699, + "num_tokens": 421515157.0, + "step": 10852 + }, + { + "epoch": 1.3806131535428063, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.01890754699707, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8850458860397339, + "num_tokens": 421550793.0, + "step": 10853 + }, + { + "epoch": 1.3807403638213969, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.88589096069336, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8869062662124634, + "num_tokens": 421587654.0, + "step": 10854 + }, + { + "epoch": 1.3808675740999874, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.076416015625, + "learning_rate": 1e-06, + "loss": 0.4176, + "mean_token_accuracy": 0.8584050536155701, + "num_tokens": 421625313.0, + "step": 10855 + }, + { + "epoch": 1.380994784378578, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.908923149108887, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.887412428855896, + "num_tokens": 421667467.0, + "step": 10856 + }, + { + "epoch": 1.3811219946571682, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.866707801818848, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8923009037971497, + "num_tokens": 421710604.0, + "step": 10857 + }, + { + "epoch": 1.3812492049357588, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.955116271972656, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8938926458358765, + "num_tokens": 421755240.0, + "step": 10858 + }, + { + "epoch": 1.3813764152143493, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.89829158782959, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8969055414199829, + "num_tokens": 421795103.0, + "step": 10859 + }, + { + "epoch": 1.3815036254929398, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.061423301696777, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8738092184066772, + "num_tokens": 421836531.0, + "step": 10860 + }, + { + "epoch": 1.3816308357715303, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.057652473449707, + "learning_rate": 1e-06, + "loss": 0.4065, + "mean_token_accuracy": 0.8615884780883789, + "num_tokens": 421878873.0, + "step": 10861 + }, + { + "epoch": 1.3817580460501209, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.001859664916992, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8932371735572815, + "num_tokens": 421911229.0, + "step": 10862 + }, + { + "epoch": 1.3818852563287114, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.971572875976562, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.896072268486023, + "num_tokens": 421943727.0, + "step": 10863 + }, + { + "epoch": 1.382012466607302, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.906946182250977, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8899514675140381, + "num_tokens": 421986478.0, + "step": 10864 + }, + { + "epoch": 1.3821396768858925, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.1021575927734375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.945854187011719, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8914111852645874, + "num_tokens": 422027495.0, + "step": 10865 + }, + { + "epoch": 1.3822668871644828, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.987424850463867, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8866376876831055, + "num_tokens": 422062790.0, + "step": 10866 + }, + { + "epoch": 1.3823940974430733, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.063199996948242, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8905229568481445, + "num_tokens": 422095422.0, + "step": 10867 + }, + { + "epoch": 1.3825213077216638, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.904121398925781, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8938069939613342, + "num_tokens": 422138290.0, + "step": 10868 + }, + { + "epoch": 1.3826485180002543, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.055033683776855, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8787763118743896, + "num_tokens": 422174976.0, + "step": 10869 + }, + { + "epoch": 1.3827757282788449, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.890222549438477, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8986674547195435, + "num_tokens": 422216429.0, + "step": 10870 + }, + { + "epoch": 1.3829029385574354, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.918130874633789, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8898674845695496, + "num_tokens": 422260051.0, + "step": 10871 + }, + { + "epoch": 1.383030148836026, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.853290557861328, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.9107080698013306, + "num_tokens": 422299674.0, + "step": 10872 + }, + { + "epoch": 1.3831573591146165, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.966960906982422, + "learning_rate": 1e-06, + "loss": 0.3856, + "mean_token_accuracy": 0.8711713552474976, + "num_tokens": 422345945.0, + "step": 10873 + }, + { + "epoch": 1.383284569393207, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.07062816619873, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8847605586051941, + "num_tokens": 422381160.0, + "step": 10874 + }, + { + "epoch": 1.3834117796717975, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.897435188293457, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8900569677352905, + "num_tokens": 422431224.0, + "step": 10875 + }, + { + "epoch": 1.383538989950388, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.907646179199219, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.897626519203186, + "num_tokens": 422471894.0, + "step": 10876 + }, + { + "epoch": 1.3836662002289786, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.943947792053223, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8894725441932678, + "num_tokens": 422512966.0, + "step": 10877 + }, + { + "epoch": 1.383793410507569, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.075934410095215, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8883463144302368, + "num_tokens": 422543648.0, + "step": 10878 + }, + { + "epoch": 1.3839206207861596, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.98019790649414, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8904917240142822, + "num_tokens": 422579108.0, + "step": 10879 + }, + { + "epoch": 1.3840478310647502, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.93251895904541, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8876276612281799, + "num_tokens": 422622458.0, + "step": 10880 + }, + { + "epoch": 1.3841750413433405, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.879792213439941, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8924657106399536, + "num_tokens": 422665209.0, + "step": 10881 + }, + { + "epoch": 1.384302251621931, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.944263458251953, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8966541290283203, + "num_tokens": 422703337.0, + "step": 10882 + }, + { + "epoch": 1.3844294619005215, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.891737937927246, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8985543847084045, + "num_tokens": 422740067.0, + "step": 10883 + }, + { + "epoch": 1.384556672179112, + "ewc_loss": 0.01513671875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0013580322265625e-05, + "grad_norm": 8.941523551940918, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8835791349411011, + "num_tokens": 422778862.0, + "step": 10884 + }, + { + "epoch": 1.3846838824577026, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.943328857421875, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.895506739616394, + "num_tokens": 422817346.0, + "step": 10885 + }, + { + "epoch": 1.384811092736293, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.947572708129883, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8844258785247803, + "num_tokens": 422857809.0, + "step": 10886 + }, + { + "epoch": 1.3849383030148836, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.914639472961426, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8837703466415405, + "num_tokens": 422896758.0, + "step": 10887 + }, + { + "epoch": 1.3850655132934742, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.903098106384277, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8911473751068115, + "num_tokens": 422937070.0, + "step": 10888 + }, + { + "epoch": 1.3851927235720647, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.967536926269531, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.8753733038902283, + "num_tokens": 422976929.0, + "step": 10889 + }, + { + "epoch": 1.385319933850655, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.963764190673828, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.9028280973434448, + "num_tokens": 423013604.0, + "step": 10890 + }, + { + "epoch": 1.3854471441292455, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.049101829528809, + "learning_rate": 1e-06, + "loss": 0.4082, + "mean_token_accuracy": 0.8632237911224365, + "num_tokens": 423052374.0, + "step": 10891 + }, + { + "epoch": 1.385574354407836, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.905431747436523, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8920581936836243, + "num_tokens": 423095266.0, + "step": 10892 + }, + { + "epoch": 1.3857015646864266, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.922114372253418, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9023547768592834, + "num_tokens": 423134887.0, + "step": 10893 + }, + { + "epoch": 1.385828774965017, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.879410743713379, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8823598623275757, + "num_tokens": 423177984.0, + "step": 10894 + }, + { + "epoch": 1.3859559852436076, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.969258308410645, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8930920958518982, + "num_tokens": 423214703.0, + "step": 10895 + }, + { + "epoch": 1.3860831955221982, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.029915809631348, + "learning_rate": 1e-06, + "loss": 0.3717, + "mean_token_accuracy": 0.8734217882156372, + "num_tokens": 423254055.0, + "step": 10896 + }, + { + "epoch": 1.3862104058007887, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.891839981079102, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8940726518630981, + "num_tokens": 423296079.0, + "step": 10897 + }, + { + "epoch": 1.3863376160793792, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.958955764770508, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8951051831245422, + "num_tokens": 423333395.0, + "step": 10898 + }, + { + "epoch": 1.3864648263579697, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.938180923461914, + "learning_rate": 1e-06, + "loss": 0.3771, + "mean_token_accuracy": 0.8755008578300476, + "num_tokens": 423374883.0, + "step": 10899 + }, + { + "epoch": 1.3865920366365603, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.107938766479492, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.879753828048706, + "num_tokens": 423406437.0, + "step": 10900 + }, + { + "epoch": 1.3867192469151508, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.023567199707031, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8971805572509766, + "num_tokens": 423443049.0, + "step": 10901 + }, + { + "epoch": 1.3868464571937413, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.017269134521484, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8829289078712463, + "num_tokens": 423476384.0, + "step": 10902 + }, + { + "epoch": 1.3869736674723319, + "ewc_loss": 0.01519775390625, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.074868202209473, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8780725598335266, + "num_tokens": 423513157.0, + "step": 10903 + }, + { + "epoch": 1.3871008777509224, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.947402000427246, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8907006978988647, + "num_tokens": 423550567.0, + "step": 10904 + }, + { + "epoch": 1.387228088029513, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.960712432861328, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8848161697387695, + "num_tokens": 423587489.0, + "step": 10905 + }, + { + "epoch": 1.3873552983081032, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.972570419311523, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8820430636405945, + "num_tokens": 423629883.0, + "step": 10906 + }, + { + "epoch": 1.3874825085866938, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.911128997802734, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8869035243988037, + "num_tokens": 423668134.0, + "step": 10907 + }, + { + "epoch": 1.3876097188652843, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.069585800170898, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8834660053253174, + "num_tokens": 423704517.0, + "step": 10908 + }, + { + "epoch": 1.3877369291438748, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.986945152282715, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8897488117218018, + "num_tokens": 423739266.0, + "step": 10909 + }, + { + "epoch": 1.3878641394224653, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.101318359375, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8712767958641052, + "num_tokens": 423772468.0, + "step": 10910 + }, + { + "epoch": 1.3879913497010559, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.983497619628906, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8772825002670288, + "num_tokens": 423812683.0, + "step": 10911 + }, + { + "epoch": 1.3881185599796464, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.07372760772705, + "learning_rate": 1e-06, + "loss": 0.3893, + "mean_token_accuracy": 0.8719444274902344, + "num_tokens": 423848295.0, + "step": 10912 + }, + { + "epoch": 1.388245770258237, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.883130073547363, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8887761831283569, + "num_tokens": 423890153.0, + "step": 10913 + }, + { + "epoch": 1.3883729805368275, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.985133171081543, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8900742530822754, + "num_tokens": 423926409.0, + "step": 10914 + }, + { + "epoch": 1.3885001908154178, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.974248886108398, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8953260183334351, + "num_tokens": 423960352.0, + "step": 10915 + }, + { + "epoch": 1.3886274010940083, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.125999450683594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.020069122314453, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8866837024688721, + "num_tokens": 423998082.0, + "step": 10916 + }, + { + "epoch": 1.3887546113725988, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.945898056030273, + "learning_rate": 1e-06, + "loss": 0.384, + "mean_token_accuracy": 0.8697457909584045, + "num_tokens": 424040291.0, + "step": 10917 + }, + { + "epoch": 1.3888818216511893, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.038002967834473, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8849448561668396, + "num_tokens": 424075131.0, + "step": 10918 + }, + { + "epoch": 1.3890090319297799, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.004952430725098, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8893242478370667, + "num_tokens": 424109856.0, + "step": 10919 + }, + { + "epoch": 1.3891362422083704, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.851153373718262, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8971803784370422, + "num_tokens": 424153301.0, + "step": 10920 + }, + { + "epoch": 1.389263452486961, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.974710464477539, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8872298002243042, + "num_tokens": 424192926.0, + "step": 10921 + }, + { + "epoch": 1.3893906627655515, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.0189790725708, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8830893039703369, + "num_tokens": 424229939.0, + "step": 10922 + }, + { + "epoch": 1.389517873044142, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.949995040893555, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8953423500061035, + "num_tokens": 424265152.0, + "step": 10923 + }, + { + "epoch": 1.3896450833227325, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.94477653503418, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8939296007156372, + "num_tokens": 424301967.0, + "step": 10924 + }, + { + "epoch": 1.389772293601323, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.906530380249023, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8910857439041138, + "num_tokens": 424343344.0, + "step": 10925 + }, + { + "epoch": 1.3898995038799136, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.983057975769043, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8969507217407227, + "num_tokens": 424377130.0, + "step": 10926 + }, + { + "epoch": 1.390026714158504, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.87460708618164, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8885525465011597, + "num_tokens": 424419953.0, + "step": 10927 + }, + { + "epoch": 1.3901539244370946, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.96584701538086, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8957861661911011, + "num_tokens": 424456553.0, + "step": 10928 + }, + { + "epoch": 1.3902811347156852, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.936347961425781, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8810815215110779, + "num_tokens": 424495973.0, + "step": 10929 + }, + { + "epoch": 1.3904083449942755, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.055076599121094, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8763278722763062, + "num_tokens": 424531288.0, + "step": 10930 + }, + { + "epoch": 1.390535555272866, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.964738845825195, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8920422792434692, + "num_tokens": 424567621.0, + "step": 10931 + }, + { + "epoch": 1.3906627655514565, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.950807571411133, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8810153007507324, + "num_tokens": 424606603.0, + "step": 10932 + }, + { + "epoch": 1.390789975830047, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.973077774047852, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8885661363601685, + "num_tokens": 424645055.0, + "step": 10933 + }, + { + "epoch": 1.3909171861086376, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.975044250488281, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8852487802505493, + "num_tokens": 424683766.0, + "step": 10934 + }, + { + "epoch": 1.391044396387228, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.974787712097168, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.88675457239151, + "num_tokens": 424719907.0, + "step": 10935 + }, + { + "epoch": 1.3911716066658186, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.117446899414062, + "learning_rate": 1e-06, + "loss": 0.4045, + "mean_token_accuracy": 0.8622463941574097, + "num_tokens": 424756207.0, + "step": 10936 + }, + { + "epoch": 1.3912988169444092, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.039839744567871, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8782852292060852, + "num_tokens": 424797988.0, + "step": 10937 + }, + { + "epoch": 1.3914260272229997, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.940499305725098, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8872898817062378, + "num_tokens": 424838983.0, + "step": 10938 + }, + { + "epoch": 1.39155323750159, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.943865776062012, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8881288766860962, + "num_tokens": 424882093.0, + "step": 10939 + }, + { + "epoch": 1.3916804477801805, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.072749137878418, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8778384327888489, + "num_tokens": 424917428.0, + "step": 10940 + }, + { + "epoch": 1.391807658058771, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.013786315917969, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8894705176353455, + "num_tokens": 424953063.0, + "step": 10941 + }, + { + "epoch": 1.3919348683373616, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.023322105407715, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.878312349319458, + "num_tokens": 424992053.0, + "step": 10942 + }, + { + "epoch": 1.392062078615952, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.983878135681152, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8750891089439392, + "num_tokens": 425035098.0, + "step": 10943 + }, + { + "epoch": 1.3921892888945426, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.066771507263184, + "learning_rate": 1e-06, + "loss": 0.3783, + "mean_token_accuracy": 0.8741273283958435, + "num_tokens": 425077137.0, + "step": 10944 + }, + { + "epoch": 1.3923164991731332, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.009188652038574, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8799232244491577, + "num_tokens": 425117931.0, + "step": 10945 + }, + { + "epoch": 1.3924437094517237, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.01773452758789, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8758370876312256, + "num_tokens": 425156871.0, + "step": 10946 + }, + { + "epoch": 1.3925709197303142, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.033110618591309, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8904209136962891, + "num_tokens": 425192271.0, + "step": 10947 + }, + { + "epoch": 1.3926981300089047, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.955031394958496, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8806774616241455, + "num_tokens": 425238294.0, + "step": 10948 + }, + { + "epoch": 1.3928253402874953, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.941028594970703, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8895533680915833, + "num_tokens": 425280780.0, + "step": 10949 + }, + { + "epoch": 1.3929525505660858, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.013701438903809, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8895025253295898, + "num_tokens": 425317857.0, + "step": 10950 + }, + { + "epoch": 1.3930797608446763, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.927851676940918, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8833959102630615, + "num_tokens": 425362430.0, + "step": 10951 + }, + { + "epoch": 1.3932069711232669, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.989052772521973, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8898500800132751, + "num_tokens": 425403293.0, + "step": 10952 + }, + { + "epoch": 1.3933341814018574, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.930289268493652, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8855558633804321, + "num_tokens": 425451019.0, + "step": 10953 + }, + { + "epoch": 1.393461391680448, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.009603500366211, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.880699098110199, + "num_tokens": 425495041.0, + "step": 10954 + }, + { + "epoch": 1.3935886019590382, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.932047843933105, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9060071706771851, + "num_tokens": 425529562.0, + "step": 10955 + }, + { + "epoch": 1.3937158122376287, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.030827522277832, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8912106156349182, + "num_tokens": 425568363.0, + "step": 10956 + }, + { + "epoch": 1.3938430225162193, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.014973640441895, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8929778337478638, + "num_tokens": 425603539.0, + "step": 10957 + }, + { + "epoch": 1.3939702327948098, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.929140090942383, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8926730155944824, + "num_tokens": 425643233.0, + "step": 10958 + }, + { + "epoch": 1.3940974430734003, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.940657615661621, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8959883451461792, + "num_tokens": 425684776.0, + "step": 10959 + }, + { + "epoch": 1.3942246533519909, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 8.952754020690918, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8865411877632141, + "num_tokens": 425726042.0, + "step": 10960 + }, + { + "epoch": 1.3943518636305814, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.000452041625977, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8863790035247803, + "num_tokens": 425766255.0, + "step": 10961 + }, + { + "epoch": 1.394479073909172, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.9332914352417, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8828048706054688, + "num_tokens": 425810347.0, + "step": 10962 + }, + { + "epoch": 1.3946062841877624, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.903251647949219, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8895186185836792, + "num_tokens": 425856364.0, + "step": 10963 + }, + { + "epoch": 1.3947334944663528, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.145041465759277, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8818715214729309, + "num_tokens": 425890355.0, + "step": 10964 + }, + { + "epoch": 1.3948607047449433, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.94723892211914, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8902413249015808, + "num_tokens": 425930017.0, + "step": 10965 + }, + { + "epoch": 1.3949879150235338, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.061296463012695, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8831644058227539, + "num_tokens": 425964034.0, + "step": 10966 + }, + { + "epoch": 1.3951151253021243, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.149681091308594, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8848478198051453, + "num_tokens": 425993849.0, + "step": 10967 + }, + { + "epoch": 1.3952423355807149, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.974629402160645, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.8748548626899719, + "num_tokens": 426032412.0, + "step": 10968 + }, + { + "epoch": 1.3953695458593054, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.955917358398438, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8888363838195801, + "num_tokens": 426069636.0, + "step": 10969 + }, + { + "epoch": 1.395496756137896, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.080544471740723, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8793423771858215, + "num_tokens": 426109093.0, + "step": 10970 + }, + { + "epoch": 1.3956239664164865, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.047201156616211, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8944191932678223, + "num_tokens": 426140212.0, + "step": 10971 + }, + { + "epoch": 1.395751176695077, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.029265403747559, + "learning_rate": 1e-06, + "loss": 0.3954, + "mean_token_accuracy": 0.8648257851600647, + "num_tokens": 426178651.0, + "step": 10972 + }, + { + "epoch": 1.3958783869736675, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.004444122314453, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8989928960800171, + "num_tokens": 426213846.0, + "step": 10973 + }, + { + "epoch": 1.396005597252258, + "ewc_loss": 0.0152587890625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.006999969482422, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8943029642105103, + "num_tokens": 426247637.0, + "step": 10974 + }, + { + "epoch": 1.3961328075308486, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.118585586547852, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8752366900444031, + "num_tokens": 426281196.0, + "step": 10975 + }, + { + "epoch": 1.396260017809439, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.084257125854492, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8773756623268127, + "num_tokens": 426313464.0, + "step": 10976 + }, + { + "epoch": 1.3963872280880296, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.007664680480957, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8891956806182861, + "num_tokens": 426349453.0, + "step": 10977 + }, + { + "epoch": 1.3965144383666201, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.034335136413574, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.87868332862854, + "num_tokens": 426388579.0, + "step": 10978 + }, + { + "epoch": 1.3966416486452105, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.02862548828125, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8762511610984802, + "num_tokens": 426425530.0, + "step": 10979 + }, + { + "epoch": 1.396768858923801, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.069014549255371, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8827773332595825, + "num_tokens": 426461862.0, + "step": 10980 + }, + { + "epoch": 1.3968960692023915, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.969331741333008, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8871134519577026, + "num_tokens": 426502189.0, + "step": 10981 + }, + { + "epoch": 1.397023279480982, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.925825119018555, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8919118642807007, + "num_tokens": 426546642.0, + "step": 10982 + }, + { + "epoch": 1.3971504897595726, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.030464172363281, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8932345509529114, + "num_tokens": 426582887.0, + "step": 10983 + }, + { + "epoch": 1.397277700038163, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.011335372924805, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8985909819602966, + "num_tokens": 426620918.0, + "step": 10984 + }, + { + "epoch": 1.3974049103167536, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.084551811218262, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8922663927078247, + "num_tokens": 426651203.0, + "step": 10985 + }, + { + "epoch": 1.3975321205953442, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.970795631408691, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8993737101554871, + "num_tokens": 426690001.0, + "step": 10986 + }, + { + "epoch": 1.3976593308739347, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.999835968017578, + "learning_rate": 1e-06, + "loss": 0.3892, + "mean_token_accuracy": 0.8675618171691895, + "num_tokens": 426728890.0, + "step": 10987 + }, + { + "epoch": 1.397786541152525, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.999506950378418, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8911344408988953, + "num_tokens": 426765771.0, + "step": 10988 + }, + { + "epoch": 1.3979137514311155, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.827020645141602, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8891476988792419, + "num_tokens": 426810179.0, + "step": 10989 + }, + { + "epoch": 1.398040961709706, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.076154708862305, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.883803129196167, + "num_tokens": 426843380.0, + "step": 10990 + }, + { + "epoch": 1.3981681719882966, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.978453636169434, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.9006553888320923, + "num_tokens": 426881175.0, + "step": 10991 + }, + { + "epoch": 1.398295382266887, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.983463287353516, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8802193403244019, + "num_tokens": 426917130.0, + "step": 10992 + }, + { + "epoch": 1.3984225925454776, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.990766525268555, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8853214979171753, + "num_tokens": 426959466.0, + "step": 10993 + }, + { + "epoch": 1.3985498028240682, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.885125160217285, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8951538801193237, + "num_tokens": 427001592.0, + "step": 10994 + }, + { + "epoch": 1.3986770131026587, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.14984130859375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.052507400512695, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8925700187683105, + "num_tokens": 427035705.0, + "step": 10995 + }, + { + "epoch": 1.3988042233812492, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.951838493347168, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.892333447933197, + "num_tokens": 427071250.0, + "step": 10996 + }, + { + "epoch": 1.3989314336598397, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.965314865112305, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8909775018692017, + "num_tokens": 427111482.0, + "step": 10997 + }, + { + "epoch": 1.3990586439384303, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.00122356414795, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8964577913284302, + "num_tokens": 427152193.0, + "step": 10998 + }, + { + "epoch": 1.3991858542170208, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.885045051574707, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8932396173477173, + "num_tokens": 427190804.0, + "step": 10999 + }, + { + "epoch": 1.3993130644956113, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.133401870727539, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8818639516830444, + "num_tokens": 427222566.0, + "step": 11000 + }, + { + "epoch": 1.3994402747742019, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.072511672973633, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8767564296722412, + "num_tokens": 427254321.0, + "step": 11001 + }, + { + "epoch": 1.3995674850527924, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.042840957641602, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8862720131874084, + "num_tokens": 427290413.0, + "step": 11002 + }, + { + "epoch": 1.399694695331383, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.931549072265625, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9014050960540771, + "num_tokens": 427329633.0, + "step": 11003 + }, + { + "epoch": 1.3998219056099732, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.89219856262207, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8986145853996277, + "num_tokens": 427369368.0, + "step": 11004 + }, + { + "epoch": 1.3999491158885637, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.052389144897461, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8851915001869202, + "num_tokens": 427402315.0, + "step": 11005 + }, + { + "epoch": 1.4000763261671543, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.0299654006958, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.8708468675613403, + "num_tokens": 427442988.0, + "step": 11006 + }, + { + "epoch": 1.4002035364457448, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.992711067199707, + "learning_rate": 1e-06, + "loss": 0.3769, + "mean_token_accuracy": 0.8746383190155029, + "num_tokens": 427480866.0, + "step": 11007 + }, + { + "epoch": 1.4003307467243353, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.97207260131836, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8950642943382263, + "num_tokens": 427515384.0, + "step": 11008 + }, + { + "epoch": 1.4004579570029259, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.065298080444336, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8838896751403809, + "num_tokens": 427548678.0, + "step": 11009 + }, + { + "epoch": 1.4005851672815164, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.936590194702148, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8871281147003174, + "num_tokens": 427591370.0, + "step": 11010 + }, + { + "epoch": 1.400712377560107, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.993880271911621, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8960040807723999, + "num_tokens": 427624593.0, + "step": 11011 + }, + { + "epoch": 1.4008395878386974, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.98170280456543, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8888440132141113, + "num_tokens": 427665503.0, + "step": 11012 + }, + { + "epoch": 1.4009667981172877, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.975434303283691, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8948698043823242, + "num_tokens": 427701926.0, + "step": 11013 + }, + { + "epoch": 1.4010940083958783, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.99557113647461, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8938909769058228, + "num_tokens": 427740782.0, + "step": 11014 + }, + { + "epoch": 1.4012212186744688, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.959733963012695, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.891175389289856, + "num_tokens": 427782157.0, + "step": 11015 + }, + { + "epoch": 1.4013484289530593, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.937829971313477, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.9034136533737183, + "num_tokens": 427823169.0, + "step": 11016 + }, + { + "epoch": 1.4014756392316499, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.044990539550781, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8892238140106201, + "num_tokens": 427857224.0, + "step": 11017 + }, + { + "epoch": 1.4016028495102404, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.999125480651855, + "learning_rate": 1e-06, + "loss": 0.3716, + "mean_token_accuracy": 0.8709195256233215, + "num_tokens": 427899801.0, + "step": 11018 + }, + { + "epoch": 1.401730059788831, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.173683166503906e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.990951538085938, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8980154991149902, + "num_tokens": 427935734.0, + "step": 11019 + }, + { + "epoch": 1.4018572700674214, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.102477073669434, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8851009607315063, + "num_tokens": 427975129.0, + "step": 11020 + }, + { + "epoch": 1.401984480346012, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.850187301635742, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.890948474407196, + "num_tokens": 428014882.0, + "step": 11021 + }, + { + "epoch": 1.4021116906246025, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.938850402832031, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8985706567764282, + "num_tokens": 428054091.0, + "step": 11022 + }, + { + "epoch": 1.402238900903193, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.85252571105957, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.9074978828430176, + "num_tokens": 428094104.0, + "step": 11023 + }, + { + "epoch": 1.4023661111817836, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.96333122253418, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8901854157447815, + "num_tokens": 428130341.0, + "step": 11024 + }, + { + "epoch": 1.402493321460374, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.020333290100098, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8931617736816406, + "num_tokens": 428168834.0, + "step": 11025 + }, + { + "epoch": 1.4026205317389646, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.94174575805664, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8820966482162476, + "num_tokens": 428209335.0, + "step": 11026 + }, + { + "epoch": 1.4027477420175551, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.004436492919922, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8913017511367798, + "num_tokens": 428248010.0, + "step": 11027 + }, + { + "epoch": 1.4028749522961454, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.956128120422363, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8910481929779053, + "num_tokens": 428289318.0, + "step": 11028 + }, + { + "epoch": 1.403002162574736, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.123612403869629, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8900547027587891, + "num_tokens": 428319924.0, + "step": 11029 + }, + { + "epoch": 1.4031293728533265, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.964656829833984, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8719584941864014, + "num_tokens": 428360975.0, + "step": 11030 + }, + { + "epoch": 1.403256583131917, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.00010871887207, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.892693281173706, + "num_tokens": 428399689.0, + "step": 11031 + }, + { + "epoch": 1.4033837934105076, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.966837882995605, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.890683650970459, + "num_tokens": 428439005.0, + "step": 11032 + }, + { + "epoch": 1.403511003689098, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.965866088867188, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8920921683311462, + "num_tokens": 428475516.0, + "step": 11033 + }, + { + "epoch": 1.4036382139676886, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.962475776672363, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8908044695854187, + "num_tokens": 428512796.0, + "step": 11034 + }, + { + "epoch": 1.4037654242462791, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.934749603271484, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8968933820724487, + "num_tokens": 428556931.0, + "step": 11035 + }, + { + "epoch": 1.4038926345248697, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.95301628112793, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8920900821685791, + "num_tokens": 428598869.0, + "step": 11036 + }, + { + "epoch": 1.40401984480346, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.18674087524414, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8847208619117737, + "num_tokens": 428627216.0, + "step": 11037 + }, + { + "epoch": 1.4041470550820505, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.891118049621582, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8926858901977539, + "num_tokens": 428670807.0, + "step": 11038 + }, + { + "epoch": 1.404274265360641, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.039974212646484, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8848494291305542, + "num_tokens": 428706400.0, + "step": 11039 + }, + { + "epoch": 1.4044014756392316, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.065842628479004, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8795241117477417, + "num_tokens": 428743703.0, + "step": 11040 + }, + { + "epoch": 1.404528685917822, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.048134803771973, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8942781090736389, + "num_tokens": 428778349.0, + "step": 11041 + }, + { + "epoch": 1.4046558961964126, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.964229583740234, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8813785314559937, + "num_tokens": 428819232.0, + "step": 11042 + }, + { + "epoch": 1.4047831064750032, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.076611518859863, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8827368021011353, + "num_tokens": 428854245.0, + "step": 11043 + }, + { + "epoch": 1.4049103167535937, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.025781631469727, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8873333930969238, + "num_tokens": 428885612.0, + "step": 11044 + }, + { + "epoch": 1.4050375270321842, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.044621467590332, + "learning_rate": 1e-06, + "loss": 0.3887, + "mean_token_accuracy": 0.869755208492279, + "num_tokens": 428923672.0, + "step": 11045 + }, + { + "epoch": 1.4051647373107747, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.94954776763916, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8870591521263123, + "num_tokens": 428961059.0, + "step": 11046 + }, + { + "epoch": 1.4052919475893653, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.061217308044434, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8879226446151733, + "num_tokens": 428996738.0, + "step": 11047 + }, + { + "epoch": 1.4054191578679558, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.914837837219238, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8907129168510437, + "num_tokens": 429034602.0, + "step": 11048 + }, + { + "epoch": 1.4055463681465463, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.940749168395996, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8972718715667725, + "num_tokens": 429074697.0, + "step": 11049 + }, + { + "epoch": 1.4056735784251368, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.962159156799316, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8842470645904541, + "num_tokens": 429120438.0, + "step": 11050 + }, + { + "epoch": 1.4058007887037274, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.932530403137207, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8871349692344666, + "num_tokens": 429165873.0, + "step": 11051 + }, + { + "epoch": 1.405927998982318, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.05490779876709, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8734332323074341, + "num_tokens": 429204608.0, + "step": 11052 + }, + { + "epoch": 1.4060552092609082, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.076401710510254, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8840023279190063, + "num_tokens": 429235143.0, + "step": 11053 + }, + { + "epoch": 1.4061824195394987, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.918928146362305, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.88442462682724, + "num_tokens": 429276568.0, + "step": 11054 + }, + { + "epoch": 1.4063096298180893, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.899995803833008, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8855661153793335, + "num_tokens": 429322915.0, + "step": 11055 + }, + { + "epoch": 1.4064368400966798, + "ewc_loss": 0.01531982421875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.9805326461792, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.9012812972068787, + "num_tokens": 429360853.0, + "step": 11056 + }, + { + "epoch": 1.4065640503752703, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.935184478759766, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.884454607963562, + "num_tokens": 429403949.0, + "step": 11057 + }, + { + "epoch": 1.4066912606538609, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.039753913879395, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8976007699966431, + "num_tokens": 429435739.0, + "step": 11058 + }, + { + "epoch": 1.4068184709324514, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.998174667358398, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.888863205909729, + "num_tokens": 429473668.0, + "step": 11059 + }, + { + "epoch": 1.406945681211042, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.001740455627441, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8832249641418457, + "num_tokens": 429513572.0, + "step": 11060 + }, + { + "epoch": 1.4070728914896324, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.956884384155273, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8749625086784363, + "num_tokens": 429554197.0, + "step": 11061 + }, + { + "epoch": 1.4072001017682227, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.907339096069336, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8827874660491943, + "num_tokens": 429596014.0, + "step": 11062 + }, + { + "epoch": 1.4073273120468133, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.9647216796875, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8895612359046936, + "num_tokens": 429633824.0, + "step": 11063 + }, + { + "epoch": 1.4074545223254038, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.955860137939453, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8892561197280884, + "num_tokens": 429670086.0, + "step": 11064 + }, + { + "epoch": 1.4075817326039943, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.039737701416016, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8886653184890747, + "num_tokens": 429706198.0, + "step": 11065 + }, + { + "epoch": 1.4077089428825849, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.08574104309082, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8924434185028076, + "num_tokens": 429736417.0, + "step": 11066 + }, + { + "epoch": 1.4078361531611754, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.051318168640137, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8728110194206238, + "num_tokens": 429775128.0, + "step": 11067 + }, + { + "epoch": 1.407963363439766, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.002225875854492, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8901025056838989, + "num_tokens": 429810692.0, + "step": 11068 + }, + { + "epoch": 1.4080905737183564, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.004380226135254, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8949772715568542, + "num_tokens": 429842735.0, + "step": 11069 + }, + { + "epoch": 1.408217783996947, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.96109390258789, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8816158175468445, + "num_tokens": 429879340.0, + "step": 11070 + }, + { + "epoch": 1.4083449942755375, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.013440132141113, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8775792717933655, + "num_tokens": 429916989.0, + "step": 11071 + }, + { + "epoch": 1.408472204554128, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.968579292297363, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8945728540420532, + "num_tokens": 429951320.0, + "step": 11072 + }, + { + "epoch": 1.4085994148327186, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.886566162109375, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8948466181755066, + "num_tokens": 429993111.0, + "step": 11073 + }, + { + "epoch": 1.408726625111309, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.97349739074707, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8808535933494568, + "num_tokens": 430038248.0, + "step": 11074 + }, + { + "epoch": 1.4088538353898996, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.947352409362793, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8919292688369751, + "num_tokens": 430076468.0, + "step": 11075 + }, + { + "epoch": 1.4089810456684901, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.980478286743164, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8943166732788086, + "num_tokens": 430113462.0, + "step": 11076 + }, + { + "epoch": 1.4091082559470804, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.061634063720703, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8782132267951965, + "num_tokens": 430148911.0, + "step": 11077 + }, + { + "epoch": 1.409235466225671, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.981549263000488, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8840787410736084, + "num_tokens": 430187287.0, + "step": 11078 + }, + { + "epoch": 1.4093626765042615, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.96366024017334, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8894530534744263, + "num_tokens": 430225622.0, + "step": 11079 + }, + { + "epoch": 1.409489886782852, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.951824188232422, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8847571611404419, + "num_tokens": 430266189.0, + "step": 11080 + }, + { + "epoch": 1.4096170970614426, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.963329315185547, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8954417705535889, + "num_tokens": 430305080.0, + "step": 11081 + }, + { + "epoch": 1.409744307340033, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.922760963439941, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8889055252075195, + "num_tokens": 430349408.0, + "step": 11082 + }, + { + "epoch": 1.4098715176186236, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.93260669708252, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.89345383644104, + "num_tokens": 430389792.0, + "step": 11083 + }, + { + "epoch": 1.4099987278972141, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.009933471679688, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.887915849685669, + "num_tokens": 430426600.0, + "step": 11084 + }, + { + "epoch": 1.4101259381758047, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.055229187011719, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8801438808441162, + "num_tokens": 430461257.0, + "step": 11085 + }, + { + "epoch": 1.410253148454395, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.951359748840332, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8880746364593506, + "num_tokens": 430500094.0, + "step": 11086 + }, + { + "epoch": 1.4103803587329855, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.1975250244140625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.924384117126465, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9073561429977417, + "num_tokens": 430533953.0, + "step": 11087 + }, + { + "epoch": 1.410507569011576, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.076763153076172, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8889528512954712, + "num_tokens": 430566765.0, + "step": 11088 + }, + { + "epoch": 1.4106347792901666, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.92381477355957, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8878133296966553, + "num_tokens": 430611835.0, + "step": 11089 + }, + { + "epoch": 1.410761989568757, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.989399909973145, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8850585222244263, + "num_tokens": 430647417.0, + "step": 11090 + }, + { + "epoch": 1.4108891998473476, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.949262619018555, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8806518316268921, + "num_tokens": 430687246.0, + "step": 11091 + }, + { + "epoch": 1.4110164101259381, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.942036628723145, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.892656683921814, + "num_tokens": 430730965.0, + "step": 11092 + }, + { + "epoch": 1.4111436204045287, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.969256401062012, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8897745013237, + "num_tokens": 430770057.0, + "step": 11093 + }, + { + "epoch": 1.4112708306831192, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.987290382385254, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8892829418182373, + "num_tokens": 430808737.0, + "step": 11094 + }, + { + "epoch": 1.4113980409617097, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.99079418182373, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8951933979988098, + "num_tokens": 430843391.0, + "step": 11095 + }, + { + "epoch": 1.4115252512403003, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.05567455291748, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8764479160308838, + "num_tokens": 430884158.0, + "step": 11096 + }, + { + "epoch": 1.4116524615188908, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.030613899230957, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.9011134505271912, + "num_tokens": 430914253.0, + "step": 11097 + }, + { + "epoch": 1.4117796717974813, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.018264770507812, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8863521814346313, + "num_tokens": 430954978.0, + "step": 11098 + }, + { + "epoch": 1.4119068820760718, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.064303398132324, + "learning_rate": 1e-06, + "loss": 0.3888, + "mean_token_accuracy": 0.8723816871643066, + "num_tokens": 430994925.0, + "step": 11099 + }, + { + "epoch": 1.4120340923546624, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.055468559265137, + "learning_rate": 1e-06, + "loss": 0.3776, + "mean_token_accuracy": 0.8737921714782715, + "num_tokens": 431033796.0, + "step": 11100 + }, + { + "epoch": 1.412161302633253, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.002583503723145, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8881199359893799, + "num_tokens": 431069137.0, + "step": 11101 + }, + { + "epoch": 1.4122885129118432, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.858641624450684, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.887143075466156, + "num_tokens": 431117405.0, + "step": 11102 + }, + { + "epoch": 1.4124157231904337, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.98608684539795, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8832454085350037, + "num_tokens": 431160200.0, + "step": 11103 + }, + { + "epoch": 1.4125429334690243, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.979037284851074, + "learning_rate": 1e-06, + "loss": 0.3724, + "mean_token_accuracy": 0.8752300143241882, + "num_tokens": 431200290.0, + "step": 11104 + }, + { + "epoch": 1.4126701437476148, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.924363136291504, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8924702405929565, + "num_tokens": 431235743.0, + "step": 11105 + }, + { + "epoch": 1.4127973540262053, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.97567367553711, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9002742767333984, + "num_tokens": 431274537.0, + "step": 11106 + }, + { + "epoch": 1.4129245643047958, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.927509307861328, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8991859555244446, + "num_tokens": 431314139.0, + "step": 11107 + }, + { + "epoch": 1.4130517745833864, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.912618637084961, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8985245823860168, + "num_tokens": 431358676.0, + "step": 11108 + }, + { + "epoch": 1.413178984861977, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.995193481445312, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.894271731376648, + "num_tokens": 431394495.0, + "step": 11109 + }, + { + "epoch": 1.4133061951405674, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.028860092163086, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8874167203903198, + "num_tokens": 431430228.0, + "step": 11110 + }, + { + "epoch": 1.4134334054191577, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.956195831298828, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8925098776817322, + "num_tokens": 431472844.0, + "step": 11111 + }, + { + "epoch": 1.4135606156977483, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.022076606750488, + "learning_rate": 1e-06, + "loss": 0.3831, + "mean_token_accuracy": 0.8686902523040771, + "num_tokens": 431510551.0, + "step": 11112 + }, + { + "epoch": 1.4136878259763388, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.010540008544922, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8878008127212524, + "num_tokens": 431553784.0, + "step": 11113 + }, + { + "epoch": 1.4138150362549293, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.927656173706055, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8879228830337524, + "num_tokens": 431593353.0, + "step": 11114 + }, + { + "epoch": 1.4139422465335199, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.934989929199219, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8809424042701721, + "num_tokens": 431640870.0, + "step": 11115 + }, + { + "epoch": 1.4140694568121104, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.952528953552246, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8996332883834839, + "num_tokens": 431679993.0, + "step": 11116 + }, + { + "epoch": 1.414196667090701, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.888858795166016, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8936172127723694, + "num_tokens": 431725691.0, + "step": 11117 + }, + { + "epoch": 1.4143238773692914, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.035442352294922, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8862861394882202, + "num_tokens": 431765637.0, + "step": 11118 + }, + { + "epoch": 1.414451087647882, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.90845775604248, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8837999105453491, + "num_tokens": 431807456.0, + "step": 11119 + }, + { + "epoch": 1.4145782979264725, + "ewc_loss": 0.01544189453125, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.926095008850098, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8962628841400146, + "num_tokens": 431844964.0, + "step": 11120 + }, + { + "epoch": 1.414705508205063, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.966550827026367, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.9025083780288696, + "num_tokens": 431884834.0, + "step": 11121 + }, + { + "epoch": 1.4148327184836536, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.245208740234375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.890471458435059, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.892630398273468, + "num_tokens": 431924702.0, + "step": 11122 + }, + { + "epoch": 1.414959928762244, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.023284912109375, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8897758722305298, + "num_tokens": 431962100.0, + "step": 11123 + }, + { + "epoch": 1.4150871390408346, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.886711120605469, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8987209796905518, + "num_tokens": 432007515.0, + "step": 11124 + }, + { + "epoch": 1.4152143493194251, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0073184967041016e-05, + "grad_norm": 9.042035102844238, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.8802682161331177, + "num_tokens": 432043557.0, + "step": 11125 + }, + { + "epoch": 1.4153415595980154, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.073192596435547, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8680618405342102, + "num_tokens": 432081629.0, + "step": 11126 + }, + { + "epoch": 1.415468769876606, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.91812515258789, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.886871337890625, + "num_tokens": 432125674.0, + "step": 11127 + }, + { + "epoch": 1.4155959801551965, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.974336624145508, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8819644451141357, + "num_tokens": 432163004.0, + "step": 11128 + }, + { + "epoch": 1.415723190433787, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.906929016113281, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8876644372940063, + "num_tokens": 432207611.0, + "step": 11129 + }, + { + "epoch": 1.4158504007123776, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.028854370117188, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8778549432754517, + "num_tokens": 432244071.0, + "step": 11130 + }, + { + "epoch": 1.415977610990968, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.054062843322754, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8902303576469421, + "num_tokens": 432277417.0, + "step": 11131 + }, + { + "epoch": 1.4161048212695586, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.925158500671387, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8874852657318115, + "num_tokens": 432313753.0, + "step": 11132 + }, + { + "epoch": 1.4162320315481491, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.932698249816895, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8963424563407898, + "num_tokens": 432354168.0, + "step": 11133 + }, + { + "epoch": 1.4163592418267397, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.942276954650879, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8903612494468689, + "num_tokens": 432390871.0, + "step": 11134 + }, + { + "epoch": 1.41648645210533, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.977760314941406, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8894011974334717, + "num_tokens": 432431939.0, + "step": 11135 + }, + { + "epoch": 1.4166136623839205, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.96811580657959, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8846489191055298, + "num_tokens": 432470795.0, + "step": 11136 + }, + { + "epoch": 1.416740872662511, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.981589317321777, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8938274383544922, + "num_tokens": 432510386.0, + "step": 11137 + }, + { + "epoch": 1.4168680829411016, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.909017562866211, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.903333306312561, + "num_tokens": 432548027.0, + "step": 11138 + }, + { + "epoch": 1.416995293219692, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.114280700683594, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8793379664421082, + "num_tokens": 432582862.0, + "step": 11139 + }, + { + "epoch": 1.4171225034982826, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.891535758972168, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8987561464309692, + "num_tokens": 432624899.0, + "step": 11140 + }, + { + "epoch": 1.4172497137768731, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.074822425842285, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8897430896759033, + "num_tokens": 432656137.0, + "step": 11141 + }, + { + "epoch": 1.4173769240554637, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.043074607849121, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8946294188499451, + "num_tokens": 432693526.0, + "step": 11142 + }, + { + "epoch": 1.4175041343340542, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.837053298950195, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.9045864939689636, + "num_tokens": 432730919.0, + "step": 11143 + }, + { + "epoch": 1.4176313446126447, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.958150863647461, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8911494612693787, + "num_tokens": 432774079.0, + "step": 11144 + }, + { + "epoch": 1.4177585548912353, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.975250244140625, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.884996235370636, + "num_tokens": 432813448.0, + "step": 11145 + }, + { + "epoch": 1.4178857651698258, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.106293678283691, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8812551498413086, + "num_tokens": 432850365.0, + "step": 11146 + }, + { + "epoch": 1.4180129754484163, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.990639686584473, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8796614408493042, + "num_tokens": 432889790.0, + "step": 11147 + }, + { + "epoch": 1.4181401857270068, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.959409713745117, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8860241770744324, + "num_tokens": 432927350.0, + "step": 11148 + }, + { + "epoch": 1.4182673960055974, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.94329833984375, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8857206702232361, + "num_tokens": 432969204.0, + "step": 11149 + }, + { + "epoch": 1.418394606284188, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.0127534866333, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8862359523773193, + "num_tokens": 433004805.0, + "step": 11150 + }, + { + "epoch": 1.4185218165627782, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.963190078735352, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8872919082641602, + "num_tokens": 433044179.0, + "step": 11151 + }, + { + "epoch": 1.4186490268413687, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.954638481140137, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8876028060913086, + "num_tokens": 433084061.0, + "step": 11152 + }, + { + "epoch": 1.4187762371199593, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.123391151428223, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8932235240936279, + "num_tokens": 433112671.0, + "step": 11153 + }, + { + "epoch": 1.4189034473985498, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.912280082702637, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8949800729751587, + "num_tokens": 433148709.0, + "step": 11154 + }, + { + "epoch": 1.4190306576771403, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.040815353393555, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8780259490013123, + "num_tokens": 433185589.0, + "step": 11155 + }, + { + "epoch": 1.4191578679557308, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.980515480041504, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8790143728256226, + "num_tokens": 433222302.0, + "step": 11156 + }, + { + "epoch": 1.4192850782343214, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.887495040893555, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8966182470321655, + "num_tokens": 433262017.0, + "step": 11157 + }, + { + "epoch": 1.419412288512912, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.96672534942627, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8970938920974731, + "num_tokens": 433299573.0, + "step": 11158 + }, + { + "epoch": 1.4195394987915024, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.166674613952637, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8876444101333618, + "num_tokens": 433341651.0, + "step": 11159 + }, + { + "epoch": 1.4196667090700927, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.007753372192383, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8968817591667175, + "num_tokens": 433381316.0, + "step": 11160 + }, + { + "epoch": 1.4197939193486833, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.032102584838867, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8781028389930725, + "num_tokens": 433419452.0, + "step": 11161 + }, + { + "epoch": 1.4199211296272738, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.973356246948242, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8956810235977173, + "num_tokens": 433460762.0, + "step": 11162 + }, + { + "epoch": 1.4200483399058643, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.972967147827148, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8833351731300354, + "num_tokens": 433500840.0, + "step": 11163 + }, + { + "epoch": 1.4201755501844548, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.045313835144043, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8895300030708313, + "num_tokens": 433539385.0, + "step": 11164 + }, + { + "epoch": 1.4203027604630454, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.941157341003418, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8868493437767029, + "num_tokens": 433579447.0, + "step": 11165 + }, + { + "epoch": 1.420429970741636, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.054359436035156, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8866843581199646, + "num_tokens": 433619888.0, + "step": 11166 + }, + { + "epoch": 1.4205571810202264, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.016688346862793, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8797008991241455, + "num_tokens": 433657881.0, + "step": 11167 + }, + { + "epoch": 1.420684391298817, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.030795097351074, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8811428546905518, + "num_tokens": 433696593.0, + "step": 11168 + }, + { + "epoch": 1.4208116015774075, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.932812690734863, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8805205225944519, + "num_tokens": 433739936.0, + "step": 11169 + }, + { + "epoch": 1.420938811855998, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.96712589263916, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8821874856948853, + "num_tokens": 433787768.0, + "step": 11170 + }, + { + "epoch": 1.4210660221345885, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.026135444641113, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8787022829055786, + "num_tokens": 433827081.0, + "step": 11171 + }, + { + "epoch": 1.421193232413179, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.112723350524902, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.9032003283500671, + "num_tokens": 433862800.0, + "step": 11172 + }, + { + "epoch": 1.4213204426917696, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.998353004455566, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8817979097366333, + "num_tokens": 433901485.0, + "step": 11173 + }, + { + "epoch": 1.4214476529703601, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.882896423339844, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.896040141582489, + "num_tokens": 433946117.0, + "step": 11174 + }, + { + "epoch": 1.4215748632489504, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.955116271972656, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8980836868286133, + "num_tokens": 433984725.0, + "step": 11175 + }, + { + "epoch": 1.421702073527541, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.916152954101562, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8840607404708862, + "num_tokens": 434027000.0, + "step": 11176 + }, + { + "epoch": 1.4218292838061315, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.044560432434082, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8772010803222656, + "num_tokens": 434067575.0, + "step": 11177 + }, + { + "epoch": 1.421956494084722, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.97472858428955, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8936324715614319, + "num_tokens": 434102792.0, + "step": 11178 + }, + { + "epoch": 1.4220837043633126, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.987517356872559, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8888431191444397, + "num_tokens": 434143377.0, + "step": 11179 + }, + { + "epoch": 1.422210914641903, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.000570297241211, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.9017047882080078, + "num_tokens": 434179708.0, + "step": 11180 + }, + { + "epoch": 1.4223381249204936, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.013214111328125, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8956109285354614, + "num_tokens": 434219686.0, + "step": 11181 + }, + { + "epoch": 1.4224653351990841, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.024545669555664, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8759005069732666, + "num_tokens": 434258479.0, + "step": 11182 + }, + { + "epoch": 1.4225925454776747, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.975851058959961, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8782677054405212, + "num_tokens": 434300676.0, + "step": 11183 + }, + { + "epoch": 1.422719755756265, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.010247230529785, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8895890712738037, + "num_tokens": 434342149.0, + "step": 11184 + }, + { + "epoch": 1.4228469660348555, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.018131256103516, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8826532363891602, + "num_tokens": 434383442.0, + "step": 11185 + }, + { + "epoch": 1.422974176313446, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.008490562438965, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8996593356132507, + "num_tokens": 434414584.0, + "step": 11186 + }, + { + "epoch": 1.4231013865920366, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.9273042678833, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.9017944931983948, + "num_tokens": 434455522.0, + "step": 11187 + }, + { + "epoch": 1.423228596870627, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.060532569885254, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8820600509643555, + "num_tokens": 434491123.0, + "step": 11188 + }, + { + "epoch": 1.4233558071492176, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.962553024291992, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8983958959579468, + "num_tokens": 434528451.0, + "step": 11189 + }, + { + "epoch": 1.4234830174278081, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.058229446411133, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8824037313461304, + "num_tokens": 434563186.0, + "step": 11190 + }, + { + "epoch": 1.4236102277063987, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.22362232208252, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8846790790557861, + "num_tokens": 434598191.0, + "step": 11191 + }, + { + "epoch": 1.4237374379849892, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.999256134033203, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8890681862831116, + "num_tokens": 434636782.0, + "step": 11192 + }, + { + "epoch": 1.4238646482635797, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.943161010742188, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8940296173095703, + "num_tokens": 434673797.0, + "step": 11193 + }, + { + "epoch": 1.4239918585421703, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.03183364868164, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8799194097518921, + "num_tokens": 434711417.0, + "step": 11194 + }, + { + "epoch": 1.4241190688207608, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.986603736877441, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8938289284706116, + "num_tokens": 434746997.0, + "step": 11195 + }, + { + "epoch": 1.4242462790993513, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.033721923828125, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.884062647819519, + "num_tokens": 434782536.0, + "step": 11196 + }, + { + "epoch": 1.4243734893779418, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.01423168182373, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8790438175201416, + "num_tokens": 434818526.0, + "step": 11197 + }, + { + "epoch": 1.4245006996565324, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.951445579528809, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8752508759498596, + "num_tokens": 434860162.0, + "step": 11198 + }, + { + "epoch": 1.424627909935123, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.006542205810547, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8814375400543213, + "num_tokens": 434899335.0, + "step": 11199 + }, + { + "epoch": 1.4247551202137132, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.953560829162598, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.9004672765731812, + "num_tokens": 434941214.0, + "step": 11200 + }, + { + "epoch": 1.4248823304923037, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.100798606872559, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8756844997406006, + "num_tokens": 434978418.0, + "step": 11201 + }, + { + "epoch": 1.4250095407708943, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.032533645629883, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8903302550315857, + "num_tokens": 435016409.0, + "step": 11202 + }, + { + "epoch": 1.4251367510494848, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.977389335632324, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8936828970909119, + "num_tokens": 435046795.0, + "step": 11203 + }, + { + "epoch": 1.4252639613280753, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.00192642211914, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8835413455963135, + "num_tokens": 435082303.0, + "step": 11204 + }, + { + "epoch": 1.4253911716066658, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.942298889160156, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8878520727157593, + "num_tokens": 435124734.0, + "step": 11205 + }, + { + "epoch": 1.4255183818852564, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.965404510498047, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.877835750579834, + "num_tokens": 435167610.0, + "step": 11206 + }, + { + "epoch": 1.425645592163847, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.999529838562012, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8955153226852417, + "num_tokens": 435206943.0, + "step": 11207 + }, + { + "epoch": 1.4257728024424374, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.97208309173584, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8885156512260437, + "num_tokens": 435247702.0, + "step": 11208 + }, + { + "epoch": 1.4259000127210277, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.956205368041992, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8893157243728638, + "num_tokens": 435286156.0, + "step": 11209 + }, + { + "epoch": 1.4260272229996183, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.02116870880127, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8900019526481628, + "num_tokens": 435323173.0, + "step": 11210 + }, + { + "epoch": 1.4261544332782088, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.038400650024414, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8956727981567383, + "num_tokens": 435357570.0, + "step": 11211 + }, + { + "epoch": 1.4262816435567993, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.964803695678711, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8812785148620605, + "num_tokens": 435400335.0, + "step": 11212 + }, + { + "epoch": 1.4264088538353898, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.94697093963623, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8987317681312561, + "num_tokens": 435439947.0, + "step": 11213 + }, + { + "epoch": 1.4265360641139804, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.121121406555176, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8868333697319031, + "num_tokens": 435475942.0, + "step": 11214 + }, + { + "epoch": 1.426663274392571, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.926189422607422, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8945097923278809, + "num_tokens": 435515633.0, + "step": 11215 + }, + { + "epoch": 1.4267904846711614, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.92388916015625, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9028469920158386, + "num_tokens": 435556605.0, + "step": 11216 + }, + { + "epoch": 1.426917694949752, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.9862060546875, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.881299614906311, + "num_tokens": 435600710.0, + "step": 11217 + }, + { + "epoch": 1.4270449052283425, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.937612533569336, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8888736963272095, + "num_tokens": 435642276.0, + "step": 11218 + }, + { + "epoch": 1.427172115506933, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.903091430664062, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8999146223068237, + "num_tokens": 435686959.0, + "step": 11219 + }, + { + "epoch": 1.4272993257855235, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.978235244750977, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8780679106712341, + "num_tokens": 435731363.0, + "step": 11220 + }, + { + "epoch": 1.427426536064114, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.01817512512207, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8855952024459839, + "num_tokens": 435768477.0, + "step": 11221 + }, + { + "epoch": 1.4275537463427046, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.029853820800781, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8994345664978027, + "num_tokens": 435804232.0, + "step": 11222 + }, + { + "epoch": 1.4276809566212951, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.06091594696045, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8924160003662109, + "num_tokens": 435839193.0, + "step": 11223 + }, + { + "epoch": 1.4278081668998854, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.884501457214355, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9070783853530884, + "num_tokens": 435879118.0, + "step": 11224 + }, + { + "epoch": 1.427935377178476, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.010598182678223, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.9005343914031982, + "num_tokens": 435914702.0, + "step": 11225 + }, + { + "epoch": 1.4280625874570665, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.023713111877441, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8887605667114258, + "num_tokens": 435952054.0, + "step": 11226 + }, + { + "epoch": 1.428189797735657, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.017047882080078, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8786460161209106, + "num_tokens": 435991386.0, + "step": 11227 + }, + { + "epoch": 1.4283170080142475, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.985953330993652, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.894280731678009, + "num_tokens": 436026454.0, + "step": 11228 + }, + { + "epoch": 1.428444218292838, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.95860481262207, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8881899118423462, + "num_tokens": 436070038.0, + "step": 11229 + }, + { + "epoch": 1.4285714285714286, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.032872200012207, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8859488368034363, + "num_tokens": 436112016.0, + "step": 11230 + }, + { + "epoch": 1.4286986388500191, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.879613876342773, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8846123814582825, + "num_tokens": 436156933.0, + "step": 11231 + }, + { + "epoch": 1.4288258491286097, + "ewc_loss": 0.015380859375, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.092392921447754, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8878260850906372, + "num_tokens": 436191061.0, + "step": 11232 + }, + { + "epoch": 1.4289530594072, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.075383186340332, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.892327606678009, + "num_tokens": 436224822.0, + "step": 11233 + }, + { + "epoch": 1.4290802696857905, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.017684936523438, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8808253407478333, + "num_tokens": 436263267.0, + "step": 11234 + }, + { + "epoch": 1.429207479964381, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.9562349319458, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8956984281539917, + "num_tokens": 436301261.0, + "step": 11235 + }, + { + "epoch": 1.4293346902429716, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.915940284729004, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9037726521492004, + "num_tokens": 436340885.0, + "step": 11236 + }, + { + "epoch": 1.429461900521562, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.019285202026367, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8833381533622742, + "num_tokens": 436378350.0, + "step": 11237 + }, + { + "epoch": 1.4295891108001526, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.989830017089844, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8843609094619751, + "num_tokens": 436417437.0, + "step": 11238 + }, + { + "epoch": 1.4297163210787431, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.100844383239746, + "learning_rate": 1e-06, + "loss": 0.4065, + "mean_token_accuracy": 0.8639655709266663, + "num_tokens": 436452713.0, + "step": 11239 + }, + { + "epoch": 1.4298435313573337, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.048758506774902, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8915264010429382, + "num_tokens": 436489525.0, + "step": 11240 + }, + { + "epoch": 1.4299707416359242, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.945755958557129, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8898520469665527, + "num_tokens": 436530296.0, + "step": 11241 + }, + { + "epoch": 1.4300979519145147, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.028604507446289, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8876117467880249, + "num_tokens": 436571773.0, + "step": 11242 + }, + { + "epoch": 1.4302251621931052, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.070441246032715, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.884803295135498, + "num_tokens": 436610378.0, + "step": 11243 + }, + { + "epoch": 1.4303523724716958, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.955745697021484, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8899860382080078, + "num_tokens": 436651335.0, + "step": 11244 + }, + { + "epoch": 1.4304795827502863, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.1256685256958, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8853527903556824, + "num_tokens": 436686000.0, + "step": 11245 + }, + { + "epoch": 1.4306067930288768, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.977108001708984, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8895140886306763, + "num_tokens": 436728814.0, + "step": 11246 + }, + { + "epoch": 1.4307340033074674, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.03109359741211, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8940542936325073, + "num_tokens": 436770248.0, + "step": 11247 + }, + { + "epoch": 1.430861213586058, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.04186725616455, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8912146091461182, + "num_tokens": 436804816.0, + "step": 11248 + }, + { + "epoch": 1.4309884238646482, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.084598541259766, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8733676671981812, + "num_tokens": 436845132.0, + "step": 11249 + }, + { + "epoch": 1.4311156341432387, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.994924545288086, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.885956883430481, + "num_tokens": 436887341.0, + "step": 11250 + }, + { + "epoch": 1.4312428444218293, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.984994888305664, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8999805450439453, + "num_tokens": 436920832.0, + "step": 11251 + }, + { + "epoch": 1.4313700547004198, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.070219993591309, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8801851868629456, + "num_tokens": 436955982.0, + "step": 11252 + }, + { + "epoch": 1.4314972649790103, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.1447172164917, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.8767993450164795, + "num_tokens": 436992702.0, + "step": 11253 + }, + { + "epoch": 1.4316244752576008, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.060213088989258, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8830645084381104, + "num_tokens": 437031121.0, + "step": 11254 + }, + { + "epoch": 1.4317516855361914, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.996850967407227, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8785617351531982, + "num_tokens": 437072233.0, + "step": 11255 + }, + { + "epoch": 1.431878895814782, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.008257865905762, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8889234066009521, + "num_tokens": 437112151.0, + "step": 11256 + }, + { + "epoch": 1.4320061060933724, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.989017486572266, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8866928219795227, + "num_tokens": 437153357.0, + "step": 11257 + }, + { + "epoch": 1.4321333163719627, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.993191719055176, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8930091857910156, + "num_tokens": 437192379.0, + "step": 11258 + }, + { + "epoch": 1.4322605266505533, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.059660911560059, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8894637823104858, + "num_tokens": 437226635.0, + "step": 11259 + }, + { + "epoch": 1.4323877369291438, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.108558654785156, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8933064341545105, + "num_tokens": 437259460.0, + "step": 11260 + }, + { + "epoch": 1.4325149472077343, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.976123809814453, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8872313499450684, + "num_tokens": 437297594.0, + "step": 11261 + }, + { + "epoch": 1.4326421574863248, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.2928924560546875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.884299278259277, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8993645906448364, + "num_tokens": 437339723.0, + "step": 11262 + }, + { + "epoch": 1.4327693677649154, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.112245559692383, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8851333856582642, + "num_tokens": 437372675.0, + "step": 11263 + }, + { + "epoch": 1.432896578043506, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.972168922424316, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8934594392776489, + "num_tokens": 437413022.0, + "step": 11264 + }, + { + "epoch": 1.4330237883220964, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.00638198852539, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8785614967346191, + "num_tokens": 437452573.0, + "step": 11265 + }, + { + "epoch": 1.433150998600687, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.99244213104248, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8870108127593994, + "num_tokens": 437497935.0, + "step": 11266 + }, + { + "epoch": 1.4332782088792775, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.081863403320312, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8824224472045898, + "num_tokens": 437533764.0, + "step": 11267 + }, + { + "epoch": 1.433405419157868, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.984357833862305, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8773376941680908, + "num_tokens": 437581432.0, + "step": 11268 + }, + { + "epoch": 1.4335326294364585, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.244965553283691, + "learning_rate": 1e-06, + "loss": 0.4138, + "mean_token_accuracy": 0.8657200932502747, + "num_tokens": 437614261.0, + "step": 11269 + }, + { + "epoch": 1.433659839715049, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.011831283569336, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.893304705619812, + "num_tokens": 437647513.0, + "step": 11270 + }, + { + "epoch": 1.4337870499936396, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.021035194396973, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8871921300888062, + "num_tokens": 437687208.0, + "step": 11271 + }, + { + "epoch": 1.4339142602722301, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.02979564666748, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.8774317502975464, + "num_tokens": 437727572.0, + "step": 11272 + }, + { + "epoch": 1.4340414705508204, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.031294822692871, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8956751823425293, + "num_tokens": 437767481.0, + "step": 11273 + }, + { + "epoch": 1.434168680829411, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.05240249633789, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8805519938468933, + "num_tokens": 437806485.0, + "step": 11274 + }, + { + "epoch": 1.4342958911080015, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.975854873657227, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8730974197387695, + "num_tokens": 437848409.0, + "step": 11275 + }, + { + "epoch": 1.434423101386592, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.074790954589844, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.883455753326416, + "num_tokens": 437881149.0, + "step": 11276 + }, + { + "epoch": 1.4345503116651825, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.009125709533691, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8931883573532104, + "num_tokens": 437916798.0, + "step": 11277 + }, + { + "epoch": 1.434677521943773, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.88448429107666, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9083099961280823, + "num_tokens": 437956395.0, + "step": 11278 + }, + { + "epoch": 1.4348047322223636, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.227513313293457, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8810604810714722, + "num_tokens": 437987338.0, + "step": 11279 + }, + { + "epoch": 1.4349319425009541, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 8.98502254486084, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8991469144821167, + "num_tokens": 438031610.0, + "step": 11280 + }, + { + "epoch": 1.4350591527795447, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.10605525970459, + "learning_rate": 1e-06, + "loss": 0.3825, + "mean_token_accuracy": 0.867815375328064, + "num_tokens": 438075284.0, + "step": 11281 + }, + { + "epoch": 1.435186363058135, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.11306095123291, + "learning_rate": 1e-06, + "loss": 0.3724, + "mean_token_accuracy": 0.8732357621192932, + "num_tokens": 438113432.0, + "step": 11282 + }, + { + "epoch": 1.4353135733367255, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.03750228881836, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8799014091491699, + "num_tokens": 438152020.0, + "step": 11283 + }, + { + "epoch": 1.435440783615316, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.012661933898926, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8912889957427979, + "num_tokens": 438190498.0, + "step": 11284 + }, + { + "epoch": 1.4355679938939065, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.97226619720459, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9019788503646851, + "num_tokens": 438225086.0, + "step": 11285 + }, + { + "epoch": 1.435695204172497, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.141904830932617, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.881878137588501, + "num_tokens": 438257832.0, + "step": 11286 + }, + { + "epoch": 1.4358224144510876, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.134160995483398, + "learning_rate": 1e-06, + "loss": 0.4139, + "mean_token_accuracy": 0.8610885143280029, + "num_tokens": 438296622.0, + "step": 11287 + }, + { + "epoch": 1.4359496247296781, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.022542953491211, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8832291960716248, + "num_tokens": 438333115.0, + "step": 11288 + }, + { + "epoch": 1.4360768350082687, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.94846248626709, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8853164315223694, + "num_tokens": 438372326.0, + "step": 11289 + }, + { + "epoch": 1.4362040452868592, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.08732795715332, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8863054513931274, + "num_tokens": 438409417.0, + "step": 11290 + }, + { + "epoch": 1.4363312555654497, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.05131721496582, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8784565925598145, + "num_tokens": 438443459.0, + "step": 11291 + }, + { + "epoch": 1.4364584658440402, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.012993812561035, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8940067887306213, + "num_tokens": 438479459.0, + "step": 11292 + }, + { + "epoch": 1.4365856761226308, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.946645736694336, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.899027943611145, + "num_tokens": 438515827.0, + "step": 11293 + }, + { + "epoch": 1.4367128864012213, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.04636001586914, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8758186101913452, + "num_tokens": 438553218.0, + "step": 11294 + }, + { + "epoch": 1.4368400966798118, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.949689865112305, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8958677649497986, + "num_tokens": 438590871.0, + "step": 11295 + }, + { + "epoch": 1.4369673069584024, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.97474479675293, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8747971057891846, + "num_tokens": 438632560.0, + "step": 11296 + }, + { + "epoch": 1.4370945172369929, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.927310943603516, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.903480052947998, + "num_tokens": 438669227.0, + "step": 11297 + }, + { + "epoch": 1.4372217275155832, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.045575141906738, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8806610107421875, + "num_tokens": 438707987.0, + "step": 11298 + }, + { + "epoch": 1.4373489377941737, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.874321937561035, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8872668743133545, + "num_tokens": 438750636.0, + "step": 11299 + }, + { + "epoch": 1.4374761480727642, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.979174613952637, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8857860565185547, + "num_tokens": 438793917.0, + "step": 11300 + }, + { + "epoch": 1.4376033583513548, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.945815086364746, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8941035270690918, + "num_tokens": 438839583.0, + "step": 11301 + }, + { + "epoch": 1.4377305686299453, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.007933616638184, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8903288841247559, + "num_tokens": 438881268.0, + "step": 11302 + }, + { + "epoch": 1.4378577789085358, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.0431547164917, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.885593056678772, + "num_tokens": 438922280.0, + "step": 11303 + }, + { + "epoch": 1.4379849891871264, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.984485626220703, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8859373927116394, + "num_tokens": 438962870.0, + "step": 11304 + }, + { + "epoch": 1.438112199465717, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.196980476379395, + "learning_rate": 1e-06, + "loss": 0.4047, + "mean_token_accuracy": 0.865199089050293, + "num_tokens": 439001082.0, + "step": 11305 + }, + { + "epoch": 1.4382394097443074, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.991486549377441, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8843313455581665, + "num_tokens": 439041591.0, + "step": 11306 + }, + { + "epoch": 1.4383666200228977, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.020676612854004, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8854131698608398, + "num_tokens": 439078995.0, + "step": 11307 + }, + { + "epoch": 1.4384938303014883, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.935700416564941, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9105690121650696, + "num_tokens": 439117644.0, + "step": 11308 + }, + { + "epoch": 1.4386210405800788, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.966409683227539, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8945412635803223, + "num_tokens": 439160245.0, + "step": 11309 + }, + { + "epoch": 1.4387482508586693, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.023059844970703, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8923301696777344, + "num_tokens": 439198191.0, + "step": 11310 + }, + { + "epoch": 1.4388754611372598, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.968599319458008, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8833385109901428, + "num_tokens": 439236288.0, + "step": 11311 + }, + { + "epoch": 1.4390026714158504, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.005338668823242, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8931880593299866, + "num_tokens": 439275313.0, + "step": 11312 + }, + { + "epoch": 1.439129881694441, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.973634719848633, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8988150358200073, + "num_tokens": 439315235.0, + "step": 11313 + }, + { + "epoch": 1.4392570919730314, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.988288879394531, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8942850828170776, + "num_tokens": 439351369.0, + "step": 11314 + }, + { + "epoch": 1.439384302251622, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.037302017211914, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8808091878890991, + "num_tokens": 439390559.0, + "step": 11315 + }, + { + "epoch": 1.4395115125302125, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.991249084472656, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8947673439979553, + "num_tokens": 439425435.0, + "step": 11316 + }, + { + "epoch": 1.439638722808803, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.107840538024902, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8881510496139526, + "num_tokens": 439460739.0, + "step": 11317 + }, + { + "epoch": 1.4397659330873935, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.071488380432129, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8736698627471924, + "num_tokens": 439496286.0, + "step": 11318 + }, + { + "epoch": 1.439893143365984, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.06505298614502, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8844344019889832, + "num_tokens": 439530501.0, + "step": 11319 + }, + { + "epoch": 1.4400203536445746, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.978595733642578, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8932211399078369, + "num_tokens": 439564374.0, + "step": 11320 + }, + { + "epoch": 1.4401475639231651, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.942477226257324, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8973957300186157, + "num_tokens": 439604770.0, + "step": 11321 + }, + { + "epoch": 1.4402747742017554, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.009836196899414, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8905137181282043, + "num_tokens": 439640312.0, + "step": 11322 + }, + { + "epoch": 1.440401984480346, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.01310920715332, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8849847912788391, + "num_tokens": 439682903.0, + "step": 11323 + }, + { + "epoch": 1.4405291947589365, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.040054321289062, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8780248165130615, + "num_tokens": 439720506.0, + "step": 11324 + }, + { + "epoch": 1.440656405037527, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.021575927734375, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8922088146209717, + "num_tokens": 439756291.0, + "step": 11325 + }, + { + "epoch": 1.4407836153161175, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.078009605407715, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8937462568283081, + "num_tokens": 439787382.0, + "step": 11326 + }, + { + "epoch": 1.440910825594708, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.965879440307617, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8877903819084167, + "num_tokens": 439827862.0, + "step": 11327 + }, + { + "epoch": 1.4410380358732986, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.01966667175293, + "learning_rate": 1e-06, + "loss": 0.3716, + "mean_token_accuracy": 0.8721221685409546, + "num_tokens": 439863400.0, + "step": 11328 + }, + { + "epoch": 1.4411652461518891, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.316734313964844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.964889526367188, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8857181072235107, + "num_tokens": 439907141.0, + "step": 11329 + }, + { + "epoch": 1.4412924564304797, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.95311164855957, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8952662944793701, + "num_tokens": 439947911.0, + "step": 11330 + }, + { + "epoch": 1.44141966670907, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.000223159790039, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.89229816198349, + "num_tokens": 439988256.0, + "step": 11331 + }, + { + "epoch": 1.4415468769876605, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.94786262512207, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8810889720916748, + "num_tokens": 440029922.0, + "step": 11332 + }, + { + "epoch": 1.441674087266251, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.071094512939453, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8828134536743164, + "num_tokens": 440066732.0, + "step": 11333 + }, + { + "epoch": 1.4418012975448415, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.098811149597168, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8831759691238403, + "num_tokens": 440098147.0, + "step": 11334 + }, + { + "epoch": 1.441928507823432, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.94648265838623, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8972882032394409, + "num_tokens": 440138438.0, + "step": 11335 + }, + { + "epoch": 1.4420557181020226, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.01215934753418, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8786047697067261, + "num_tokens": 440182212.0, + "step": 11336 + }, + { + "epoch": 1.4421829283806131, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.019577980041504, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8805646896362305, + "num_tokens": 440220624.0, + "step": 11337 + }, + { + "epoch": 1.4423101386592037, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.070676803588867, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8899359703063965, + "num_tokens": 440256088.0, + "step": 11338 + }, + { + "epoch": 1.4424373489377942, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.076850891113281, + "learning_rate": 1e-06, + "loss": 0.401, + "mean_token_accuracy": 0.8637750148773193, + "num_tokens": 440292555.0, + "step": 11339 + }, + { + "epoch": 1.4425645592163847, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.027973175048828, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8881480693817139, + "num_tokens": 440328898.0, + "step": 11340 + }, + { + "epoch": 1.4426917694949752, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.993654251098633, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8917316794395447, + "num_tokens": 440369424.0, + "step": 11341 + }, + { + "epoch": 1.4428189797735658, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.988255500793457, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8869141936302185, + "num_tokens": 440413944.0, + "step": 11342 + }, + { + "epoch": 1.4429461900521563, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.056553840637207, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8860951662063599, + "num_tokens": 440448702.0, + "step": 11343 + }, + { + "epoch": 1.4430734003307468, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.071599006652832, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8883290886878967, + "num_tokens": 440482558.0, + "step": 11344 + }, + { + "epoch": 1.4432006106093374, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.017473220825195, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8892567157745361, + "num_tokens": 440519693.0, + "step": 11345 + }, + { + "epoch": 1.4433278208879279, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.001227378845215, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8854727149009705, + "num_tokens": 440560372.0, + "step": 11346 + }, + { + "epoch": 1.4434550311665182, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.090679168701172, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8813415765762329, + "num_tokens": 440591857.0, + "step": 11347 + }, + { + "epoch": 1.4435822414451087, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.00981330871582, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8836648464202881, + "num_tokens": 440633560.0, + "step": 11348 + }, + { + "epoch": 1.4437094517236992, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.026090621948242, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8874531984329224, + "num_tokens": 440670606.0, + "step": 11349 + }, + { + "epoch": 1.4438366620022898, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.965002059936523, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8843851685523987, + "num_tokens": 440718952.0, + "step": 11350 + }, + { + "epoch": 1.4439638722808803, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.084344863891602, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8927457332611084, + "num_tokens": 440750300.0, + "step": 11351 + }, + { + "epoch": 1.4440910825594708, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.038884162902832, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.879059910774231, + "num_tokens": 440787661.0, + "step": 11352 + }, + { + "epoch": 1.4442182928380614, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.978554725646973, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8901174068450928, + "num_tokens": 440826227.0, + "step": 11353 + }, + { + "epoch": 1.4443455031166519, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.164674758911133, + "learning_rate": 1e-06, + "loss": 0.3671, + "mean_token_accuracy": 0.8773103952407837, + "num_tokens": 440861674.0, + "step": 11354 + }, + { + "epoch": 1.4444727133952424, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.142141342163086, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8810921311378479, + "num_tokens": 440895407.0, + "step": 11355 + }, + { + "epoch": 1.4445999236738327, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.995577812194824, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8897536993026733, + "num_tokens": 440936452.0, + "step": 11356 + }, + { + "epoch": 1.4447271339524232, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.056126594543457, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8876845836639404, + "num_tokens": 440971951.0, + "step": 11357 + }, + { + "epoch": 1.4448543442310138, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.006485939025879, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8964433670043945, + "num_tokens": 441006341.0, + "step": 11358 + }, + { + "epoch": 1.4449815545096043, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.035161972045898, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.9038162231445312, + "num_tokens": 441036182.0, + "step": 11359 + }, + { + "epoch": 1.4451087647881948, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 8.969192504882812, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8922380208969116, + "num_tokens": 441077671.0, + "step": 11360 + }, + { + "epoch": 1.4452359750667854, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.088645935058594, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8778970837593079, + "num_tokens": 441114064.0, + "step": 11361 + }, + { + "epoch": 1.445363185345376, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.046481132507324, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8989231586456299, + "num_tokens": 441148596.0, + "step": 11362 + }, + { + "epoch": 1.4454903956239664, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.96150016784668, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8960433602333069, + "num_tokens": 441188746.0, + "step": 11363 + }, + { + "epoch": 1.445617605902557, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.04738712310791, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8944845199584961, + "num_tokens": 441226991.0, + "step": 11364 + }, + { + "epoch": 1.4457448161811475, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.030324935913086, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8758143186569214, + "num_tokens": 441270019.0, + "step": 11365 + }, + { + "epoch": 1.445872026459738, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.048273086547852, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8866571187973022, + "num_tokens": 441306084.0, + "step": 11366 + }, + { + "epoch": 1.4459992367383285, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.072872161865234, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.882483720779419, + "num_tokens": 441345942.0, + "step": 11367 + }, + { + "epoch": 1.446126447016919, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.99350643157959, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8857707977294922, + "num_tokens": 441384925.0, + "step": 11368 + }, + { + "epoch": 1.4462536572955096, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.055807113647461, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8818561434745789, + "num_tokens": 441424672.0, + "step": 11369 + }, + { + "epoch": 1.4463808675741001, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.592473983764648, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8917155265808105, + "num_tokens": 441463337.0, + "step": 11370 + }, + { + "epoch": 1.4465080778526904, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.013166427612305, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8899596929550171, + "num_tokens": 441502966.0, + "step": 11371 + }, + { + "epoch": 1.446635288131281, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.984983444213867, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8798928260803223, + "num_tokens": 441542364.0, + "step": 11372 + }, + { + "epoch": 1.4467624984098715, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.117572784423828, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8881353735923767, + "num_tokens": 441579113.0, + "step": 11373 + }, + { + "epoch": 1.446889708688462, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.906890869140625, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9031506776809692, + "num_tokens": 441619640.0, + "step": 11374 + }, + { + "epoch": 1.4470169189670525, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.030038833618164, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8881516456604004, + "num_tokens": 441655805.0, + "step": 11375 + }, + { + "epoch": 1.447144129245643, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.078132629394531, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8920240998268127, + "num_tokens": 441693521.0, + "step": 11376 + }, + { + "epoch": 1.4472713395242336, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.019545555114746, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.876312255859375, + "num_tokens": 441729787.0, + "step": 11377 + }, + { + "epoch": 1.4473985498028241, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.06998062133789, + "learning_rate": 1e-06, + "loss": 0.3936, + "mean_token_accuracy": 0.8658688068389893, + "num_tokens": 441770262.0, + "step": 11378 + }, + { + "epoch": 1.4475257600814146, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.982124328613281, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8957563042640686, + "num_tokens": 441808197.0, + "step": 11379 + }, + { + "epoch": 1.447652970360005, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.994989395141602, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.894427478313446, + "num_tokens": 441845464.0, + "step": 11380 + }, + { + "epoch": 1.4477801806385955, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.010641098022461, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8877767324447632, + "num_tokens": 441881645.0, + "step": 11381 + }, + { + "epoch": 1.447907390917186, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.065093994140625, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8809394240379333, + "num_tokens": 441923713.0, + "step": 11382 + }, + { + "epoch": 1.4480346011957765, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.012574195861816, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8878589868545532, + "num_tokens": 441959988.0, + "step": 11383 + }, + { + "epoch": 1.448161811474367, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.984886169433594, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8816975355148315, + "num_tokens": 442000796.0, + "step": 11384 + }, + { + "epoch": 1.4482890217529576, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.996113777160645, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.887466311454773, + "num_tokens": 442040507.0, + "step": 11385 + }, + { + "epoch": 1.4484162320315481, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.01280403137207, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8977784514427185, + "num_tokens": 442076541.0, + "step": 11386 + }, + { + "epoch": 1.4485434423101387, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.973891258239746, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8774170875549316, + "num_tokens": 442122888.0, + "step": 11387 + }, + { + "epoch": 1.4486706525887292, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.988844871520996, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9024626016616821, + "num_tokens": 442162899.0, + "step": 11388 + }, + { + "epoch": 1.4487978628673197, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.066137313842773, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8875340223312378, + "num_tokens": 442200350.0, + "step": 11389 + }, + { + "epoch": 1.4489250731459102, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.03497314453125, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8710342645645142, + "num_tokens": 442239699.0, + "step": 11390 + }, + { + "epoch": 1.4490522834245008, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.039761543273926, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8843085169792175, + "num_tokens": 442278363.0, + "step": 11391 + }, + { + "epoch": 1.4491794937030913, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.958121299743652, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8958536386489868, + "num_tokens": 442319584.0, + "step": 11392 + }, + { + "epoch": 1.4493067039816818, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.992021560668945, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8970353007316589, + "num_tokens": 442358584.0, + "step": 11393 + }, + { + "epoch": 1.4494339142602723, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.9169282913208, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8995305895805359, + "num_tokens": 442403722.0, + "step": 11394 + }, + { + "epoch": 1.4495611245388629, + "ewc_loss": 0.0155029296875, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.936873435974121, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8938522338867188, + "num_tokens": 442446227.0, + "step": 11395 + }, + { + "epoch": 1.4496883348174532, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.987058639526367, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.901968240737915, + "num_tokens": 442486869.0, + "step": 11396 + }, + { + "epoch": 1.4498155450960437, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.155022621154785, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8986708521842957, + "num_tokens": 442518015.0, + "step": 11397 + }, + { + "epoch": 1.4499427553746342, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.004867553710938, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8833013772964478, + "num_tokens": 442557447.0, + "step": 11398 + }, + { + "epoch": 1.4500699656532248, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.998209953308105, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.895937979221344, + "num_tokens": 442596641.0, + "step": 11399 + }, + { + "epoch": 1.4501971759318153, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.096722602844238, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.881676197052002, + "num_tokens": 442633432.0, + "step": 11400 + }, + { + "epoch": 1.4503243862104058, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.977916717529297, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9063223600387573, + "num_tokens": 442664525.0, + "step": 11401 + }, + { + "epoch": 1.4504515964889964, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.047219276428223, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8834657669067383, + "num_tokens": 442705314.0, + "step": 11402 + }, + { + "epoch": 1.4505788067675869, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.340576171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.006478309631348, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8858588337898254, + "num_tokens": 442745555.0, + "step": 11403 + }, + { + "epoch": 1.4507060170461774, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.083720207214355, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8713629245758057, + "num_tokens": 442785847.0, + "step": 11404 + }, + { + "epoch": 1.4508332273247677, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.06860065460205, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8846688270568848, + "num_tokens": 442823110.0, + "step": 11405 + }, + { + "epoch": 1.4509604376033582, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.019949913024902, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8986514806747437, + "num_tokens": 442857530.0, + "step": 11406 + }, + { + "epoch": 1.4510876478819488, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.00434398651123, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8980348110198975, + "num_tokens": 442889658.0, + "step": 11407 + }, + { + "epoch": 1.4512148581605393, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.03467845916748, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9037432670593262, + "num_tokens": 442922863.0, + "step": 11408 + }, + { + "epoch": 1.4513420684391298, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.952858924865723, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8963502645492554, + "num_tokens": 442963948.0, + "step": 11409 + }, + { + "epoch": 1.4514692787177204, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.894247055053711, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.89543217420578, + "num_tokens": 443004978.0, + "step": 11410 + }, + { + "epoch": 1.4515964889963109, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.100436210632324, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8847495913505554, + "num_tokens": 443042114.0, + "step": 11411 + }, + { + "epoch": 1.4517236992749014, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.03387451171875, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8838591575622559, + "num_tokens": 443082368.0, + "step": 11412 + }, + { + "epoch": 1.451850909553492, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.083361625671387, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8856226205825806, + "num_tokens": 443115422.0, + "step": 11413 + }, + { + "epoch": 1.4519781198320825, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.005050659179688, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8859571218490601, + "num_tokens": 443158338.0, + "step": 11414 + }, + { + "epoch": 1.452105330110673, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.05224609375, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8818190097808838, + "num_tokens": 443195990.0, + "step": 11415 + }, + { + "epoch": 1.4522325403892635, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.121978759765625, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8842446804046631, + "num_tokens": 443225765.0, + "step": 11416 + }, + { + "epoch": 1.452359750667854, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.01263427734375, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.8718003034591675, + "num_tokens": 443265875.0, + "step": 11417 + }, + { + "epoch": 1.4524869609464446, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.017831802368164, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8895221948623657, + "num_tokens": 443303725.0, + "step": 11418 + }, + { + "epoch": 1.4526141712250351, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.985997200012207, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8887420296669006, + "num_tokens": 443346371.0, + "step": 11419 + }, + { + "epoch": 1.4527413815036254, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.974226951599121, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8874899744987488, + "num_tokens": 443384143.0, + "step": 11420 + }, + { + "epoch": 1.452868591782216, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.169260025024414, + "learning_rate": 1e-06, + "loss": 0.3889, + "mean_token_accuracy": 0.8732729554176331, + "num_tokens": 443420129.0, + "step": 11421 + }, + { + "epoch": 1.4529958020608065, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.906577110290527, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8951610922813416, + "num_tokens": 443463570.0, + "step": 11422 + }, + { + "epoch": 1.453123012339397, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.042479515075684, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8860164284706116, + "num_tokens": 443503173.0, + "step": 11423 + }, + { + "epoch": 1.4532502226179875, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.019936561584473, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.8817332983016968, + "num_tokens": 443542901.0, + "step": 11424 + }, + { + "epoch": 1.453377432896578, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.010054588317871, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8955000042915344, + "num_tokens": 443582284.0, + "step": 11425 + }, + { + "epoch": 1.4535046431751686, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.00894546508789, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8924435973167419, + "num_tokens": 443618270.0, + "step": 11426 + }, + { + "epoch": 1.4536318534537591, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.067058563232422, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8814688920974731, + "num_tokens": 443659313.0, + "step": 11427 + }, + { + "epoch": 1.4537590637323496, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.091692924499512, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8979936242103577, + "num_tokens": 443692474.0, + "step": 11428 + }, + { + "epoch": 1.45388627401094, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.958757400512695, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8842560648918152, + "num_tokens": 443733914.0, + "step": 11429 + }, + { + "epoch": 1.4540134842895305, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.113494873046875, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8737794756889343, + "num_tokens": 443773715.0, + "step": 11430 + }, + { + "epoch": 1.454140694568121, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.090697288513184, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.8747099041938782, + "num_tokens": 443809653.0, + "step": 11431 + }, + { + "epoch": 1.4542679048467115, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.971146583557129, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.8979647159576416, + "num_tokens": 443847515.0, + "step": 11432 + }, + { + "epoch": 1.454395115125302, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.017439842224121, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8883251547813416, + "num_tokens": 443889037.0, + "step": 11433 + }, + { + "epoch": 1.4545223254038926, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.00023365020752, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8837140798568726, + "num_tokens": 443927021.0, + "step": 11434 + }, + { + "epoch": 1.4546495356824831, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.982813835144043, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8860424160957336, + "num_tokens": 443972706.0, + "step": 11435 + }, + { + "epoch": 1.4547767459610736, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.04778003692627, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8953532576560974, + "num_tokens": 444008526.0, + "step": 11436 + }, + { + "epoch": 1.4549039562396642, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.004764556884766, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8877496719360352, + "num_tokens": 444050062.0, + "step": 11437 + }, + { + "epoch": 1.4550311665182547, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.9603853225708, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8986589908599854, + "num_tokens": 444094880.0, + "step": 11438 + }, + { + "epoch": 1.4551583767968452, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.033815383911133, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8835458755493164, + "num_tokens": 444135698.0, + "step": 11439 + }, + { + "epoch": 1.4552855870754358, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.024917602539062, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8959159851074219, + "num_tokens": 444172584.0, + "step": 11440 + }, + { + "epoch": 1.4554127973540263, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.951610565185547, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8888803124427795, + "num_tokens": 444220492.0, + "step": 11441 + }, + { + "epoch": 1.4555400076326168, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.136818885803223, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8888641595840454, + "num_tokens": 444256161.0, + "step": 11442 + }, + { + "epoch": 1.4556672179112073, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.074528694152832, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8797168731689453, + "num_tokens": 444296291.0, + "step": 11443 + }, + { + "epoch": 1.4557944281897979, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.13154411315918, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8771026730537415, + "num_tokens": 444332233.0, + "step": 11444 + }, + { + "epoch": 1.4559216384683882, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.10600471496582, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8977656960487366, + "num_tokens": 444364074.0, + "step": 11445 + }, + { + "epoch": 1.4560488487469787, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.972856521606445, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8963068723678589, + "num_tokens": 444409698.0, + "step": 11446 + }, + { + "epoch": 1.4561760590255692, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 9.040173530578613, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8860482573509216, + "num_tokens": 444452940.0, + "step": 11447 + }, + { + "epoch": 1.4563032693041598, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.029165267944336, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8828654289245605, + "num_tokens": 444490052.0, + "step": 11448 + }, + { + "epoch": 1.4564304795827503, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.004022598266602, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.896848201751709, + "num_tokens": 444527404.0, + "step": 11449 + }, + { + "epoch": 1.4565576898613408, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.936298370361328, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8960751295089722, + "num_tokens": 444569207.0, + "step": 11450 + }, + { + "epoch": 1.4566849001399313, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.065396308898926, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8794022798538208, + "num_tokens": 444602344.0, + "step": 11451 + }, + { + "epoch": 1.4568121104185219, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.052337646484375, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8821154236793518, + "num_tokens": 444641412.0, + "step": 11452 + }, + { + "epoch": 1.4569393206971124, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.177421569824219, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8724312782287598, + "num_tokens": 444676570.0, + "step": 11453 + }, + { + "epoch": 1.4570665309757027, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.977534294128418, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8966252207756042, + "num_tokens": 444715477.0, + "step": 11454 + }, + { + "epoch": 1.4571937412542932, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.061332702636719, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8796118497848511, + "num_tokens": 444753058.0, + "step": 11455 + }, + { + "epoch": 1.4573209515328838, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.026342391967773, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.9018991589546204, + "num_tokens": 444791492.0, + "step": 11456 + }, + { + "epoch": 1.4574481618114743, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.106127738952637, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8933135271072388, + "num_tokens": 444820345.0, + "step": 11457 + }, + { + "epoch": 1.4575753720900648, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.075348854064941, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8842315673828125, + "num_tokens": 444859689.0, + "step": 11458 + }, + { + "epoch": 1.4577025823686554, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.924169540405273, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8879233598709106, + "num_tokens": 444898224.0, + "step": 11459 + }, + { + "epoch": 1.4578297926472459, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.032742500305176, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8975290060043335, + "num_tokens": 444937670.0, + "step": 11460 + }, + { + "epoch": 1.4579570029258364, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.953669548034668, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8819690942764282, + "num_tokens": 444976572.0, + "step": 11461 + }, + { + "epoch": 1.458084213204427, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.921688079833984, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8998109102249146, + "num_tokens": 445016961.0, + "step": 11462 + }, + { + "epoch": 1.4582114234830175, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.949939727783203, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.893852710723877, + "num_tokens": 445056038.0, + "step": 11463 + }, + { + "epoch": 1.458338633761608, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.07162857055664, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8791805505752563, + "num_tokens": 445094727.0, + "step": 11464 + }, + { + "epoch": 1.4584658440401985, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.014589309692383, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.875356912612915, + "num_tokens": 445135762.0, + "step": 11465 + }, + { + "epoch": 1.458593054318789, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.004327774047852, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8881024718284607, + "num_tokens": 445179567.0, + "step": 11466 + }, + { + "epoch": 1.4587202645973796, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.962177276611328, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8796201348304749, + "num_tokens": 445220124.0, + "step": 11467 + }, + { + "epoch": 1.45884747487597, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.985892295837402, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8820346593856812, + "num_tokens": 445260161.0, + "step": 11468 + }, + { + "epoch": 1.4589746851545604, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.021634101867676, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.9040673971176147, + "num_tokens": 445298425.0, + "step": 11469 + }, + { + "epoch": 1.459101895433151, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.992058753967285, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8781304359436035, + "num_tokens": 445337134.0, + "step": 11470 + }, + { + "epoch": 1.4592291057117415, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.952959060668945, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8941689729690552, + "num_tokens": 445379553.0, + "step": 11471 + }, + { + "epoch": 1.459356315990332, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.004217147827148, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8903679847717285, + "num_tokens": 445421425.0, + "step": 11472 + }, + { + "epoch": 1.4594835262689225, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.97976016998291, + "learning_rate": 1e-06, + "loss": 0.3825, + "mean_token_accuracy": 0.8722093105316162, + "num_tokens": 445468150.0, + "step": 11473 + }, + { + "epoch": 1.459610736547513, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.072280883789062, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8864939212799072, + "num_tokens": 445503158.0, + "step": 11474 + }, + { + "epoch": 1.4597379468261036, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.081846237182617, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8828680515289307, + "num_tokens": 445541140.0, + "step": 11475 + }, + { + "epoch": 1.4598651571046941, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.00583267211914, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8823474645614624, + "num_tokens": 445580182.0, + "step": 11476 + }, + { + "epoch": 1.4599923673832846, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.047632217407227, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8967245817184448, + "num_tokens": 445615316.0, + "step": 11477 + }, + { + "epoch": 1.460119577661875, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.958061218261719, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8933684825897217, + "num_tokens": 445659279.0, + "step": 11478 + }, + { + "epoch": 1.4602467879404655, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.073775291442871, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8926515579223633, + "num_tokens": 445696068.0, + "step": 11479 + }, + { + "epoch": 1.460373998219056, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.984441757202148, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8941900134086609, + "num_tokens": 445740958.0, + "step": 11480 + }, + { + "epoch": 1.4605012084976465, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.032230377197266, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8896982669830322, + "num_tokens": 445776117.0, + "step": 11481 + }, + { + "epoch": 1.460628418776237, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.143889427185059, + "learning_rate": 1e-06, + "loss": 0.384, + "mean_token_accuracy": 0.8715344071388245, + "num_tokens": 445815860.0, + "step": 11482 + }, + { + "epoch": 1.4607556290548276, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.976974487304688, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.9042650461196899, + "num_tokens": 445853992.0, + "step": 11483 + }, + { + "epoch": 1.4608828393334181, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.954546928405762, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8932199478149414, + "num_tokens": 445893163.0, + "step": 11484 + }, + { + "epoch": 1.4610100496120086, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.159381866455078, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8766018152236938, + "num_tokens": 445926127.0, + "step": 11485 + }, + { + "epoch": 1.4611372598905992, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.966312408447266, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8941973447799683, + "num_tokens": 445968037.0, + "step": 11486 + }, + { + "epoch": 1.4612644701691897, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.986498832702637, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8844695091247559, + "num_tokens": 446008656.0, + "step": 11487 + }, + { + "epoch": 1.4613916804477802, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.050093650817871, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8794865608215332, + "num_tokens": 446047715.0, + "step": 11488 + }, + { + "epoch": 1.4615188907263708, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.964248657226562, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.892092227935791, + "num_tokens": 446085743.0, + "step": 11489 + }, + { + "epoch": 1.4616461010049613, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.063600540161133, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8730201721191406, + "num_tokens": 446125418.0, + "step": 11490 + }, + { + "epoch": 1.4617733112835518, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.053661346435547, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8895929455757141, + "num_tokens": 446165420.0, + "step": 11491 + }, + { + "epoch": 1.4619005215621423, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.945582389831543, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.889357328414917, + "num_tokens": 446208592.0, + "step": 11492 + }, + { + "epoch": 1.4620277318407329, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.111928939819336, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8818281888961792, + "num_tokens": 446244116.0, + "step": 11493 + }, + { + "epoch": 1.4621549421193232, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.972047805786133, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8878011107444763, + "num_tokens": 446285991.0, + "step": 11494 + }, + { + "epoch": 1.4622821523979137, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.008023262023926, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8785707950592041, + "num_tokens": 446328161.0, + "step": 11495 + }, + { + "epoch": 1.4624093626765042, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.033523559570312, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.880874514579773, + "num_tokens": 446369894.0, + "step": 11496 + }, + { + "epoch": 1.4625365729550948, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.068453788757324, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8786015510559082, + "num_tokens": 446406824.0, + "step": 11497 + }, + { + "epoch": 1.4626637832336853, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.061202049255371, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8873496055603027, + "num_tokens": 446444791.0, + "step": 11498 + }, + { + "epoch": 1.4627909935122758, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.034993171691895, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8873907327651978, + "num_tokens": 446482643.0, + "step": 11499 + }, + { + "epoch": 1.4629182037908663, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.01711368560791, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8938301801681519, + "num_tokens": 446520799.0, + "step": 11500 + }, + { + "epoch": 1.4630454140694569, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.988423347473145, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8911260366439819, + "num_tokens": 446559781.0, + "step": 11501 + }, + { + "epoch": 1.4631726243480474, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.039006233215332, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8833794593811035, + "num_tokens": 446601023.0, + "step": 11502 + }, + { + "epoch": 1.4632998346266377, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.048439979553223, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8801321387290955, + "num_tokens": 446643402.0, + "step": 11503 + }, + { + "epoch": 1.4634270449052282, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.022345542907715, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8898105621337891, + "num_tokens": 446680236.0, + "step": 11504 + }, + { + "epoch": 1.4635542551838188, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.994970321655273, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8770759701728821, + "num_tokens": 446720235.0, + "step": 11505 + }, + { + "epoch": 1.4636814654624093, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.065410614013672, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8983204364776611, + "num_tokens": 446753947.0, + "step": 11506 + }, + { + "epoch": 1.4638086757409998, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.990313529968262, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8869922161102295, + "num_tokens": 446791719.0, + "step": 11507 + }, + { + "epoch": 1.4639358860195903, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.993401527404785, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.894680917263031, + "num_tokens": 446828437.0, + "step": 11508 + }, + { + "epoch": 1.4640630962981809, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.967302322387695, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8953768014907837, + "num_tokens": 446865345.0, + "step": 11509 + }, + { + "epoch": 1.4641903065767714, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.893999099731445, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8917755484580994, + "num_tokens": 446909045.0, + "step": 11510 + }, + { + "epoch": 1.464317516855362, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.094090461730957, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8908123970031738, + "num_tokens": 446946149.0, + "step": 11511 + }, + { + "epoch": 1.4644447271339525, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.033777236938477, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8906354308128357, + "num_tokens": 446980936.0, + "step": 11512 + }, + { + "epoch": 1.464571937412543, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.091592788696289, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8831733465194702, + "num_tokens": 447020518.0, + "step": 11513 + }, + { + "epoch": 1.4646991476911335, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.963102340698242, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8933731317520142, + "num_tokens": 447063630.0, + "step": 11514 + }, + { + "epoch": 1.464826357969724, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.025764465332031, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8904761672019958, + "num_tokens": 447100429.0, + "step": 11515 + }, + { + "epoch": 1.4649535682483146, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.958770751953125, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8900114297866821, + "num_tokens": 447141873.0, + "step": 11516 + }, + { + "epoch": 1.465080778526905, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.364418029785156e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.143043518066406, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8864846229553223, + "num_tokens": 447179325.0, + "step": 11517 + }, + { + "epoch": 1.4652079888054954, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.972496032714844, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8846015334129333, + "num_tokens": 447226876.0, + "step": 11518 + }, + { + "epoch": 1.465335199084086, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.068568229675293, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8809964060783386, + "num_tokens": 447265999.0, + "step": 11519 + }, + { + "epoch": 1.4654624093626765, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.022594451904297, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8970547914505005, + "num_tokens": 447301849.0, + "step": 11520 + }, + { + "epoch": 1.465589619641267, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.18661117553711, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.895889401435852, + "num_tokens": 447329793.0, + "step": 11521 + }, + { + "epoch": 1.4657168299198575, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.966242790222168, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8925520181655884, + "num_tokens": 447369826.0, + "step": 11522 + }, + { + "epoch": 1.465844040198448, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.023634910583496, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8750380277633667, + "num_tokens": 447408158.0, + "step": 11523 + }, + { + "epoch": 1.4659712504770386, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.055829048156738, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8852184414863586, + "num_tokens": 447444917.0, + "step": 11524 + }, + { + "epoch": 1.466098460755629, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.974688529968262, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8940162062644958, + "num_tokens": 447481720.0, + "step": 11525 + }, + { + "epoch": 1.4662256710342196, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.002986907958984, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8818362951278687, + "num_tokens": 447527721.0, + "step": 11526 + }, + { + "epoch": 1.46635288131281, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.969534873962402, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.9042230248451233, + "num_tokens": 447566914.0, + "step": 11527 + }, + { + "epoch": 1.4664800915914005, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.033305168151855, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8830980658531189, + "num_tokens": 447604651.0, + "step": 11528 + }, + { + "epoch": 1.466607301869991, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.053516387939453, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.883760929107666, + "num_tokens": 447640185.0, + "step": 11529 + }, + { + "epoch": 1.4667345121485815, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.995408058166504, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8868932723999023, + "num_tokens": 447681756.0, + "step": 11530 + }, + { + "epoch": 1.466861722427172, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.05533218383789, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8830009698867798, + "num_tokens": 447720122.0, + "step": 11531 + }, + { + "epoch": 1.4669889327057626, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.080978393554688, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8874533176422119, + "num_tokens": 447754931.0, + "step": 11532 + }, + { + "epoch": 1.467116142984353, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.054490089416504, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8897862434387207, + "num_tokens": 447790084.0, + "step": 11533 + }, + { + "epoch": 1.4672433532629436, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.01892375946045, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8836917877197266, + "num_tokens": 447832068.0, + "step": 11534 + }, + { + "epoch": 1.4673705635415342, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.131464004516602, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8940274119377136, + "num_tokens": 447866843.0, + "step": 11535 + }, + { + "epoch": 1.4674977738201247, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.104757308959961, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8897427320480347, + "num_tokens": 447901821.0, + "step": 11536 + }, + { + "epoch": 1.4676249840987152, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.053308486938477, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8907694816589355, + "num_tokens": 447939305.0, + "step": 11537 + }, + { + "epoch": 1.4677521943773058, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.99743938446045, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8895952701568604, + "num_tokens": 447980890.0, + "step": 11538 + }, + { + "epoch": 1.4678794046558963, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.018436431884766, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.887688934803009, + "num_tokens": 448019258.0, + "step": 11539 + }, + { + "epoch": 1.4680066149344868, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.124183654785156, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8902137875556946, + "num_tokens": 448054887.0, + "step": 11540 + }, + { + "epoch": 1.4681338252130773, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.141683578491211, + "learning_rate": 1e-06, + "loss": 0.368, + "mean_token_accuracy": 0.8783674836158752, + "num_tokens": 448090006.0, + "step": 11541 + }, + { + "epoch": 1.4682610354916679, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.994644165039062, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8874262571334839, + "num_tokens": 448129153.0, + "step": 11542 + }, + { + "epoch": 1.4683882457702582, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.03699779510498, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8896848559379578, + "num_tokens": 448168614.0, + "step": 11543 + }, + { + "epoch": 1.4685154560488487, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.122847557067871, + "learning_rate": 1e-06, + "loss": 0.3783, + "mean_token_accuracy": 0.8716854453086853, + "num_tokens": 448209413.0, + "step": 11544 + }, + { + "epoch": 1.4686426663274392, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.094643592834473, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8827917575836182, + "num_tokens": 448244017.0, + "step": 11545 + }, + { + "epoch": 1.4687698766060298, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.958786964416504, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8960838317871094, + "num_tokens": 448286949.0, + "step": 11546 + }, + { + "epoch": 1.4688970868846203, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.034857749938965, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8931158781051636, + "num_tokens": 448322417.0, + "step": 11547 + }, + { + "epoch": 1.4690242971632108, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.3882598876953125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 8.984302520751953, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8975594639778137, + "num_tokens": 448361648.0, + "step": 11548 + }, + { + "epoch": 1.4691515074418013, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.11116886138916, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8994162082672119, + "num_tokens": 448392963.0, + "step": 11549 + }, + { + "epoch": 1.4692787177203919, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.04806137084961, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8952197432518005, + "num_tokens": 448430104.0, + "step": 11550 + }, + { + "epoch": 1.4694059279989824, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.903855323791504, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8994787335395813, + "num_tokens": 448475250.0, + "step": 11551 + }, + { + "epoch": 1.4695331382775727, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.0360107421875, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.894752025604248, + "num_tokens": 448513423.0, + "step": 11552 + }, + { + "epoch": 1.4696603485561632, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.056703567504883, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8842968344688416, + "num_tokens": 448552638.0, + "step": 11553 + }, + { + "epoch": 1.4697875588347538, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.072733879089355, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8751026391983032, + "num_tokens": 448592197.0, + "step": 11554 + }, + { + "epoch": 1.4699147691133443, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.966551780700684, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8943114876747131, + "num_tokens": 448635166.0, + "step": 11555 + }, + { + "epoch": 1.4700419793919348, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.003307342529297, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8897411823272705, + "num_tokens": 448675973.0, + "step": 11556 + }, + { + "epoch": 1.4701691896705253, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.043190956115723, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8853713870048523, + "num_tokens": 448713705.0, + "step": 11557 + }, + { + "epoch": 1.4702963999491159, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.03955078125, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8884537220001221, + "num_tokens": 448755691.0, + "step": 11558 + }, + { + "epoch": 1.4704236102277064, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.177506446838379, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.8729441165924072, + "num_tokens": 448790804.0, + "step": 11559 + }, + { + "epoch": 1.470550820506297, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.96587085723877, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.891157865524292, + "num_tokens": 448835989.0, + "step": 11560 + }, + { + "epoch": 1.4706780307848875, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.072778701782227, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8885999321937561, + "num_tokens": 448870136.0, + "step": 11561 + }, + { + "epoch": 1.470805241063478, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.005912780761719, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8882489204406738, + "num_tokens": 448915533.0, + "step": 11562 + }, + { + "epoch": 1.4709324513420685, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.980022430419922, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8840645551681519, + "num_tokens": 448954895.0, + "step": 11563 + }, + { + "epoch": 1.471059661620659, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.043656349182129, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8948321342468262, + "num_tokens": 448993555.0, + "step": 11564 + }, + { + "epoch": 1.4711868718992496, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.015069961547852, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8863536715507507, + "num_tokens": 449033569.0, + "step": 11565 + }, + { + "epoch": 1.47131408217784, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.074073791503906, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8989032506942749, + "num_tokens": 449065199.0, + "step": 11566 + }, + { + "epoch": 1.4714412924564304, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.120058059692383, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8893613219261169, + "num_tokens": 449099319.0, + "step": 11567 + }, + { + "epoch": 1.471568502735021, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.98575496673584, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8943507671356201, + "num_tokens": 449140416.0, + "step": 11568 + }, + { + "epoch": 1.4716957130136115, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.140241622924805, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8734470009803772, + "num_tokens": 449175548.0, + "step": 11569 + }, + { + "epoch": 1.471822923292202, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.080440521240234, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8808314204216003, + "num_tokens": 449215926.0, + "step": 11570 + }, + { + "epoch": 1.4719501335707925, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.178638458251953, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.8716981410980225, + "num_tokens": 449248557.0, + "step": 11571 + }, + { + "epoch": 1.472077343849383, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.972248077392578, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8875627517700195, + "num_tokens": 449290796.0, + "step": 11572 + }, + { + "epoch": 1.4722045541279736, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.038902282714844, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8782482147216797, + "num_tokens": 449330144.0, + "step": 11573 + }, + { + "epoch": 1.472331764406564, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.002205848693848, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8987756967544556, + "num_tokens": 449370125.0, + "step": 11574 + }, + { + "epoch": 1.4724589746851546, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.05156135559082, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8939729928970337, + "num_tokens": 449408612.0, + "step": 11575 + }, + { + "epoch": 1.472586184963745, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.148244857788086, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8804216980934143, + "num_tokens": 449446933.0, + "step": 11576 + }, + { + "epoch": 1.4727133952423355, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.939641952514648, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8886286020278931, + "num_tokens": 449489837.0, + "step": 11577 + }, + { + "epoch": 1.472840605520926, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.030754089355469, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8854048252105713, + "num_tokens": 449531492.0, + "step": 11578 + }, + { + "epoch": 1.4729678157995165, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.022597312927246, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8794812560081482, + "num_tokens": 449574663.0, + "step": 11579 + }, + { + "epoch": 1.473095026078107, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.074392318725586, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8890696167945862, + "num_tokens": 449610172.0, + "step": 11580 + }, + { + "epoch": 1.4732222363566976, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.97076416015625, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8949047327041626, + "num_tokens": 449657017.0, + "step": 11581 + }, + { + "epoch": 1.473349446635288, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.00780200958252, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.9061675071716309, + "num_tokens": 449699051.0, + "step": 11582 + }, + { + "epoch": 1.4734766569138786, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.093999862670898, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8834520578384399, + "num_tokens": 449738786.0, + "step": 11583 + }, + { + "epoch": 1.4736038671924692, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.94605541229248, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8993242979049683, + "num_tokens": 449784763.0, + "step": 11584 + }, + { + "epoch": 1.4737310774710597, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.050288200378418, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8920530676841736, + "num_tokens": 449822766.0, + "step": 11585 + }, + { + "epoch": 1.4738582877496502, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.068090438842773, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8888101577758789, + "num_tokens": 449863056.0, + "step": 11586 + }, + { + "epoch": 1.4739854980282407, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.97722053527832, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.901983380317688, + "num_tokens": 449902890.0, + "step": 11587 + }, + { + "epoch": 1.4741127083068313, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.072394371032715, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8924440145492554, + "num_tokens": 449940138.0, + "step": 11588 + }, + { + "epoch": 1.4742399185854218, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.045516014099121, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8998984098434448, + "num_tokens": 449976958.0, + "step": 11589 + }, + { + "epoch": 1.4743671288640123, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.118501663208008, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8900136351585388, + "num_tokens": 450011187.0, + "step": 11590 + }, + { + "epoch": 1.4744943391426026, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.131983757019043, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8927211165428162, + "num_tokens": 450046631.0, + "step": 11591 + }, + { + "epoch": 1.4746215494211932, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.018016815185547, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8921066522598267, + "num_tokens": 450083348.0, + "step": 11592 + }, + { + "epoch": 1.4747487596997837, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.988852500915527, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8963111639022827, + "num_tokens": 450119248.0, + "step": 11593 + }, + { + "epoch": 1.4748759699783742, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.16841983795166, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8865404725074768, + "num_tokens": 450154511.0, + "step": 11594 + }, + { + "epoch": 1.4750031802569648, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.302288055419922, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.888900637626648, + "num_tokens": 450197404.0, + "step": 11595 + }, + { + "epoch": 1.4751303905355553, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.99417495727539, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8941342830657959, + "num_tokens": 450233102.0, + "step": 11596 + }, + { + "epoch": 1.4752576008141458, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.957874298095703, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8971360921859741, + "num_tokens": 450274103.0, + "step": 11597 + }, + { + "epoch": 1.4753848110927363, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.002902030944824, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8924182653427124, + "num_tokens": 450316671.0, + "step": 11598 + }, + { + "epoch": 1.4755120213713269, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.999495506286621, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8832393884658813, + "num_tokens": 450356077.0, + "step": 11599 + }, + { + "epoch": 1.4756392316499174, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.96906566619873, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8943915367126465, + "num_tokens": 450399811.0, + "step": 11600 + }, + { + "epoch": 1.4757664419285077, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.023184776306152, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8967775702476501, + "num_tokens": 450437067.0, + "step": 11601 + }, + { + "epoch": 1.4758936522070982, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.027556419372559, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8898594379425049, + "num_tokens": 450476252.0, + "step": 11602 + }, + { + "epoch": 1.4760208624856888, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.029396057128906, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8796647787094116, + "num_tokens": 450514778.0, + "step": 11603 + }, + { + "epoch": 1.4761480727642793, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.948976516723633, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8949199318885803, + "num_tokens": 450554682.0, + "step": 11604 + }, + { + "epoch": 1.4762752830428698, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.98680591583252, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8990395069122314, + "num_tokens": 450593738.0, + "step": 11605 + }, + { + "epoch": 1.4764024933214603, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.940986633300781, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8957057595252991, + "num_tokens": 450636809.0, + "step": 11606 + }, + { + "epoch": 1.4765297036000509, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.936594009399414, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8874026536941528, + "num_tokens": 450680326.0, + "step": 11607 + }, + { + "epoch": 1.4766569138786414, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 9.028557777404785, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8897022604942322, + "num_tokens": 450722577.0, + "step": 11608 + }, + { + "epoch": 1.476784124157232, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.0614595413208, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.9016480445861816, + "num_tokens": 450754971.0, + "step": 11609 + }, + { + "epoch": 1.4769113344358225, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.996532440185547, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8857299089431763, + "num_tokens": 450798806.0, + "step": 11610 + }, + { + "epoch": 1.477038544714413, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.12387466430664, + "learning_rate": 1e-06, + "loss": 0.3631, + "mean_token_accuracy": 0.8777821660041809, + "num_tokens": 450835290.0, + "step": 11611 + }, + { + "epoch": 1.4771657549930035, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.975618362426758, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8891481161117554, + "num_tokens": 450880781.0, + "step": 11612 + }, + { + "epoch": 1.477292965271594, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.069046974182129, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8799786567687988, + "num_tokens": 450920385.0, + "step": 11613 + }, + { + "epoch": 1.4774201755501846, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.949538230895996, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9143107533454895, + "num_tokens": 450959446.0, + "step": 11614 + }, + { + "epoch": 1.477547385828775, + "ewc_loss": 0.01556396484375, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0132789611816406e-05, + "grad_norm": 8.944576263427734, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8942164182662964, + "num_tokens": 451001887.0, + "step": 11615 + }, + { + "epoch": 1.4776745961073654, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.151793479919434, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8823078870773315, + "num_tokens": 451036811.0, + "step": 11616 + }, + { + "epoch": 1.477801806385956, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.988472938537598, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8813983201980591, + "num_tokens": 451082016.0, + "step": 11617 + }, + { + "epoch": 1.4779290166645465, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.024198532104492, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8936225175857544, + "num_tokens": 451119721.0, + "step": 11618 + }, + { + "epoch": 1.478056226943137, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.117563247680664, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.885939359664917, + "num_tokens": 451154261.0, + "step": 11619 + }, + { + "epoch": 1.4781834372217275, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.005083084106445, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8845525979995728, + "num_tokens": 451203501.0, + "step": 11620 + }, + { + "epoch": 1.478310647500318, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.032159805297852, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8832912445068359, + "num_tokens": 451243632.0, + "step": 11621 + }, + { + "epoch": 1.4784378577789086, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.146117210388184, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8928113579750061, + "num_tokens": 451277322.0, + "step": 11622 + }, + { + "epoch": 1.478565068057499, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.108833312988281, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8812668323516846, + "num_tokens": 451320272.0, + "step": 11623 + }, + { + "epoch": 1.4786922783360896, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.48493480682373, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8879231214523315, + "num_tokens": 451355730.0, + "step": 11624 + }, + { + "epoch": 1.47881948861468, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.050107955932617, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8893336057662964, + "num_tokens": 451391710.0, + "step": 11625 + }, + { + "epoch": 1.4789466988932705, + "ewc_loss": 0.015625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0192394256591797e-05, + "grad_norm": 8.889655113220215, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.9060506820678711, + "num_tokens": 451430709.0, + "step": 11626 + }, + { + "epoch": 1.479073909171861, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.084964752197266, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8840471506118774, + "num_tokens": 451465629.0, + "step": 11627 + }, + { + "epoch": 1.4792011194504515, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.935161590576172, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8810761570930481, + "num_tokens": 451504986.0, + "step": 11628 + }, + { + "epoch": 1.479328329729042, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.170059204101562, + "learning_rate": 1e-06, + "loss": 0.4185, + "mean_token_accuracy": 0.8600886464118958, + "num_tokens": 451547938.0, + "step": 11629 + }, + { + "epoch": 1.4794555400076326, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.109278678894043, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8813608288764954, + "num_tokens": 451580184.0, + "step": 11630 + }, + { + "epoch": 1.479582750286223, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.009984016418457, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8801475763320923, + "num_tokens": 451619403.0, + "step": 11631 + }, + { + "epoch": 1.4797099605648136, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.03329086303711, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8776158094406128, + "num_tokens": 451659209.0, + "step": 11632 + }, + { + "epoch": 1.4798371708434042, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.061880111694336, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8873884677886963, + "num_tokens": 451697248.0, + "step": 11633 + }, + { + "epoch": 1.4799643811219947, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.026408195495605, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8876891732215881, + "num_tokens": 451734863.0, + "step": 11634 + }, + { + "epoch": 1.4800915914005852, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.108366966247559, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8961913585662842, + "num_tokens": 451768009.0, + "step": 11635 + }, + { + "epoch": 1.4802188016791757, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.055122375488281, + "learning_rate": 1e-06, + "loss": 0.3744, + "mean_token_accuracy": 0.87286376953125, + "num_tokens": 451813940.0, + "step": 11636 + }, + { + "epoch": 1.4803460119577663, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.062424659729004, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8975252509117126, + "num_tokens": 451851220.0, + "step": 11637 + }, + { + "epoch": 1.4804732222363568, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.946208000183105, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8939120173454285, + "num_tokens": 451891201.0, + "step": 11638 + }, + { + "epoch": 1.4806004325149473, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.412101745605469e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.005803108215332, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8770006895065308, + "num_tokens": 451933325.0, + "step": 11639 + }, + { + "epoch": 1.4807276427935376, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.029435157775879, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8869389891624451, + "num_tokens": 451977539.0, + "step": 11640 + }, + { + "epoch": 1.4808548530721282, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.00226879119873, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.89060378074646, + "num_tokens": 452015395.0, + "step": 11641 + }, + { + "epoch": 1.4809820633507187, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.435943603515625e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.109670639038086, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8846085071563721, + "num_tokens": 452051590.0, + "step": 11642 + }, + { + "epoch": 1.4811092736293092, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.01341438293457, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8839424252510071, + "num_tokens": 452094011.0, + "step": 11643 + }, + { + "epoch": 1.4812364839078997, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.055863380432129, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8736568689346313, + "num_tokens": 452135337.0, + "step": 11644 + }, + { + "epoch": 1.4813636941864903, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.094635963439941, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8928111791610718, + "num_tokens": 452165130.0, + "step": 11645 + }, + { + "epoch": 1.4814909044650808, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.971664428710938, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8987438678741455, + "num_tokens": 452204966.0, + "step": 11646 + }, + { + "epoch": 1.4816181147436713, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.044480323791504, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8855091333389282, + "num_tokens": 452244200.0, + "step": 11647 + }, + { + "epoch": 1.4817453250222619, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.027141571044922, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8921247720718384, + "num_tokens": 452283043.0, + "step": 11648 + }, + { + "epoch": 1.4818725353008524, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.045533180236816, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8812390565872192, + "num_tokens": 452321095.0, + "step": 11649 + }, + { + "epoch": 1.4819997455794427, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.03446102142334, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8896491527557373, + "num_tokens": 452368925.0, + "step": 11650 + }, + { + "epoch": 1.4821269558580332, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.051365852355957, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8848716616630554, + "num_tokens": 452407049.0, + "step": 11651 + }, + { + "epoch": 1.4822541661366238, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.152318000793457, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8781962990760803, + "num_tokens": 452446288.0, + "step": 11652 + }, + { + "epoch": 1.4823813764152143, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.00082778930664, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8867565989494324, + "num_tokens": 452491245.0, + "step": 11653 + }, + { + "epoch": 1.4825085866938048, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.047618865966797, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8865519762039185, + "num_tokens": 452533360.0, + "step": 11654 + }, + { + "epoch": 1.4826357969723953, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.155660629272461, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8872723579406738, + "num_tokens": 452566878.0, + "step": 11655 + }, + { + "epoch": 1.4827630072509859, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.979391098022461, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8894398808479309, + "num_tokens": 452605235.0, + "step": 11656 + }, + { + "epoch": 1.4828902175295764, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.086600303649902, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8846259713172913, + "num_tokens": 452640251.0, + "step": 11657 + }, + { + "epoch": 1.483017427808167, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.218745231628418, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8894894123077393, + "num_tokens": 452667006.0, + "step": 11658 + }, + { + "epoch": 1.4831446380867574, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.947233200073242, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8929624557495117, + "num_tokens": 452711772.0, + "step": 11659 + }, + { + "epoch": 1.483271848365348, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.99421501159668, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8870025873184204, + "num_tokens": 452751663.0, + "step": 11660 + }, + { + "epoch": 1.4833990586439385, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.035703659057617, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8925875425338745, + "num_tokens": 452791050.0, + "step": 11661 + }, + { + "epoch": 1.483526268922529, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.045555114746094, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8784307837486267, + "num_tokens": 452834413.0, + "step": 11662 + }, + { + "epoch": 1.4836534792011196, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.930874824523926, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.9002411365509033, + "num_tokens": 452872750.0, + "step": 11663 + }, + { + "epoch": 1.48378068947971, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.025650978088379, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8912028074264526, + "num_tokens": 452919226.0, + "step": 11664 + }, + { + "epoch": 1.4839078997583004, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.052265167236328, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8937532305717468, + "num_tokens": 452956456.0, + "step": 11665 + }, + { + "epoch": 1.484035110036891, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.050471305847168, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8973816633224487, + "num_tokens": 452997544.0, + "step": 11666 + }, + { + "epoch": 1.4841623203154815, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.16046142578125, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8863341808319092, + "num_tokens": 453028609.0, + "step": 11667 + }, + { + "epoch": 1.484289530594072, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.971173286437988, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8900080919265747, + "num_tokens": 453075648.0, + "step": 11668 + }, + { + "epoch": 1.4844167408726625, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.964487075805664, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8957552313804626, + "num_tokens": 453115892.0, + "step": 11669 + }, + { + "epoch": 1.484543951151253, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.006224632263184, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8950836658477783, + "num_tokens": 453156978.0, + "step": 11670 + }, + { + "epoch": 1.4846711614298436, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.016518592834473, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8944190740585327, + "num_tokens": 453195273.0, + "step": 11671 + }, + { + "epoch": 1.484798371708434, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.08071517944336, + "learning_rate": 1e-06, + "loss": 0.3921, + "mean_token_accuracy": 0.866732120513916, + "num_tokens": 453236189.0, + "step": 11672 + }, + { + "epoch": 1.4849255819870246, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.059263229370117, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8828058838844299, + "num_tokens": 453271670.0, + "step": 11673 + }, + { + "epoch": 1.485052792265615, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.95217514038086, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.893937885761261, + "num_tokens": 453311451.0, + "step": 11674 + }, + { + "epoch": 1.4851800025442055, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.05079460144043, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.889750599861145, + "num_tokens": 453349235.0, + "step": 11675 + }, + { + "epoch": 1.485307212822796, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.971604347229004, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8969525098800659, + "num_tokens": 453389038.0, + "step": 11676 + }, + { + "epoch": 1.4854344231013865, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.038525581359863, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9000028371810913, + "num_tokens": 453425135.0, + "step": 11677 + }, + { + "epoch": 1.485561633379977, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.115254402160645, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8776094913482666, + "num_tokens": 453460884.0, + "step": 11678 + }, + { + "epoch": 1.4856888436585676, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.079998016357422, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8807217478752136, + "num_tokens": 453495820.0, + "step": 11679 + }, + { + "epoch": 1.485816053937158, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.05007553100586, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8758646249771118, + "num_tokens": 453537765.0, + "step": 11680 + }, + { + "epoch": 1.4859432642157486, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.016730308532715, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8887403011322021, + "num_tokens": 453577406.0, + "step": 11681 + }, + { + "epoch": 1.4860704744943392, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.998650550842285, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8870381116867065, + "num_tokens": 453619477.0, + "step": 11682 + }, + { + "epoch": 1.4861976847729297, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.020331382751465, + "learning_rate": 1e-06, + "loss": 0.379, + "mean_token_accuracy": 0.8712787628173828, + "num_tokens": 453663195.0, + "step": 11683 + }, + { + "epoch": 1.4863248950515202, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.001261711120605, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8869701027870178, + "num_tokens": 453701764.0, + "step": 11684 + }, + { + "epoch": 1.4864521053301107, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.046277046203613, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8895177245140076, + "num_tokens": 453742979.0, + "step": 11685 + }, + { + "epoch": 1.4865793156087013, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.039811134338379, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8908296823501587, + "num_tokens": 453779596.0, + "step": 11686 + }, + { + "epoch": 1.4867065258872918, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.000511169433594, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.9001496434211731, + "num_tokens": 453814367.0, + "step": 11687 + }, + { + "epoch": 1.4868337361658823, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.954636573791504, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.8993541598320007, + "num_tokens": 453854285.0, + "step": 11688 + }, + { + "epoch": 1.4869609464444726, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.027117729187012, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.9012418389320374, + "num_tokens": 453891419.0, + "step": 11689 + }, + { + "epoch": 1.4870881567230632, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.958823204040527, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8972551822662354, + "num_tokens": 453932873.0, + "step": 11690 + }, + { + "epoch": 1.4872153670016537, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.186813354492188, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8861873745918274, + "num_tokens": 453965695.0, + "step": 11691 + }, + { + "epoch": 1.4873425772802442, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.909642219543457, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8888717293739319, + "num_tokens": 454006560.0, + "step": 11692 + }, + { + "epoch": 1.4874697875588347, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.069149017333984, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8933901786804199, + "num_tokens": 454040921.0, + "step": 11693 + }, + { + "epoch": 1.4875969978374253, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.980876922607422, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8816710114479065, + "num_tokens": 454087067.0, + "step": 11694 + }, + { + "epoch": 1.4877242081160158, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.023481369018555, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8874489068984985, + "num_tokens": 454124292.0, + "step": 11695 + }, + { + "epoch": 1.4878514183946063, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.076695442199707, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8810119032859802, + "num_tokens": 454161838.0, + "step": 11696 + }, + { + "epoch": 1.4879786286731969, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.168096542358398, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8871859312057495, + "num_tokens": 454193442.0, + "step": 11697 + }, + { + "epoch": 1.4881058389517874, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.951163291931152, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.9053540229797363, + "num_tokens": 454229770.0, + "step": 11698 + }, + { + "epoch": 1.4882330492303777, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.090171813964844, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8764945268630981, + "num_tokens": 454269504.0, + "step": 11699 + }, + { + "epoch": 1.4883602595089682, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.001791000366211, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8895647525787354, + "num_tokens": 454310391.0, + "step": 11700 + }, + { + "epoch": 1.4884874697875587, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.977778434753418, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8888251781463623, + "num_tokens": 454352587.0, + "step": 11701 + }, + { + "epoch": 1.4886146800661493, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.089282035827637, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.9010988473892212, + "num_tokens": 454389746.0, + "step": 11702 + }, + { + "epoch": 1.4887418903447398, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.195911407470703, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8901523351669312, + "num_tokens": 454421887.0, + "step": 11703 + }, + { + "epoch": 1.4888691006233303, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.971986770629883, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8920155763626099, + "num_tokens": 454463086.0, + "step": 11704 + }, + { + "epoch": 1.4889963109019209, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.059148788452148, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8849559426307678, + "num_tokens": 454498098.0, + "step": 11705 + }, + { + "epoch": 1.4891235211805114, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.055328369140625, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8953012824058533, + "num_tokens": 454535756.0, + "step": 11706 + }, + { + "epoch": 1.489250731459102, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.96562671661377, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8920414447784424, + "num_tokens": 454575194.0, + "step": 11707 + }, + { + "epoch": 1.4893779417376924, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.088171005249023, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8782894611358643, + "num_tokens": 454617452.0, + "step": 11708 + }, + { + "epoch": 1.489505152016283, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.031890869140625, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.9003797173500061, + "num_tokens": 454655545.0, + "step": 11709 + }, + { + "epoch": 1.4896323622948735, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.459785461425781e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.04497241973877, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8906201124191284, + "num_tokens": 454697788.0, + "step": 11710 + }, + { + "epoch": 1.489759572573464, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.12782096862793, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8975380659103394, + "num_tokens": 454731191.0, + "step": 11711 + }, + { + "epoch": 1.4898867828520546, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.050130844116211, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.88286954164505, + "num_tokens": 454774151.0, + "step": 11712 + }, + { + "epoch": 1.490013993130645, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.022917747497559, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8910688757896423, + "num_tokens": 454810238.0, + "step": 11713 + }, + { + "epoch": 1.4901412034092354, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.928901672363281, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8940396904945374, + "num_tokens": 454851197.0, + "step": 11714 + }, + { + "epoch": 1.490268413687826, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.130928993225098, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8776249885559082, + "num_tokens": 454889890.0, + "step": 11715 + }, + { + "epoch": 1.4903956239664164, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.966821670532227, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8878089785575867, + "num_tokens": 454929496.0, + "step": 11716 + }, + { + "epoch": 1.490522834245007, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.028635025024414, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8840757608413696, + "num_tokens": 454967128.0, + "step": 11717 + }, + { + "epoch": 1.4906500445235975, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.983548164367676, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.9037920236587524, + "num_tokens": 455007841.0, + "step": 11718 + }, + { + "epoch": 1.490777254802188, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.029627799987793, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8946101665496826, + "num_tokens": 455042339.0, + "step": 11719 + }, + { + "epoch": 1.4909044650807786, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.11264705657959, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8768660426139832, + "num_tokens": 455080575.0, + "step": 11720 + }, + { + "epoch": 1.491031675359369, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.970450401306152, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.9007653594017029, + "num_tokens": 455114994.0, + "step": 11721 + }, + { + "epoch": 1.4911588856379596, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.044941902160645, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8726500272750854, + "num_tokens": 455156733.0, + "step": 11722 + }, + { + "epoch": 1.49128609591655, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.184157371520996, + "learning_rate": 1e-06, + "loss": 0.387, + "mean_token_accuracy": 0.8683802485466003, + "num_tokens": 455191861.0, + "step": 11723 + }, + { + "epoch": 1.4914133061951405, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.037921905517578, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8790061473846436, + "num_tokens": 455234165.0, + "step": 11724 + }, + { + "epoch": 1.491540516473731, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.004972457885742, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8864096999168396, + "num_tokens": 455271695.0, + "step": 11725 + }, + { + "epoch": 1.4916677267523215, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.996278762817383, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9021831154823303, + "num_tokens": 455313172.0, + "step": 11726 + }, + { + "epoch": 1.491794937030912, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.031874656677246, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8806967735290527, + "num_tokens": 455351828.0, + "step": 11727 + }, + { + "epoch": 1.4919221473095026, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.02220630645752, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8982147574424744, + "num_tokens": 455391125.0, + "step": 11728 + }, + { + "epoch": 1.492049357588093, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.948328018188477, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8887965679168701, + "num_tokens": 455435401.0, + "step": 11729 + }, + { + "epoch": 1.4921765678666836, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.104974746704102, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8914257287979126, + "num_tokens": 455476015.0, + "step": 11730 + }, + { + "epoch": 1.4923037781452742, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.038731575012207, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8825401067733765, + "num_tokens": 455517670.0, + "step": 11731 + }, + { + "epoch": 1.4924309884238647, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.133535385131836, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8823665976524353, + "num_tokens": 455551566.0, + "step": 11732 + }, + { + "epoch": 1.4925581987024552, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.132770538330078, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8835795521736145, + "num_tokens": 455589230.0, + "step": 11733 + }, + { + "epoch": 1.4926854089810457, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.036641120910645, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8839349150657654, + "num_tokens": 455630341.0, + "step": 11734 + }, + { + "epoch": 1.4928126192596363, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.010869979858398, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8923841714859009, + "num_tokens": 455675219.0, + "step": 11735 + }, + { + "epoch": 1.4929398295382268, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.045271873474121, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8882399201393127, + "num_tokens": 455712641.0, + "step": 11736 + }, + { + "epoch": 1.4930670398168173, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.090924263000488, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8945519328117371, + "num_tokens": 455750841.0, + "step": 11737 + }, + { + "epoch": 1.4931942500954076, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.052933692932129, + "learning_rate": 1e-06, + "loss": 0.3647, + "mean_token_accuracy": 0.8752644658088684, + "num_tokens": 455791101.0, + "step": 11738 + }, + { + "epoch": 1.4933214603739982, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.028932571411133, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.9006742835044861, + "num_tokens": 455828816.0, + "step": 11739 + }, + { + "epoch": 1.4934486706525887, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.024100303649902, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.892142653465271, + "num_tokens": 455867475.0, + "step": 11740 + }, + { + "epoch": 1.4935758809311792, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.109487533569336, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.878230094909668, + "num_tokens": 455908494.0, + "step": 11741 + }, + { + "epoch": 1.4937030912097697, + "ewc_loss": 0.0157470703125, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.059473037719727, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8780554533004761, + "num_tokens": 455947649.0, + "step": 11742 + }, + { + "epoch": 1.4938303014883603, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.153829574584961, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8706941604614258, + "num_tokens": 455978800.0, + "step": 11743 + }, + { + "epoch": 1.4939575117669508, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.114034652709961, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8810256123542786, + "num_tokens": 456011924.0, + "step": 11744 + }, + { + "epoch": 1.4940847220455413, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.965649604797363, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.885048508644104, + "num_tokens": 456056420.0, + "step": 11745 + }, + { + "epoch": 1.4942119323241319, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.158199310302734, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8806988596916199, + "num_tokens": 456087440.0, + "step": 11746 + }, + { + "epoch": 1.4943391426027224, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.02610969543457, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8766844272613525, + "num_tokens": 456131084.0, + "step": 11747 + }, + { + "epoch": 1.4944663528813127, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.993264198303223, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8949487209320068, + "num_tokens": 456164860.0, + "step": 11748 + }, + { + "epoch": 1.4945935631599032, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.055903434753418, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8947647213935852, + "num_tokens": 456202496.0, + "step": 11749 + }, + { + "epoch": 1.4947207734384937, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.016761779785156, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8870763778686523, + "num_tokens": 456243947.0, + "step": 11750 + }, + { + "epoch": 1.4948479837170843, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.097582817077637, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8873323202133179, + "num_tokens": 456278801.0, + "step": 11751 + }, + { + "epoch": 1.4949751939956748, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.01391887664795, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8961670994758606, + "num_tokens": 456317176.0, + "step": 11752 + }, + { + "epoch": 1.4951024042742653, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.033722877502441, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.892801821231842, + "num_tokens": 456357608.0, + "step": 11753 + }, + { + "epoch": 1.4952296145528559, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.9906644821167, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8858352899551392, + "num_tokens": 456398384.0, + "step": 11754 + }, + { + "epoch": 1.4953568248314464, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.193388938903809, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8938323259353638, + "num_tokens": 456431425.0, + "step": 11755 + }, + { + "epoch": 1.495484035110037, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.04629898071289, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8965103626251221, + "num_tokens": 456466147.0, + "step": 11756 + }, + { + "epoch": 1.4956112453886274, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.022248268127441, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9030176401138306, + "num_tokens": 456502294.0, + "step": 11757 + }, + { + "epoch": 1.495738455667218, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.048964500427246, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.9007542133331299, + "num_tokens": 456537243.0, + "step": 11758 + }, + { + "epoch": 1.4958656659458085, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.025224685668945, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8946830034255981, + "num_tokens": 456577910.0, + "step": 11759 + }, + { + "epoch": 1.495992876224399, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.084721565246582, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.883275032043457, + "num_tokens": 456620807.0, + "step": 11760 + }, + { + "epoch": 1.4961200865029896, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.998085021972656, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8872464895248413, + "num_tokens": 456662438.0, + "step": 11761 + }, + { + "epoch": 1.49624729678158, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.049745559692383, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8775159120559692, + "num_tokens": 456704981.0, + "step": 11762 + }, + { + "epoch": 1.4963745070601704, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.125799179077148, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8850836753845215, + "num_tokens": 456741630.0, + "step": 11763 + }, + { + "epoch": 1.496501717338761, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.036364555358887, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8796811103820801, + "num_tokens": 456783591.0, + "step": 11764 + }, + { + "epoch": 1.4966289276173514, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.074466705322266, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8838275671005249, + "num_tokens": 456822142.0, + "step": 11765 + }, + { + "epoch": 1.496756137895942, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.030980110168457, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8722164630889893, + "num_tokens": 456869927.0, + "step": 11766 + }, + { + "epoch": 1.4968833481745325, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.997117042541504, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9025864601135254, + "num_tokens": 456903014.0, + "step": 11767 + }, + { + "epoch": 1.497010558453123, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.077156066894531, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8823114037513733, + "num_tokens": 456945180.0, + "step": 11768 + }, + { + "epoch": 1.4971377687317136, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.01410961151123, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8962973356246948, + "num_tokens": 456979452.0, + "step": 11769 + }, + { + "epoch": 1.497264979010304, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.140313148498535, + "learning_rate": 1e-06, + "loss": 0.3849, + "mean_token_accuracy": 0.8753758668899536, + "num_tokens": 457018833.0, + "step": 11770 + }, + { + "epoch": 1.4973921892888946, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.132552146911621, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8848382234573364, + "num_tokens": 457055700.0, + "step": 11771 + }, + { + "epoch": 1.497519399567485, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.098493576049805, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8781283497810364, + "num_tokens": 457091766.0, + "step": 11772 + }, + { + "epoch": 1.4976466098460754, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.156627655029297, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.8947091102600098, + "num_tokens": 457126223.0, + "step": 11773 + }, + { + "epoch": 1.497773820124666, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.019944190979004, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8988417387008667, + "num_tokens": 457166273.0, + "step": 11774 + }, + { + "epoch": 1.4979010304032565, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.054546356201172, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8841555118560791, + "num_tokens": 457202054.0, + "step": 11775 + }, + { + "epoch": 1.498028240681847, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.022612571716309, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8872395157814026, + "num_tokens": 457243313.0, + "step": 11776 + }, + { + "epoch": 1.4981554509604376, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.08186149597168, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8845661878585815, + "num_tokens": 457282306.0, + "step": 11777 + }, + { + "epoch": 1.498282661239028, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.09359073638916, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8937766551971436, + "num_tokens": 457318020.0, + "step": 11778 + }, + { + "epoch": 1.4984098715176186, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.032331466674805, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9063866138458252, + "num_tokens": 457354037.0, + "step": 11779 + }, + { + "epoch": 1.4985370817962091, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.0845365524292, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8815519213676453, + "num_tokens": 457392363.0, + "step": 11780 + }, + { + "epoch": 1.4986642920747997, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.962136268615723, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.9000321626663208, + "num_tokens": 457432088.0, + "step": 11781 + }, + { + "epoch": 1.4987915023533902, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.170584678649902, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.882082462310791, + "num_tokens": 457466621.0, + "step": 11782 + }, + { + "epoch": 1.4989187126319807, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.072489738464355, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8872129321098328, + "num_tokens": 457504728.0, + "step": 11783 + }, + { + "epoch": 1.4990459229105713, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.102924346923828, + "learning_rate": 1e-06, + "loss": 0.389, + "mean_token_accuracy": 0.8670291900634766, + "num_tokens": 457545650.0, + "step": 11784 + }, + { + "epoch": 1.4991731331891618, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.059033393859863, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8737624883651733, + "num_tokens": 457587450.0, + "step": 11785 + }, + { + "epoch": 1.4993003434677523, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.987231254577637, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8829526901245117, + "num_tokens": 457628901.0, + "step": 11786 + }, + { + "epoch": 1.4994275537463426, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.023731231689453, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8819791078567505, + "num_tokens": 457670660.0, + "step": 11787 + }, + { + "epoch": 1.4995547640249332, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.948984146118164, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8894546031951904, + "num_tokens": 457714188.0, + "step": 11788 + }, + { + "epoch": 1.4996819743035237, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.97690200805664, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8891609311103821, + "num_tokens": 457756595.0, + "step": 11789 + }, + { + "epoch": 1.4998091845821142, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.115246772766113, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8909017443656921, + "num_tokens": 457790613.0, + "step": 11790 + }, + { + "epoch": 1.4999363948607047, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.101966857910156, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.8793902397155762, + "num_tokens": 457832543.0, + "step": 11791 + }, + { + "epoch": 1.5000636051392953, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.977452278137207, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8894668817520142, + "num_tokens": 457875437.0, + "step": 11792 + }, + { + "epoch": 1.5001908154178858, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.063004493713379, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8861571550369263, + "num_tokens": 457917013.0, + "step": 11793 + }, + { + "epoch": 1.5003180256964763, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.04586124420166, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8865571022033691, + "num_tokens": 457957773.0, + "step": 11794 + }, + { + "epoch": 1.5004452359750666, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.102276802062988, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8722845315933228, + "num_tokens": 458000909.0, + "step": 11795 + }, + { + "epoch": 1.5005724462536572, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.507469177246094e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.97658634185791, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8999620676040649, + "num_tokens": 458039326.0, + "step": 11796 + }, + { + "epoch": 1.5006996565322477, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.116700172424316, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8939974904060364, + "num_tokens": 458074788.0, + "step": 11797 + }, + { + "epoch": 1.5008268668108382, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.555152893066406e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.185232162475586, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8765822649002075, + "num_tokens": 458109005.0, + "step": 11798 + }, + { + "epoch": 1.5009540770894287, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.033297538757324, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8928152918815613, + "num_tokens": 458146227.0, + "step": 11799 + }, + { + "epoch": 1.5010812873680193, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.194531440734863, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8796869516372681, + "num_tokens": 458180354.0, + "step": 11800 + }, + { + "epoch": 1.5012084976466098, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.079684257507324, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8878315687179565, + "num_tokens": 458221948.0, + "step": 11801 + }, + { + "epoch": 1.5013357079252003, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.059549331665039, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8821294903755188, + "num_tokens": 458261805.0, + "step": 11802 + }, + { + "epoch": 1.5014629182037909, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.087836265563965, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8878710865974426, + "num_tokens": 458306704.0, + "step": 11803 + }, + { + "epoch": 1.5015901284823814, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.097638130187988, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8697856068611145, + "num_tokens": 458350451.0, + "step": 11804 + }, + { + "epoch": 1.501717338760972, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.051819801330566, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8901562094688416, + "num_tokens": 458390337.0, + "step": 11805 + }, + { + "epoch": 1.5018445490395624, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.114226341247559, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8745926022529602, + "num_tokens": 458430471.0, + "step": 11806 + }, + { + "epoch": 1.501971759318153, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.95947551727295, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8958848714828491, + "num_tokens": 458473808.0, + "step": 11807 + }, + { + "epoch": 1.5020989695967435, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.160325050354004, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8931377530097961, + "num_tokens": 458503793.0, + "step": 11808 + }, + { + "epoch": 1.502226179875334, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.074788093566895, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8900851011276245, + "num_tokens": 458545442.0, + "step": 11809 + }, + { + "epoch": 1.5023533901539246, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.07535171508789, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8889853954315186, + "num_tokens": 458578113.0, + "step": 11810 + }, + { + "epoch": 1.502480600432515, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.181729316711426, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8806238770484924, + "num_tokens": 458616750.0, + "step": 11811 + }, + { + "epoch": 1.5026078107111056, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.168562889099121, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8754757046699524, + "num_tokens": 458650534.0, + "step": 11812 + }, + { + "epoch": 1.502735020989696, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.194314956665039, + "learning_rate": 1e-06, + "loss": 0.3671, + "mean_token_accuracy": 0.8774245977401733, + "num_tokens": 458681275.0, + "step": 11813 + }, + { + "epoch": 1.5028622312682864, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.02065372467041, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8879682421684265, + "num_tokens": 458722590.0, + "step": 11814 + }, + { + "epoch": 1.502989441546877, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.044195175170898, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8917660713195801, + "num_tokens": 458758309.0, + "step": 11815 + }, + { + "epoch": 1.5031166518254675, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.03347110748291, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8917946815490723, + "num_tokens": 458801433.0, + "step": 11816 + }, + { + "epoch": 1.503243862104058, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.077495574951172, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8865941762924194, + "num_tokens": 458840623.0, + "step": 11817 + }, + { + "epoch": 1.5033710723826486, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.957921981811523, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.9033690094947815, + "num_tokens": 458885077.0, + "step": 11818 + }, + { + "epoch": 1.503498282661239, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.19560718536377, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8940215110778809, + "num_tokens": 458914326.0, + "step": 11819 + }, + { + "epoch": 1.5036254929398294, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.08754825592041, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.8766803741455078, + "num_tokens": 458955312.0, + "step": 11820 + }, + { + "epoch": 1.50375270321842, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.150769233703613, + "learning_rate": 1e-06, + "loss": 0.3898, + "mean_token_accuracy": 0.8708124160766602, + "num_tokens": 458990965.0, + "step": 11821 + }, + { + "epoch": 1.5038799134970104, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.021303176879883, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8827293515205383, + "num_tokens": 459033187.0, + "step": 11822 + }, + { + "epoch": 1.504007123775601, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.034199714660645, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8977444171905518, + "num_tokens": 459072539.0, + "step": 11823 + }, + { + "epoch": 1.5041343340541915, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.156506538391113, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8848798274993896, + "num_tokens": 459109582.0, + "step": 11824 + }, + { + "epoch": 1.504261544332782, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.083940505981445, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8840961456298828, + "num_tokens": 459151009.0, + "step": 11825 + }, + { + "epoch": 1.5043887546113726, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.04988956451416, + "learning_rate": 1e-06, + "loss": 0.3769, + "mean_token_accuracy": 0.8757467865943909, + "num_tokens": 459190569.0, + "step": 11826 + }, + { + "epoch": 1.504515964889963, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.06279468536377, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8846310973167419, + "num_tokens": 459230817.0, + "step": 11827 + }, + { + "epoch": 1.5046431751685536, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.92248821258545, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.9044666290283203, + "num_tokens": 459271911.0, + "step": 11828 + }, + { + "epoch": 1.5047703854471441, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.12199878692627, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8851655721664429, + "num_tokens": 459315824.0, + "step": 11829 + }, + { + "epoch": 1.5048975957257347, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.045791625976562, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8958624601364136, + "num_tokens": 459352223.0, + "step": 11830 + }, + { + "epoch": 1.5050248060043252, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.142888069152832, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8837834596633911, + "num_tokens": 459384758.0, + "step": 11831 + }, + { + "epoch": 1.5051520162829157, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.060711860656738, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8969359397888184, + "num_tokens": 459420782.0, + "step": 11832 + }, + { + "epoch": 1.5052792265615063, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.978055000305176, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.9053522348403931, + "num_tokens": 459461461.0, + "step": 11833 + }, + { + "epoch": 1.5054064368400968, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.149898529052734, + "learning_rate": 1e-06, + "loss": 0.3927, + "mean_token_accuracy": 0.8699319362640381, + "num_tokens": 459500632.0, + "step": 11834 + }, + { + "epoch": 1.5055336471186873, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.095592498779297, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.8756458759307861, + "num_tokens": 459538636.0, + "step": 11835 + }, + { + "epoch": 1.5056608573972778, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.033376693725586, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8871963620185852, + "num_tokens": 459578221.0, + "step": 11836 + }, + { + "epoch": 1.5057880676758684, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.11092758178711, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8866652250289917, + "num_tokens": 459615081.0, + "step": 11837 + }, + { + "epoch": 1.5059152779544587, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.127921104431152, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8861958980560303, + "num_tokens": 459649243.0, + "step": 11838 + }, + { + "epoch": 1.5060424882330492, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.003259658813477, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8934067487716675, + "num_tokens": 459688719.0, + "step": 11839 + }, + { + "epoch": 1.5061696985116397, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.04223918914795, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.874198317527771, + "num_tokens": 459732973.0, + "step": 11840 + }, + { + "epoch": 1.5062969087902303, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.065495491027832, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8768441677093506, + "num_tokens": 459772465.0, + "step": 11841 + }, + { + "epoch": 1.5064241190688208, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.049214363098145, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8855383396148682, + "num_tokens": 459814031.0, + "step": 11842 + }, + { + "epoch": 1.5065513293474113, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.092086791992188, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.877748966217041, + "num_tokens": 459852980.0, + "step": 11843 + }, + { + "epoch": 1.5066785396260016, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.060744285583496, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8974183797836304, + "num_tokens": 459893384.0, + "step": 11844 + }, + { + "epoch": 1.5068057499045922, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.013679504394531, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.886178195476532, + "num_tokens": 459933618.0, + "step": 11845 + }, + { + "epoch": 1.5069329601831827, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.075292587280273, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8888673782348633, + "num_tokens": 459976359.0, + "step": 11846 + }, + { + "epoch": 1.5070601704617732, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.10601806640625, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8935251235961914, + "num_tokens": 460016485.0, + "step": 11847 + }, + { + "epoch": 1.5071873807403637, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.139504432678223, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8845046162605286, + "num_tokens": 460056429.0, + "step": 11848 + }, + { + "epoch": 1.5073145910189543, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.10490608215332, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.875616729259491, + "num_tokens": 460098971.0, + "step": 11849 + }, + { + "epoch": 1.5074418012975448, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 8.987532615661621, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8892694115638733, + "num_tokens": 460138975.0, + "step": 11850 + }, + { + "epoch": 1.5075690115761353, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.085929870605469, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8770459890365601, + "num_tokens": 460181236.0, + "step": 11851 + }, + { + "epoch": 1.5076962218547258, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.020989418029785, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.893031656742096, + "num_tokens": 460218222.0, + "step": 11852 + }, + { + "epoch": 1.5078234321333164, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.038835525512695, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8966111540794373, + "num_tokens": 460254498.0, + "step": 11853 + }, + { + "epoch": 1.507950642411907, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.989459991455078, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9014895558357239, + "num_tokens": 460291316.0, + "step": 11854 + }, + { + "epoch": 1.5080778526904974, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.022656440734863, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8846757411956787, + "num_tokens": 460338451.0, + "step": 11855 + }, + { + "epoch": 1.508205062969088, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.136781692504883, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8977192640304565, + "num_tokens": 460375772.0, + "step": 11856 + }, + { + "epoch": 1.5083322732476785, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.073284149169922, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8930679559707642, + "num_tokens": 460413757.0, + "step": 11857 + }, + { + "epoch": 1.508459483526269, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.059321403503418, + "learning_rate": 1e-06, + "loss": 0.3475, + "mean_token_accuracy": 0.8799025416374207, + "num_tokens": 460452393.0, + "step": 11858 + }, + { + "epoch": 1.5085866938048595, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.189397811889648, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8901237845420837, + "num_tokens": 460484775.0, + "step": 11859 + }, + { + "epoch": 1.50871390408345, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.017462730407715, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8864954113960266, + "num_tokens": 460523898.0, + "step": 11860 + }, + { + "epoch": 1.5088411143620406, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.063902854919434, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8960359692573547, + "num_tokens": 460557347.0, + "step": 11861 + }, + { + "epoch": 1.508968324640631, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.103303909301758, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8931757211685181, + "num_tokens": 460593895.0, + "step": 11862 + }, + { + "epoch": 1.5090955349192214, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.137930870056152, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8892381191253662, + "num_tokens": 460630205.0, + "step": 11863 + }, + { + "epoch": 1.509222745197812, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.057751655578613, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8741714954376221, + "num_tokens": 460669604.0, + "step": 11864 + }, + { + "epoch": 1.5093499554764025, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.96308708190918, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8927181959152222, + "num_tokens": 460709852.0, + "step": 11865 + }, + { + "epoch": 1.509477165754993, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.034561157226562, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8867786526679993, + "num_tokens": 460752226.0, + "step": 11866 + }, + { + "epoch": 1.5096043760335836, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.086727142333984, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8833068609237671, + "num_tokens": 460793846.0, + "step": 11867 + }, + { + "epoch": 1.509731586312174, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.027515411376953, + "learning_rate": 1e-06, + "loss": 0.3867, + "mean_token_accuracy": 0.8744809627532959, + "num_tokens": 460840516.0, + "step": 11868 + }, + { + "epoch": 1.5098587965907644, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.03604507446289, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8767914175987244, + "num_tokens": 460881923.0, + "step": 11869 + }, + { + "epoch": 1.509986006869355, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.010720252990723, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8923708200454712, + "num_tokens": 460921598.0, + "step": 11870 + }, + { + "epoch": 1.5101132171479454, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.114986419677734, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8864660263061523, + "num_tokens": 460959409.0, + "step": 11871 + }, + { + "epoch": 1.510240427426536, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.062458992004395, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8846603631973267, + "num_tokens": 461000302.0, + "step": 11872 + }, + { + "epoch": 1.5103676377051265, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.02800464630127, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8924483060836792, + "num_tokens": 461038107.0, + "step": 11873 + }, + { + "epoch": 1.510494847983717, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.13412857055664, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8697539567947388, + "num_tokens": 461077281.0, + "step": 11874 + }, + { + "epoch": 1.5106220582623076, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.122050285339355, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8947382569313049, + "num_tokens": 461112028.0, + "step": 11875 + }, + { + "epoch": 1.510749268540898, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.988755226135254, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8870804309844971, + "num_tokens": 461154037.0, + "step": 11876 + }, + { + "epoch": 1.5108764788194886, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.059212684631348, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8944329619407654, + "num_tokens": 461193964.0, + "step": 11877 + }, + { + "epoch": 1.5110036890980791, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.040875434875488, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8882059454917908, + "num_tokens": 461230715.0, + "step": 11878 + }, + { + "epoch": 1.5111308993766697, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.022294044494629, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.898453414440155, + "num_tokens": 461273784.0, + "step": 11879 + }, + { + "epoch": 1.5112581096552602, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.148716926574707, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.877743124961853, + "num_tokens": 461310163.0, + "step": 11880 + }, + { + "epoch": 1.5113853199338507, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.06064224243164, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8989492058753967, + "num_tokens": 461344166.0, + "step": 11881 + }, + { + "epoch": 1.5115125302124413, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.085363388061523, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8840199708938599, + "num_tokens": 461385268.0, + "step": 11882 + }, + { + "epoch": 1.5116397404910318, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.023460388183594, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8928868770599365, + "num_tokens": 461422495.0, + "step": 11883 + }, + { + "epoch": 1.5117669507696223, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.067816734313965, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.877842903137207, + "num_tokens": 461462573.0, + "step": 11884 + }, + { + "epoch": 1.5118941610482128, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.084108352661133, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.877884566783905, + "num_tokens": 461503938.0, + "step": 11885 + }, + { + "epoch": 1.5120213713268034, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.075142860412598, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8901214599609375, + "num_tokens": 461541525.0, + "step": 11886 + }, + { + "epoch": 1.5121485816053937, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.162750244140625, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8994489312171936, + "num_tokens": 461573531.0, + "step": 11887 + }, + { + "epoch": 1.5122757918839842, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.015602111816406, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.888218879699707, + "num_tokens": 461618828.0, + "step": 11888 + }, + { + "epoch": 1.5124030021625747, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.034266471862793, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.903628945350647, + "num_tokens": 461656634.0, + "step": 11889 + }, + { + "epoch": 1.5125302124411653, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.119037628173828, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8870710134506226, + "num_tokens": 461696300.0, + "step": 11890 + }, + { + "epoch": 1.5126574227197558, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.0852632522583, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8901098966598511, + "num_tokens": 461732553.0, + "step": 11891 + }, + { + "epoch": 1.5127846329983463, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.08970832824707, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8864263892173767, + "num_tokens": 461773111.0, + "step": 11892 + }, + { + "epoch": 1.5129118432769366, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.148001670837402, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8929147720336914, + "num_tokens": 461805852.0, + "step": 11893 + }, + { + "epoch": 1.5130390535555271, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.07591438293457, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8956769704818726, + "num_tokens": 461843743.0, + "step": 11894 + }, + { + "epoch": 1.5131662638341177, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.04815673828125, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8952977657318115, + "num_tokens": 461877326.0, + "step": 11895 + }, + { + "epoch": 1.5132934741127082, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.984540939331055, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.9008979201316833, + "num_tokens": 461916535.0, + "step": 11896 + }, + { + "epoch": 1.5134206843912987, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.038015365600586, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8957911729812622, + "num_tokens": 461952836.0, + "step": 11897 + }, + { + "epoch": 1.5135478946698893, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.995634078979492, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8921695351600647, + "num_tokens": 461995760.0, + "step": 11898 + }, + { + "epoch": 1.5136751049484798, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.983250617980957, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8888950943946838, + "num_tokens": 462042470.0, + "step": 11899 + }, + { + "epoch": 1.5138023152270703, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.063870429992676, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8797870874404907, + "num_tokens": 462085899.0, + "step": 11900 + }, + { + "epoch": 1.5139295255056608, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.052833557128906, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.8766077160835266, + "num_tokens": 462131083.0, + "step": 11901 + }, + { + "epoch": 1.5140567357842514, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.017265319824219, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.9003428220748901, + "num_tokens": 462171530.0, + "step": 11902 + }, + { + "epoch": 1.514183946062842, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.075885772705078, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8851292729377747, + "num_tokens": 462209526.0, + "step": 11903 + }, + { + "epoch": 1.5143111563414324, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.094359397888184, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8761397004127502, + "num_tokens": 462246998.0, + "step": 11904 + }, + { + "epoch": 1.514438366620023, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.035691261291504, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.895577609539032, + "num_tokens": 462285182.0, + "step": 11905 + }, + { + "epoch": 1.5145655768986135, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.602836608886719e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.150824546813965, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8823915123939514, + "num_tokens": 462319306.0, + "step": 11906 + }, + { + "epoch": 1.514692787177204, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.057583808898926, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8926706314086914, + "num_tokens": 462357006.0, + "step": 11907 + }, + { + "epoch": 1.5148199974557945, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.092326164245605, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.877191424369812, + "num_tokens": 462395020.0, + "step": 11908 + }, + { + "epoch": 1.514947207734385, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.098609924316406, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8769184350967407, + "num_tokens": 462432652.0, + "step": 11909 + }, + { + "epoch": 1.5150744180129756, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.065049171447754, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8893604874610901, + "num_tokens": 462468257.0, + "step": 11910 + }, + { + "epoch": 1.515201628291566, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.0868501663208, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8861709237098694, + "num_tokens": 462504580.0, + "step": 11911 + }, + { + "epoch": 1.5153288385701564, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.04509449005127, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8770318031311035, + "num_tokens": 462549539.0, + "step": 11912 + }, + { + "epoch": 1.515456048848747, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.998449325561523, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.895708441734314, + "num_tokens": 462588781.0, + "step": 11913 + }, + { + "epoch": 1.5155832591273375, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.11205005645752, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8972097635269165, + "num_tokens": 462623298.0, + "step": 11914 + }, + { + "epoch": 1.515710469405928, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.177651405334473, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8825082778930664, + "num_tokens": 462655479.0, + "step": 11915 + }, + { + "epoch": 1.5158376796845185, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.00321102142334, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8828800916671753, + "num_tokens": 462698350.0, + "step": 11916 + }, + { + "epoch": 1.515964889963109, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.03803825378418, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8820970058441162, + "num_tokens": 462744438.0, + "step": 11917 + }, + { + "epoch": 1.5160921002416994, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.999866485595703, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8952512741088867, + "num_tokens": 462784394.0, + "step": 11918 + }, + { + "epoch": 1.51621931052029, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.960871696472168, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8875184059143066, + "num_tokens": 462825439.0, + "step": 11919 + }, + { + "epoch": 1.5163465207988804, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.135887145996094, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8809230327606201, + "num_tokens": 462861825.0, + "step": 11920 + }, + { + "epoch": 1.516473731077471, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.98432445526123, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.9031431674957275, + "num_tokens": 462901529.0, + "step": 11921 + }, + { + "epoch": 1.5166009413560615, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.098087310791016, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8814622163772583, + "num_tokens": 462940058.0, + "step": 11922 + }, + { + "epoch": 1.516728151634652, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.077946662902832, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8946584463119507, + "num_tokens": 462975762.0, + "step": 11923 + }, + { + "epoch": 1.5168553619132426, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.989506721496582, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8919826745986938, + "num_tokens": 463016382.0, + "step": 11924 + }, + { + "epoch": 1.516982572191833, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.05509090423584, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8896889686584473, + "num_tokens": 463055189.0, + "step": 11925 + }, + { + "epoch": 1.5171097824704236, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.015029907226562, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8886682987213135, + "num_tokens": 463098141.0, + "step": 11926 + }, + { + "epoch": 1.5172369927490141, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.057039260864258, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8949445486068726, + "num_tokens": 463134785.0, + "step": 11927 + }, + { + "epoch": 1.5173642030276047, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.034235954284668, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8914996385574341, + "num_tokens": 463176324.0, + "step": 11928 + }, + { + "epoch": 1.5174914133061952, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.146484375, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8829797506332397, + "num_tokens": 463212877.0, + "step": 11929 + }, + { + "epoch": 1.5176186235847857, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.99932861328125, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.9019520878791809, + "num_tokens": 463254102.0, + "step": 11930 + }, + { + "epoch": 1.5177458338633762, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.02397632598877, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.886680006980896, + "num_tokens": 463292236.0, + "step": 11931 + }, + { + "epoch": 1.5178730441419668, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.029122352600098, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.890697717666626, + "num_tokens": 463330335.0, + "step": 11932 + }, + { + "epoch": 1.5180002544205573, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.102919578552246, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8851631879806519, + "num_tokens": 463365735.0, + "step": 11933 + }, + { + "epoch": 1.5181274646991478, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.038565635681152, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8863116502761841, + "num_tokens": 463404723.0, + "step": 11934 + }, + { + "epoch": 1.5182546749777384, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.985962867736816, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.89894700050354, + "num_tokens": 463447029.0, + "step": 11935 + }, + { + "epoch": 1.5183818852563287, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.122528076171875, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.877230167388916, + "num_tokens": 463488355.0, + "step": 11936 + }, + { + "epoch": 1.5185090955349192, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.167037963867188, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8814107179641724, + "num_tokens": 463524066.0, + "step": 11937 + }, + { + "epoch": 1.5186363058135097, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.050704956054688, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8815045952796936, + "num_tokens": 463573087.0, + "step": 11938 + }, + { + "epoch": 1.5187635160921003, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.180307388305664, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8867602348327637, + "num_tokens": 463608540.0, + "step": 11939 + }, + { + "epoch": 1.5188907263706908, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.13365650177002, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8833414316177368, + "num_tokens": 463643015.0, + "step": 11940 + }, + { + "epoch": 1.5190179366492813, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.941279411315918, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8905405402183533, + "num_tokens": 463689740.0, + "step": 11941 + }, + { + "epoch": 1.5191451469278716, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.086931228637695, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8733758330345154, + "num_tokens": 463730928.0, + "step": 11942 + }, + { + "epoch": 1.5192723572064621, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.024130821228027, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8889782428741455, + "num_tokens": 463770201.0, + "step": 11943 + }, + { + "epoch": 1.5193995674850527, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.047505378723145, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.8662388324737549, + "num_tokens": 463811910.0, + "step": 11944 + }, + { + "epoch": 1.5195267777636432, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.018449783325195, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8845166563987732, + "num_tokens": 463858017.0, + "step": 11945 + }, + { + "epoch": 1.5196539880422337, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.052618980407715, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.890344500541687, + "num_tokens": 463897709.0, + "step": 11946 + }, + { + "epoch": 1.5197811983208243, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.111641883850098, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8873709440231323, + "num_tokens": 463931496.0, + "step": 11947 + }, + { + "epoch": 1.5199084085994148, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.055757522583008, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8830834627151489, + "num_tokens": 463971650.0, + "step": 11948 + }, + { + "epoch": 1.5200356188780053, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.006087303161621, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8951735496520996, + "num_tokens": 464012665.0, + "step": 11949 + }, + { + "epoch": 1.5201628291565958, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.113825798034668, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.879473090171814, + "num_tokens": 464055382.0, + "step": 11950 + }, + { + "epoch": 1.5202900394351864, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.114845275878906, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.89262455701828, + "num_tokens": 464090876.0, + "step": 11951 + }, + { + "epoch": 1.520417249713777, + "ewc_loss": 0.015869140625, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.005910873413086, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.9022279977798462, + "num_tokens": 464128205.0, + "step": 11952 + }, + { + "epoch": 1.5205444599923674, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.004876136779785, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8928966522216797, + "num_tokens": 464171741.0, + "step": 11953 + }, + { + "epoch": 1.520671670270958, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.172370910644531, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8721047043800354, + "num_tokens": 464203008.0, + "step": 11954 + }, + { + "epoch": 1.5207988805495485, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.188619613647461, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8786774277687073, + "num_tokens": 464242666.0, + "step": 11955 + }, + { + "epoch": 1.520926090828139, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.982684135437012, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8837095499038696, + "num_tokens": 464283706.0, + "step": 11956 + }, + { + "epoch": 1.5210533011067295, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.13944149017334, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8867000341415405, + "num_tokens": 464320525.0, + "step": 11957 + }, + { + "epoch": 1.52118051138532, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.127266883850098, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8883079290390015, + "num_tokens": 464352842.0, + "step": 11958 + }, + { + "epoch": 1.5213077216639106, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.035199165344238, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.888272762298584, + "num_tokens": 464390410.0, + "step": 11959 + }, + { + "epoch": 1.521434931942501, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.147270202636719, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.886228621006012, + "num_tokens": 464423527.0, + "step": 11960 + }, + { + "epoch": 1.5215621422210914, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.114959716796875, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8919159770011902, + "num_tokens": 464455538.0, + "step": 11961 + }, + { + "epoch": 1.521689352499682, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.117854118347168, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8726602792739868, + "num_tokens": 464492628.0, + "step": 11962 + }, + { + "epoch": 1.5218165627782725, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.023893356323242, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8967978358268738, + "num_tokens": 464533495.0, + "step": 11963 + }, + { + "epoch": 1.521943773056863, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.013915061950684, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.891288161277771, + "num_tokens": 464578737.0, + "step": 11964 + }, + { + "epoch": 1.5220709833354535, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.136621475219727, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8957424163818359, + "num_tokens": 464611868.0, + "step": 11965 + }, + { + "epoch": 1.5221981936140438, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.003609657287598, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8803212642669678, + "num_tokens": 464656462.0, + "step": 11966 + }, + { + "epoch": 1.5223254038926344, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.060842514038086, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8974034786224365, + "num_tokens": 464691049.0, + "step": 11967 + }, + { + "epoch": 1.522452614171225, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.134922981262207, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8976798057556152, + "num_tokens": 464723045.0, + "step": 11968 + }, + { + "epoch": 1.5225798244498154, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.051170349121094, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8789383172988892, + "num_tokens": 464766207.0, + "step": 11969 + }, + { + "epoch": 1.522707034728406, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.062520980834961, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8847947120666504, + "num_tokens": 464804557.0, + "step": 11970 + }, + { + "epoch": 1.5228342450069965, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.100021362304688, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8891443610191345, + "num_tokens": 464841596.0, + "step": 11971 + }, + { + "epoch": 1.522961455285587, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 8.98083209991455, + "learning_rate": 1e-06, + "loss": 0.3804, + "mean_token_accuracy": 0.8690989017486572, + "num_tokens": 464887357.0, + "step": 11972 + }, + { + "epoch": 1.5230886655641775, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.005447387695312, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8927339315414429, + "num_tokens": 464925239.0, + "step": 11973 + }, + { + "epoch": 1.523215875842768, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.034626960754395, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8849005699157715, + "num_tokens": 464963369.0, + "step": 11974 + }, + { + "epoch": 1.5233430861213586, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.154860496520996, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8876067399978638, + "num_tokens": 464999815.0, + "step": 11975 + }, + { + "epoch": 1.5234702963999491, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.044137001037598, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9035208225250244, + "num_tokens": 465034137.0, + "step": 11976 + }, + { + "epoch": 1.5235975066785397, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.015572547912598, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8818864822387695, + "num_tokens": 465081754.0, + "step": 11977 + }, + { + "epoch": 1.5237247169571302, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.05744743347168, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8834768533706665, + "num_tokens": 465120779.0, + "step": 11978 + }, + { + "epoch": 1.5238519272357207, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.067622184753418, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8922191858291626, + "num_tokens": 465162861.0, + "step": 11979 + }, + { + "epoch": 1.5239791375143112, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.031237602233887, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.9009098410606384, + "num_tokens": 465203838.0, + "step": 11980 + }, + { + "epoch": 1.5241063477929018, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.048279762268066, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8934478759765625, + "num_tokens": 465241552.0, + "step": 11981 + }, + { + "epoch": 1.5242335580714923, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.090929985046387, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8861710429191589, + "num_tokens": 465285110.0, + "step": 11982 + }, + { + "epoch": 1.5243607683500828, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.084332466125488, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.887700617313385, + "num_tokens": 465322878.0, + "step": 11983 + }, + { + "epoch": 1.5244879786286734, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.110877990722656, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9003216028213501, + "num_tokens": 465358626.0, + "step": 11984 + }, + { + "epoch": 1.5246151889072637, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.159374237060547, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.898343563079834, + "num_tokens": 465399563.0, + "step": 11985 + }, + { + "epoch": 1.5247423991858542, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.038204193115234, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8759477138519287, + "num_tokens": 465446700.0, + "step": 11986 + }, + { + "epoch": 1.5248696094644447, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.043035507202148, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8919574618339539, + "num_tokens": 465482911.0, + "step": 11987 + }, + { + "epoch": 1.5249968197430352, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.650520324707031e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.15627670288086, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8867418766021729, + "num_tokens": 465516646.0, + "step": 11988 + }, + { + "epoch": 1.5251240300216258, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.120861053466797, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8821736574172974, + "num_tokens": 465556529.0, + "step": 11989 + }, + { + "epoch": 1.5252512403002163, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.965733528137207, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8992154598236084, + "num_tokens": 465597860.0, + "step": 11990 + }, + { + "epoch": 1.5253784505788066, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.076594352722168, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8885573148727417, + "num_tokens": 465636928.0, + "step": 11991 + }, + { + "epoch": 1.5255056608573971, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.139344215393066, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.889028787612915, + "num_tokens": 465669986.0, + "step": 11992 + }, + { + "epoch": 1.5256328711359877, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.097872734069824, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.904781699180603, + "num_tokens": 465701331.0, + "step": 11993 + }, + { + "epoch": 1.5257600814145782, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.223182678222656, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8900579214096069, + "num_tokens": 465739788.0, + "step": 11994 + }, + { + "epoch": 1.5258872916931687, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.086705207824707, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8872723579406738, + "num_tokens": 465771938.0, + "step": 11995 + }, + { + "epoch": 1.5260145019717593, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.050676345825195, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8970240354537964, + "num_tokens": 465810604.0, + "step": 11996 + }, + { + "epoch": 1.5261417122503498, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.085946083068848, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.894105851650238, + "num_tokens": 465846321.0, + "step": 11997 + }, + { + "epoch": 1.5262689225289403, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.04410171508789, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8945252299308777, + "num_tokens": 465887590.0, + "step": 11998 + }, + { + "epoch": 1.5263961328075308, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.111250877380371, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8939856886863708, + "num_tokens": 465926801.0, + "step": 11999 + }, + { + "epoch": 1.5265233430861214, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.108912467956543, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8920590877532959, + "num_tokens": 465963082.0, + "step": 12000 + }, + { + "epoch": 1.526650553364712, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.045482635498047, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8926844596862793, + "num_tokens": 466000704.0, + "step": 12001 + }, + { + "epoch": 1.5267777636433024, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.093134880065918, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8907839059829712, + "num_tokens": 466039012.0, + "step": 12002 + }, + { + "epoch": 1.526904973921893, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.081541061401367, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8873902559280396, + "num_tokens": 466080109.0, + "step": 12003 + }, + { + "epoch": 1.5270321842004835, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.093064308166504, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8861313462257385, + "num_tokens": 466121150.0, + "step": 12004 + }, + { + "epoch": 1.527159394479074, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.072595596313477, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.887864351272583, + "num_tokens": 466166022.0, + "step": 12005 + }, + { + "epoch": 1.5272866047576645, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.17268180847168, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8821394443511963, + "num_tokens": 466205467.0, + "step": 12006 + }, + { + "epoch": 1.527413815036255, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.106623649597168, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.883536159992218, + "num_tokens": 466240348.0, + "step": 12007 + }, + { + "epoch": 1.5275410253148456, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.086383819580078, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.888412356376648, + "num_tokens": 466280458.0, + "step": 12008 + }, + { + "epoch": 1.527668235593436, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.117402076721191, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8882203102111816, + "num_tokens": 466318654.0, + "step": 12009 + }, + { + "epoch": 1.5277954458720264, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.098870277404785, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8836725354194641, + "num_tokens": 466359903.0, + "step": 12010 + }, + { + "epoch": 1.527922656150617, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.061561584472656, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.893612802028656, + "num_tokens": 466396527.0, + "step": 12011 + }, + { + "epoch": 1.5280498664292075, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.033888816833496, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8843057155609131, + "num_tokens": 466437734.0, + "step": 12012 + }, + { + "epoch": 1.528177076707798, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.024295806884766, + "learning_rate": 1e-06, + "loss": 0.3631, + "mean_token_accuracy": 0.8779824376106262, + "num_tokens": 466484069.0, + "step": 12013 + }, + { + "epoch": 1.5283042869863885, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.12826919555664, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8925763368606567, + "num_tokens": 466518773.0, + "step": 12014 + }, + { + "epoch": 1.5284314972649788, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.179474830627441, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8900653123855591, + "num_tokens": 466548700.0, + "step": 12015 + }, + { + "epoch": 1.5285587075435694, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.05932331085205, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8857593536376953, + "num_tokens": 466589796.0, + "step": 12016 + }, + { + "epoch": 1.52868591782216, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.079737663269043, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8864097595214844, + "num_tokens": 466629637.0, + "step": 12017 + }, + { + "epoch": 1.5288131281007504, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.130980491638184, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.894891083240509, + "num_tokens": 466668307.0, + "step": 12018 + }, + { + "epoch": 1.528940338379341, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.994897842407227, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.886300265789032, + "num_tokens": 466712663.0, + "step": 12019 + }, + { + "epoch": 1.5290675486579315, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.08560848236084, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8950282335281372, + "num_tokens": 466755794.0, + "step": 12020 + }, + { + "epoch": 1.529194758936522, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.038360595703125, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8949416279792786, + "num_tokens": 466791707.0, + "step": 12021 + }, + { + "epoch": 1.5293219692151125, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.15233039855957, + "learning_rate": 1e-06, + "loss": 0.4039, + "mean_token_accuracy": 0.8633425235748291, + "num_tokens": 466831834.0, + "step": 12022 + }, + { + "epoch": 1.529449179493703, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.12869644165039, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8886704444885254, + "num_tokens": 466865887.0, + "step": 12023 + }, + { + "epoch": 1.5295763897722936, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.083396911621094, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8915776610374451, + "num_tokens": 466902420.0, + "step": 12024 + }, + { + "epoch": 1.5297036000508841, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.039076805114746, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8893520832061768, + "num_tokens": 466942842.0, + "step": 12025 + }, + { + "epoch": 1.5298308103294747, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.051493644714355, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8860733509063721, + "num_tokens": 466981228.0, + "step": 12026 + }, + { + "epoch": 1.5299580206080652, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.157488822937012, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8866656422615051, + "num_tokens": 467015368.0, + "step": 12027 + }, + { + "epoch": 1.5300852308866557, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.109437942504883, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8785396814346313, + "num_tokens": 467049496.0, + "step": 12028 + }, + { + "epoch": 1.5302124411652462, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.063918113708496, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.9036486744880676, + "num_tokens": 467086486.0, + "step": 12029 + }, + { + "epoch": 1.5303396514438368, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.968426704406738, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8944206833839417, + "num_tokens": 467127393.0, + "step": 12030 + }, + { + "epoch": 1.5304668617224273, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.093716621398926, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8913888335227966, + "num_tokens": 467165277.0, + "step": 12031 + }, + { + "epoch": 1.5305940720010178, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.6743621826171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.032268524169922, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8818011283874512, + "num_tokens": 467206244.0, + "step": 12032 + }, + { + "epoch": 1.5307212822796084, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.080613136291504, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.9018643498420715, + "num_tokens": 467244037.0, + "step": 12033 + }, + { + "epoch": 1.5308484925581987, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.178520202636719, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8860235214233398, + "num_tokens": 467278466.0, + "step": 12034 + }, + { + "epoch": 1.5309757028367892, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.04427433013916, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9036767482757568, + "num_tokens": 467315994.0, + "step": 12035 + }, + { + "epoch": 1.5311029131153797, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.11880111694336, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.9002489447593689, + "num_tokens": 467349974.0, + "step": 12036 + }, + { + "epoch": 1.5312301233939702, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.098297119140625, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.890532374382019, + "num_tokens": 467388204.0, + "step": 12037 + }, + { + "epoch": 1.5313573336725608, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.06727123260498, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8857173919677734, + "num_tokens": 467424685.0, + "step": 12038 + }, + { + "epoch": 1.5314845439511513, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.099204063415527, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8885754346847534, + "num_tokens": 467464508.0, + "step": 12039 + }, + { + "epoch": 1.5316117542297416, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.059063911437988, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8817480802536011, + "num_tokens": 467507603.0, + "step": 12040 + }, + { + "epoch": 1.5317389645083321, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.143523216247559, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8966886401176453, + "num_tokens": 467542622.0, + "step": 12041 + }, + { + "epoch": 1.5318661747869227, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.329601287841797, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.874710202217102, + "num_tokens": 467575705.0, + "step": 12042 + }, + { + "epoch": 1.5319933850655132, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.09097671508789, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8878223299980164, + "num_tokens": 467612438.0, + "step": 12043 + }, + { + "epoch": 1.5321205953441037, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.067919731140137, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8963000774383545, + "num_tokens": 467647329.0, + "step": 12044 + }, + { + "epoch": 1.5322478056226942, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.078774452209473, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8945194482803345, + "num_tokens": 467689818.0, + "step": 12045 + }, + { + "epoch": 1.5323750159012848, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.119085311889648, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8775515556335449, + "num_tokens": 467726652.0, + "step": 12046 + }, + { + "epoch": 1.5325022261798753, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.039392471313477, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.889372706413269, + "num_tokens": 467771074.0, + "step": 12047 + }, + { + "epoch": 1.5326294364584658, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.068703651428223, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.877494752407074, + "num_tokens": 467814164.0, + "step": 12048 + }, + { + "epoch": 1.5327566467370564, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.144780158996582, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8974723815917969, + "num_tokens": 467852815.0, + "step": 12049 + }, + { + "epoch": 1.532883857015647, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 8.95511531829834, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8947626352310181, + "num_tokens": 467900257.0, + "step": 12050 + }, + { + "epoch": 1.5330110672942374, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.10888957977295, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8944015502929688, + "num_tokens": 467936412.0, + "step": 12051 + }, + { + "epoch": 1.533138277572828, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.12637996673584, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8795856237411499, + "num_tokens": 467974320.0, + "step": 12052 + }, + { + "epoch": 1.5332654878514185, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.062626838684082, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.9025139212608337, + "num_tokens": 468011066.0, + "step": 12053 + }, + { + "epoch": 1.533392698130009, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.041555404663086, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8953012824058533, + "num_tokens": 468052727.0, + "step": 12054 + }, + { + "epoch": 1.5335199084085995, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.157347679138184, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.892297625541687, + "num_tokens": 468084905.0, + "step": 12055 + }, + { + "epoch": 1.53364711868719, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.08625316619873, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8987222909927368, + "num_tokens": 468118551.0, + "step": 12056 + }, + { + "epoch": 1.5337743289657806, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.076215744018555, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8930794596672058, + "num_tokens": 468156760.0, + "step": 12057 + }, + { + "epoch": 1.533901539244371, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.093600273132324, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8901193737983704, + "num_tokens": 468195280.0, + "step": 12058 + }, + { + "epoch": 1.5340287495229614, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.060812950134277, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8784301280975342, + "num_tokens": 468235535.0, + "step": 12059 + }, + { + "epoch": 1.534155959801552, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.082465171813965, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8773605823516846, + "num_tokens": 468275501.0, + "step": 12060 + }, + { + "epoch": 1.5342831700801425, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.09262466430664, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8835065364837646, + "num_tokens": 468314922.0, + "step": 12061 + }, + { + "epoch": 1.534410380358733, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.002898216247559, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8863939046859741, + "num_tokens": 468359992.0, + "step": 12062 + }, + { + "epoch": 1.5345375906373235, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.146783828735352, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8810967206954956, + "num_tokens": 468400057.0, + "step": 12063 + }, + { + "epoch": 1.5346648009159138, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.095722198486328, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8921645879745483, + "num_tokens": 468433596.0, + "step": 12064 + }, + { + "epoch": 1.5347920111945044, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.103915214538574, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8876979351043701, + "num_tokens": 468468539.0, + "step": 12065 + }, + { + "epoch": 1.534919221473095, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.09257984161377, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8904617428779602, + "num_tokens": 468509517.0, + "step": 12066 + }, + { + "epoch": 1.5350464317516854, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.153928756713867, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8789855241775513, + "num_tokens": 468552342.0, + "step": 12067 + }, + { + "epoch": 1.535173642030276, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.078722953796387, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8920691013336182, + "num_tokens": 468594175.0, + "step": 12068 + }, + { + "epoch": 1.5353008523088665, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.10042953491211, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8936070799827576, + "num_tokens": 468628431.0, + "step": 12069 + }, + { + "epoch": 1.535428062587457, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.110928535461426, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8844480514526367, + "num_tokens": 468668329.0, + "step": 12070 + }, + { + "epoch": 1.5355552728660475, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.035805702209473, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8807158470153809, + "num_tokens": 468708260.0, + "step": 12071 + }, + { + "epoch": 1.535682483144638, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.134382247924805, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8937958478927612, + "num_tokens": 468747989.0, + "step": 12072 + }, + { + "epoch": 1.5358096934232286, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.044686317443848, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8821481466293335, + "num_tokens": 468791530.0, + "step": 12073 + }, + { + "epoch": 1.5359369037018191, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.095776557922363, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8700981140136719, + "num_tokens": 468836227.0, + "step": 12074 + }, + { + "epoch": 1.5360641139804097, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.020891189575195, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8821922540664673, + "num_tokens": 468876870.0, + "step": 12075 + }, + { + "epoch": 1.5361913242590002, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.143235206604004, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8820469975471497, + "num_tokens": 468912292.0, + "step": 12076 + }, + { + "epoch": 1.5363185345375907, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.10776138305664, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.878338098526001, + "num_tokens": 468950690.0, + "step": 12077 + }, + { + "epoch": 1.5364457448161812, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.074774742126465, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8827755451202393, + "num_tokens": 468991469.0, + "step": 12078 + }, + { + "epoch": 1.5365729550947718, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.177704811096191, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8815816640853882, + "num_tokens": 469027643.0, + "step": 12079 + }, + { + "epoch": 1.5367001653733623, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.044885635375977, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8916678428649902, + "num_tokens": 469063838.0, + "step": 12080 + }, + { + "epoch": 1.5368273756519528, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.131118774414062, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8776670694351196, + "num_tokens": 469103021.0, + "step": 12081 + }, + { + "epoch": 1.5369545859305433, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.19924545288086, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8797765374183655, + "num_tokens": 469136592.0, + "step": 12082 + }, + { + "epoch": 1.5370817962091337, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.060569763183594, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8819785118103027, + "num_tokens": 469176133.0, + "step": 12083 + }, + { + "epoch": 1.5372090064877242, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.13890266418457, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8764889240264893, + "num_tokens": 469217358.0, + "step": 12084 + }, + { + "epoch": 1.5373362167663147, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.150826454162598, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.8746668100357056, + "num_tokens": 469255510.0, + "step": 12085 + }, + { + "epoch": 1.5374634270449052, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.027162551879883, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8878981471061707, + "num_tokens": 469296533.0, + "step": 12086 + }, + { + "epoch": 1.5375906373234958, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.047435760498047, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8813593983650208, + "num_tokens": 469339448.0, + "step": 12087 + }, + { + "epoch": 1.5377178476020863, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.994747161865234, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.9002082347869873, + "num_tokens": 469377583.0, + "step": 12088 + }, + { + "epoch": 1.5378450578806766, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.08946418762207, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8834335803985596, + "num_tokens": 469415995.0, + "step": 12089 + }, + { + "epoch": 1.5379722681592671, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.050677299499512, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8970577716827393, + "num_tokens": 469457426.0, + "step": 12090 + }, + { + "epoch": 1.5380994784378577, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.170876502990723, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8802981376647949, + "num_tokens": 469494208.0, + "step": 12091 + }, + { + "epoch": 1.5382266887164482, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.054350852966309, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8908788561820984, + "num_tokens": 469530840.0, + "step": 12092 + }, + { + "epoch": 1.5383538989950387, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.081672668457031, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8977946043014526, + "num_tokens": 469567121.0, + "step": 12093 + }, + { + "epoch": 1.5384811092736292, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.089544296264648, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.883362352848053, + "num_tokens": 469604314.0, + "step": 12094 + }, + { + "epoch": 1.5386083195522198, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.11691665649414, + "learning_rate": 1e-06, + "loss": 0.3714, + "mean_token_accuracy": 0.8712378144264221, + "num_tokens": 469639722.0, + "step": 12095 + }, + { + "epoch": 1.5387355298308103, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.113832473754883, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8776267170906067, + "num_tokens": 469681561.0, + "step": 12096 + }, + { + "epoch": 1.5388627401094008, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.081581115722656, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8869458436965942, + "num_tokens": 469723677.0, + "step": 12097 + }, + { + "epoch": 1.5389899503879914, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.107096672058105, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8812280893325806, + "num_tokens": 469765549.0, + "step": 12098 + }, + { + "epoch": 1.5391171606665819, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.169110298156738, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8777360916137695, + "num_tokens": 469800726.0, + "step": 12099 + }, + { + "epoch": 1.5392443709451724, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.058011054992676, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8848525285720825, + "num_tokens": 469840404.0, + "step": 12100 + }, + { + "epoch": 1.539371581223763, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.066490173339844, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8854800462722778, + "num_tokens": 469878983.0, + "step": 12101 + }, + { + "epoch": 1.5394987915023535, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.073827743530273, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8853249549865723, + "num_tokens": 469919836.0, + "step": 12102 + }, + { + "epoch": 1.539626001780944, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.054590225219727, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8972281813621521, + "num_tokens": 469957392.0, + "step": 12103 + }, + { + "epoch": 1.5397532120595345, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.268617630004883, + "learning_rate": 1e-06, + "loss": 0.3809, + "mean_token_accuracy": 0.873167097568512, + "num_tokens": 469995630.0, + "step": 12104 + }, + { + "epoch": 1.539880422338125, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.960945129394531, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8991145491600037, + "num_tokens": 470037305.0, + "step": 12105 + }, + { + "epoch": 1.5400076326167156, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.040558815002441, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8794127702713013, + "num_tokens": 470079982.0, + "step": 12106 + }, + { + "epoch": 1.5401348428953059, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.179722785949707, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8910712003707886, + "num_tokens": 470116468.0, + "step": 12107 + }, + { + "epoch": 1.5402620531738964, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.109546661376953, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8917069435119629, + "num_tokens": 470152581.0, + "step": 12108 + }, + { + "epoch": 1.540389263452487, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.068046569824219, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.9007077217102051, + "num_tokens": 470192251.0, + "step": 12109 + }, + { + "epoch": 1.5405164737310775, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.05482292175293, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8907620906829834, + "num_tokens": 470235958.0, + "step": 12110 + }, + { + "epoch": 1.540643684009668, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.088522911071777, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8917286396026611, + "num_tokens": 470275552.0, + "step": 12111 + }, + { + "epoch": 1.5407708942882585, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.132490158081055, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8975816965103149, + "num_tokens": 470310650.0, + "step": 12112 + }, + { + "epoch": 1.5408981045668488, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.17107105255127, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8885642290115356, + "num_tokens": 470348466.0, + "step": 12113 + }, + { + "epoch": 1.5410253148454394, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.972230911254883, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8861085176467896, + "num_tokens": 470395996.0, + "step": 12114 + }, + { + "epoch": 1.54115252512403, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.103281021118164, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8858227729797363, + "num_tokens": 470437363.0, + "step": 12115 + }, + { + "epoch": 1.5412797354026204, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.128090858459473, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8826169371604919, + "num_tokens": 470479208.0, + "step": 12116 + }, + { + "epoch": 1.541406945681211, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.10537052154541, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8899528384208679, + "num_tokens": 470516966.0, + "step": 12117 + }, + { + "epoch": 1.5415341559598015, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.034598350524902, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8862895369529724, + "num_tokens": 470561751.0, + "step": 12118 + }, + { + "epoch": 1.541661366238392, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.192787170410156, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8822883367538452, + "num_tokens": 470598795.0, + "step": 12119 + }, + { + "epoch": 1.5417885765169825, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.698204040527344e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.996881484985352, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8946810364723206, + "num_tokens": 470643067.0, + "step": 12120 + }, + { + "epoch": 1.541915786795573, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.148457527160645, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8866689205169678, + "num_tokens": 470683032.0, + "step": 12121 + }, + { + "epoch": 1.5420429970741636, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.285243034362793, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8818880319595337, + "num_tokens": 470714089.0, + "step": 12122 + }, + { + "epoch": 1.5421702073527541, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.051648139953613, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8879046440124512, + "num_tokens": 470752477.0, + "step": 12123 + }, + { + "epoch": 1.5422974176313446, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.04602336883545, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8886257410049438, + "num_tokens": 470793798.0, + "step": 12124 + }, + { + "epoch": 1.5424246279099352, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.166864395141602, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8962622880935669, + "num_tokens": 470826403.0, + "step": 12125 + }, + { + "epoch": 1.5425518381885257, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.077155113220215, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8871640563011169, + "num_tokens": 470866440.0, + "step": 12126 + }, + { + "epoch": 1.5426790484671162, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.22159194946289, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8836491107940674, + "num_tokens": 470899122.0, + "step": 12127 + }, + { + "epoch": 1.5428062587457068, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.140400886535645, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9002209305763245, + "num_tokens": 470930936.0, + "step": 12128 + }, + { + "epoch": 1.5429334690242973, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.091978073120117, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.892500102519989, + "num_tokens": 470966920.0, + "step": 12129 + }, + { + "epoch": 1.5430606793028878, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.088739395141602, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8900750279426575, + "num_tokens": 471008163.0, + "step": 12130 + }, + { + "epoch": 1.5431878895814783, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.069565773010254, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8864196538925171, + "num_tokens": 471052910.0, + "step": 12131 + }, + { + "epoch": 1.5433150998600687, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.114826202392578, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.888249397277832, + "num_tokens": 471091547.0, + "step": 12132 + }, + { + "epoch": 1.5434423101386592, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.063340187072754, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8886575698852539, + "num_tokens": 471135853.0, + "step": 12133 + }, + { + "epoch": 1.5435695204172497, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.048428535461426, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.893627405166626, + "num_tokens": 471176367.0, + "step": 12134 + }, + { + "epoch": 1.5436967306958402, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.188302040100098, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.895460844039917, + "num_tokens": 471210881.0, + "step": 12135 + }, + { + "epoch": 1.5438239409744308, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.189502716064453, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8808406591415405, + "num_tokens": 471246921.0, + "step": 12136 + }, + { + "epoch": 1.5439511512530213, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.167628288269043, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.889255940914154, + "num_tokens": 471283022.0, + "step": 12137 + }, + { + "epoch": 1.5440783615316116, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.051573753356934, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9005512595176697, + "num_tokens": 471319801.0, + "step": 12138 + }, + { + "epoch": 1.5442055718102021, + "ewc_loss": 0.0159912109375, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0251998901367188e-05, + "grad_norm": 9.108109474182129, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8768373727798462, + "num_tokens": 471358641.0, + "step": 12139 + }, + { + "epoch": 1.5443327820887927, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.09617805480957, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8922368288040161, + "num_tokens": 471397581.0, + "step": 12140 + }, + { + "epoch": 1.5444599923673832, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.160853385925293, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8899370431900024, + "num_tokens": 471436775.0, + "step": 12141 + }, + { + "epoch": 1.5445872026459737, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.206443786621094, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8809945583343506, + "num_tokens": 471477149.0, + "step": 12142 + }, + { + "epoch": 1.5447144129245642, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.012988090515137, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8870604038238525, + "num_tokens": 471518738.0, + "step": 12143 + }, + { + "epoch": 1.5448416232031548, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.137890815734863, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8955885767936707, + "num_tokens": 471555845.0, + "step": 12144 + }, + { + "epoch": 1.5449688334817453, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.195191383361816, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8927362561225891, + "num_tokens": 471588680.0, + "step": 12145 + }, + { + "epoch": 1.5450960437603358, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.141587257385254, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8967117071151733, + "num_tokens": 471624621.0, + "step": 12146 + }, + { + "epoch": 1.5452232540389264, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.086670875549316, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8958725333213806, + "num_tokens": 471664120.0, + "step": 12147 + }, + { + "epoch": 1.5453504643175169, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.22708511352539, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8856854438781738, + "num_tokens": 471697096.0, + "step": 12148 + }, + { + "epoch": 1.5454776745961074, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.987783432006836, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8851922750473022, + "num_tokens": 471742102.0, + "step": 12149 + }, + { + "epoch": 1.545604884874698, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.219314575195312, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8844183087348938, + "num_tokens": 471776461.0, + "step": 12150 + }, + { + "epoch": 1.5457320951532885, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.195350646972656, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8772553205490112, + "num_tokens": 471808370.0, + "step": 12151 + }, + { + "epoch": 1.545859305431879, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.096644401550293, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.887019157409668, + "num_tokens": 471842306.0, + "step": 12152 + }, + { + "epoch": 1.5459865157104695, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.075935363769531, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8971433639526367, + "num_tokens": 471881820.0, + "step": 12153 + }, + { + "epoch": 1.54611372598906, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.118132591247559, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8775045871734619, + "num_tokens": 471918655.0, + "step": 12154 + }, + { + "epoch": 1.5462409362676506, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.178306579589844, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8697614669799805, + "num_tokens": 471960819.0, + "step": 12155 + }, + { + "epoch": 1.5463681465462409, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.115946769714355, + "learning_rate": 1e-06, + "loss": 0.393, + "mean_token_accuracy": 0.8698034286499023, + "num_tokens": 471998923.0, + "step": 12156 + }, + { + "epoch": 1.5464953568248314, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.00625228881836, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9002042412757874, + "num_tokens": 472047885.0, + "step": 12157 + }, + { + "epoch": 1.546622567103422, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.046216011047363, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8878982067108154, + "num_tokens": 472091016.0, + "step": 12158 + }, + { + "epoch": 1.5467497773820125, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.130714416503906, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8861762285232544, + "num_tokens": 472128162.0, + "step": 12159 + }, + { + "epoch": 1.546876987660603, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.051105499267578, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8851590156555176, + "num_tokens": 472168127.0, + "step": 12160 + }, + { + "epoch": 1.5470041979391935, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.092024803161621, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8923656940460205, + "num_tokens": 472209777.0, + "step": 12161 + }, + { + "epoch": 1.5471314082177838, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.99414348602295, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8907502889633179, + "num_tokens": 472259903.0, + "step": 12162 + }, + { + "epoch": 1.5472586184963744, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.133437156677246, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8745747804641724, + "num_tokens": 472301073.0, + "step": 12163 + }, + { + "epoch": 1.5473858287749649, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.043135643005371, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8831471800804138, + "num_tokens": 472346033.0, + "step": 12164 + }, + { + "epoch": 1.5475130390535554, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.131156921386719, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8886594772338867, + "num_tokens": 472382542.0, + "step": 12165 + }, + { + "epoch": 1.547640249332146, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.027518272399902, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8972434997558594, + "num_tokens": 472423718.0, + "step": 12166 + }, + { + "epoch": 1.5477674596107365, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.191280364990234, + "learning_rate": 1e-06, + "loss": 0.3776, + "mean_token_accuracy": 0.8728988170623779, + "num_tokens": 472463173.0, + "step": 12167 + }, + { + "epoch": 1.547894669889327, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.028286933898926, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8984810709953308, + "num_tokens": 472504898.0, + "step": 12168 + }, + { + "epoch": 1.5480218801679175, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.128223419189453, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.882225513458252, + "num_tokens": 472541461.0, + "step": 12169 + }, + { + "epoch": 1.548149090446508, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.049251556396484, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8831352591514587, + "num_tokens": 472585123.0, + "step": 12170 + }, + { + "epoch": 1.5482763007250986, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.150039672851562, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8825124502182007, + "num_tokens": 472617860.0, + "step": 12171 + }, + { + "epoch": 1.5484035110036891, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.07967472076416, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8859446048736572, + "num_tokens": 472660436.0, + "step": 12172 + }, + { + "epoch": 1.5485307212822796, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.144155502319336, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8871311545372009, + "num_tokens": 472692214.0, + "step": 12173 + }, + { + "epoch": 1.5486579315608702, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.082053184509277, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8792440891265869, + "num_tokens": 472738518.0, + "step": 12174 + }, + { + "epoch": 1.5487851418394607, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.025452613830566, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8926278352737427, + "num_tokens": 472777731.0, + "step": 12175 + }, + { + "epoch": 1.5489123521180512, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.245347023010254, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8873990178108215, + "num_tokens": 472811227.0, + "step": 12176 + }, + { + "epoch": 1.5490395623966418, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.097066879272461, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8869256377220154, + "num_tokens": 472846462.0, + "step": 12177 + }, + { + "epoch": 1.5491667726752323, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.301247596740723, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8867689967155457, + "num_tokens": 472877217.0, + "step": 12178 + }, + { + "epoch": 1.5492939829538228, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.015522956848145, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.883346438407898, + "num_tokens": 472919116.0, + "step": 12179 + }, + { + "epoch": 1.5494211932324133, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.004271507263184, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.889171838760376, + "num_tokens": 472960788.0, + "step": 12180 + }, + { + "epoch": 1.5495484035110036, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.111627578735352, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8922110795974731, + "num_tokens": 472998371.0, + "step": 12181 + }, + { + "epoch": 1.5496756137895942, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.056103706359863, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8882225155830383, + "num_tokens": 473034046.0, + "step": 12182 + }, + { + "epoch": 1.5498028240681847, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.154895782470703, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.879988431930542, + "num_tokens": 473069058.0, + "step": 12183 + }, + { + "epoch": 1.5499300343467752, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.135587692260742, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8781343698501587, + "num_tokens": 473107172.0, + "step": 12184 + }, + { + "epoch": 1.5500572446253658, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.080636978149414, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8937973976135254, + "num_tokens": 473146155.0, + "step": 12185 + }, + { + "epoch": 1.5501844549039563, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.143656730651855, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8893111348152161, + "num_tokens": 473181888.0, + "step": 12186 + }, + { + "epoch": 1.5503116651825466, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.059069633483887, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8970091342926025, + "num_tokens": 473220482.0, + "step": 12187 + }, + { + "epoch": 1.5504388754611371, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.156461715698242, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8938102722167969, + "num_tokens": 473266515.0, + "step": 12188 + }, + { + "epoch": 1.5505660857397277, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.108098983764648, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8822675943374634, + "num_tokens": 473304466.0, + "step": 12189 + }, + { + "epoch": 1.5506932960183182, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 8.987676620483398, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8950796127319336, + "num_tokens": 473346626.0, + "step": 12190 + }, + { + "epoch": 1.5508205062969087, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.210561752319336, + "learning_rate": 1e-06, + "loss": 0.3942, + "mean_token_accuracy": 0.8700268864631653, + "num_tokens": 473384453.0, + "step": 12191 + }, + { + "epoch": 1.5509477165754992, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.0283842086792, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8937057256698608, + "num_tokens": 473424506.0, + "step": 12192 + }, + { + "epoch": 1.5510749268540898, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 8.990385055541992, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8986001014709473, + "num_tokens": 473467184.0, + "step": 12193 + }, + { + "epoch": 1.5512021371326803, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.15555191040039, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.886969804763794, + "num_tokens": 473505880.0, + "step": 12194 + }, + { + "epoch": 1.5513293474112708, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.121526718139648, + "learning_rate": 1e-06, + "loss": 0.386, + "mean_token_accuracy": 0.8681278228759766, + "num_tokens": 473544550.0, + "step": 12195 + }, + { + "epoch": 1.5514565576898613, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.055537223815918, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8876715898513794, + "num_tokens": 473583763.0, + "step": 12196 + }, + { + "epoch": 1.5515837679684519, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.124736785888672, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8834991455078125, + "num_tokens": 473621111.0, + "step": 12197 + }, + { + "epoch": 1.5517109782470424, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.115964889526367, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8972250819206238, + "num_tokens": 473654559.0, + "step": 12198 + }, + { + "epoch": 1.551838188525633, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.06649112701416, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.891608476638794, + "num_tokens": 473697304.0, + "step": 12199 + }, + { + "epoch": 1.5519653988042235, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.18016242980957, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8857293128967285, + "num_tokens": 473733436.0, + "step": 12200 + }, + { + "epoch": 1.552092609082814, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.01772689819336, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.9020322561264038, + "num_tokens": 473771442.0, + "step": 12201 + }, + { + "epoch": 1.5522198193614045, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.132267951965332, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8951077461242676, + "num_tokens": 473809989.0, + "step": 12202 + }, + { + "epoch": 1.552347029639995, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.02839183807373, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8911830186843872, + "num_tokens": 473851871.0, + "step": 12203 + }, + { + "epoch": 1.5524742399185856, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.251198768615723, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8853609561920166, + "num_tokens": 473884775.0, + "step": 12204 + }, + { + "epoch": 1.5526014501971759, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.101759910583496, + "learning_rate": 1e-06, + "loss": 0.3884, + "mean_token_accuracy": 0.8708163499832153, + "num_tokens": 473925516.0, + "step": 12205 + }, + { + "epoch": 1.5527286604757664, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.179824829101562, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8754245638847351, + "num_tokens": 473960048.0, + "step": 12206 + }, + { + "epoch": 1.552855870754357, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.168265342712402, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8924528360366821, + "num_tokens": 473995326.0, + "step": 12207 + }, + { + "epoch": 1.5529830810329475, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.135783195495605, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8996068835258484, + "num_tokens": 474027869.0, + "step": 12208 + }, + { + "epoch": 1.553110291311538, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.069677352905273, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8914999961853027, + "num_tokens": 474069506.0, + "step": 12209 + }, + { + "epoch": 1.5532375015901285, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7220458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.143352508544922, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8976888656616211, + "num_tokens": 474105314.0, + "step": 12210 + }, + { + "epoch": 1.5533647118687188, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.134215354919434, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.895302951335907, + "num_tokens": 474140956.0, + "step": 12211 + }, + { + "epoch": 1.5534919221473094, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.186574935913086, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8801933526992798, + "num_tokens": 474177834.0, + "step": 12212 + }, + { + "epoch": 1.5536191324258999, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.06149673461914, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8977712392807007, + "num_tokens": 474216640.0, + "step": 12213 + }, + { + "epoch": 1.5537463427044904, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.135429382324219, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8844464421272278, + "num_tokens": 474254829.0, + "step": 12214 + }, + { + "epoch": 1.553873552983081, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.177190780639648, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8911833167076111, + "num_tokens": 474289766.0, + "step": 12215 + }, + { + "epoch": 1.5540007632616715, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.074166297912598, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8712705969810486, + "num_tokens": 474330749.0, + "step": 12216 + }, + { + "epoch": 1.554127973540262, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.138439178466797, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8732889890670776, + "num_tokens": 474370846.0, + "step": 12217 + }, + { + "epoch": 1.5542551838188525, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.986342430114746, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8958776593208313, + "num_tokens": 474412082.0, + "step": 12218 + }, + { + "epoch": 1.554382394097443, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.188261985778809, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8812113404273987, + "num_tokens": 474449818.0, + "step": 12219 + }, + { + "epoch": 1.5545096043760336, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.106133460998535, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8930419087409973, + "num_tokens": 474486109.0, + "step": 12220 + }, + { + "epoch": 1.554636814654624, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.08426284790039, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8935969471931458, + "num_tokens": 474523071.0, + "step": 12221 + }, + { + "epoch": 1.5547640249332146, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.273920059204102, + "learning_rate": 1e-06, + "loss": 0.3899, + "mean_token_accuracy": 0.8686583042144775, + "num_tokens": 474555430.0, + "step": 12222 + }, + { + "epoch": 1.5548912352118052, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.071056365966797, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.886768639087677, + "num_tokens": 474599778.0, + "step": 12223 + }, + { + "epoch": 1.5550184454903957, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.057089805603027, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8942396640777588, + "num_tokens": 474640875.0, + "step": 12224 + }, + { + "epoch": 1.5551456557689862, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.11961555480957, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8955508470535278, + "num_tokens": 474678171.0, + "step": 12225 + }, + { + "epoch": 1.5552728660475768, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.10649299621582, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8779428601264954, + "num_tokens": 474719600.0, + "step": 12226 + }, + { + "epoch": 1.5554000763261673, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.218612670898438, + "learning_rate": 1e-06, + "loss": 0.385, + "mean_token_accuracy": 0.8712372779846191, + "num_tokens": 474761600.0, + "step": 12227 + }, + { + "epoch": 1.5555272866047578, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.164381980895996, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8827053308486938, + "num_tokens": 474801466.0, + "step": 12228 + }, + { + "epoch": 1.5556544968833483, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.111538887023926, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.887584924697876, + "num_tokens": 474838845.0, + "step": 12229 + }, + { + "epoch": 1.5557817071619386, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.104249954223633, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8828564882278442, + "num_tokens": 474882856.0, + "step": 12230 + }, + { + "epoch": 1.5559089174405292, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.065327644348145, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.894734799861908, + "num_tokens": 474923049.0, + "step": 12231 + }, + { + "epoch": 1.5560361277191197, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.253235816955566, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.8705499172210693, + "num_tokens": 474958268.0, + "step": 12232 + }, + { + "epoch": 1.5561633379977102, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.117825508117676, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8768061399459839, + "num_tokens": 474994833.0, + "step": 12233 + }, + { + "epoch": 1.5562905482763008, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.156702995300293, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8793904781341553, + "num_tokens": 475031299.0, + "step": 12234 + }, + { + "epoch": 1.5564177585548913, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.062704086303711, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.894249439239502, + "num_tokens": 475069282.0, + "step": 12235 + }, + { + "epoch": 1.5565449688334816, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.169861793518066, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8880103826522827, + "num_tokens": 475107124.0, + "step": 12236 + }, + { + "epoch": 1.5566721791120721, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.170207977294922, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.890394389629364, + "num_tokens": 475143891.0, + "step": 12237 + }, + { + "epoch": 1.5567993893906626, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.106706619262695, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8795570731163025, + "num_tokens": 475188018.0, + "step": 12238 + }, + { + "epoch": 1.5569265996692532, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.084929466247559, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8823856711387634, + "num_tokens": 475231034.0, + "step": 12239 + }, + { + "epoch": 1.5570538099478437, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.169632911682129, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8821060657501221, + "num_tokens": 475268742.0, + "step": 12240 + }, + { + "epoch": 1.5571810202264342, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.13525104522705, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8907142877578735, + "num_tokens": 475304401.0, + "step": 12241 + }, + { + "epoch": 1.5573082305050248, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.106082916259766, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8863916397094727, + "num_tokens": 475346299.0, + "step": 12242 + }, + { + "epoch": 1.5574354407836153, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.122925758361816, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8871232867240906, + "num_tokens": 475381317.0, + "step": 12243 + }, + { + "epoch": 1.5575626510622058, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.089256286621094, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8962042927742004, + "num_tokens": 475419876.0, + "step": 12244 + }, + { + "epoch": 1.5576898613407963, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.109297752380371, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8873866200447083, + "num_tokens": 475460967.0, + "step": 12245 + }, + { + "epoch": 1.5578170716193869, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.06296443939209, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8942709565162659, + "num_tokens": 475502926.0, + "step": 12246 + }, + { + "epoch": 1.5579442818979774, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.025604248046875, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8966753482818604, + "num_tokens": 475545255.0, + "step": 12247 + }, + { + "epoch": 1.558071492176568, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.12473201751709, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8935220241546631, + "num_tokens": 475582200.0, + "step": 12248 + }, + { + "epoch": 1.5581987024551585, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.957822799682617, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.9017240405082703, + "num_tokens": 475625692.0, + "step": 12249 + }, + { + "epoch": 1.558325912733749, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.180132865905762, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.885177731513977, + "num_tokens": 475664501.0, + "step": 12250 + }, + { + "epoch": 1.5584531230123395, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.074849128723145, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9030242562294006, + "num_tokens": 475704292.0, + "step": 12251 + }, + { + "epoch": 1.55858033329093, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.067924499511719, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8768884539604187, + "num_tokens": 475748197.0, + "step": 12252 + }, + { + "epoch": 1.5587075435695206, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.21390151977539, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8854784965515137, + "num_tokens": 475786814.0, + "step": 12253 + }, + { + "epoch": 1.5588347538481109, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.056171417236328, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9080743789672852, + "num_tokens": 475820554.0, + "step": 12254 + }, + { + "epoch": 1.5589619641267014, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.106951713562012, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.9019889235496521, + "num_tokens": 475859387.0, + "step": 12255 + }, + { + "epoch": 1.559089174405292, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.087432861328125, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.893147349357605, + "num_tokens": 475896997.0, + "step": 12256 + }, + { + "epoch": 1.5592163846838825, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.048531532287598, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8892699480056763, + "num_tokens": 475934991.0, + "step": 12257 + }, + { + "epoch": 1.559343594962473, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.156683921813965, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8909478187561035, + "num_tokens": 475975458.0, + "step": 12258 + }, + { + "epoch": 1.5594708052410635, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.07837963104248, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8811999559402466, + "num_tokens": 476015786.0, + "step": 12259 + }, + { + "epoch": 1.5595980155196538, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.179668426513672, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.877158522605896, + "num_tokens": 476052353.0, + "step": 12260 + }, + { + "epoch": 1.5597252257982444, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.11611557006836, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8808313608169556, + "num_tokens": 476089247.0, + "step": 12261 + }, + { + "epoch": 1.5598524360768349, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.093753814697266, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8767648935317993, + "num_tokens": 476132645.0, + "step": 12262 + }, + { + "epoch": 1.5599796463554254, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.18198013305664, + "learning_rate": 1e-06, + "loss": 0.381, + "mean_token_accuracy": 0.8677436113357544, + "num_tokens": 476167939.0, + "step": 12263 + }, + { + "epoch": 1.560106856634016, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.103668212890625, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8834524154663086, + "num_tokens": 476207675.0, + "step": 12264 + }, + { + "epoch": 1.5602340669126065, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.167964935302734, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8897882103919983, + "num_tokens": 476240867.0, + "step": 12265 + }, + { + "epoch": 1.560361277191197, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.117709159851074, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8839771747589111, + "num_tokens": 476278309.0, + "step": 12266 + }, + { + "epoch": 1.5604884874697875, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.082841873168945, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8947840332984924, + "num_tokens": 476314588.0, + "step": 12267 + }, + { + "epoch": 1.560615697748378, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.076620101928711, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.898120641708374, + "num_tokens": 476349357.0, + "step": 12268 + }, + { + "epoch": 1.5607429080269686, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.123558044433594, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8980953693389893, + "num_tokens": 476381895.0, + "step": 12269 + }, + { + "epoch": 1.560870118305559, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.096323013305664, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8856840133666992, + "num_tokens": 476417591.0, + "step": 12270 + }, + { + "epoch": 1.5609973285841496, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.12263298034668, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8867748379707336, + "num_tokens": 476458452.0, + "step": 12271 + }, + { + "epoch": 1.5611245388627402, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.042133331298828, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.9002641439437866, + "num_tokens": 476492566.0, + "step": 12272 + }, + { + "epoch": 1.5612517491413307, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.110825538635254, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8831400275230408, + "num_tokens": 476531236.0, + "step": 12273 + }, + { + "epoch": 1.5613789594199212, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.056977272033691, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8864226937294006, + "num_tokens": 476574013.0, + "step": 12274 + }, + { + "epoch": 1.5615061696985117, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.097076416015625, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8834037184715271, + "num_tokens": 476612001.0, + "step": 12275 + }, + { + "epoch": 1.5616333799771023, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.211977005004883, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8855178356170654, + "num_tokens": 476647250.0, + "step": 12276 + }, + { + "epoch": 1.5617605902556928, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.037433624267578, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8899863362312317, + "num_tokens": 476692078.0, + "step": 12277 + }, + { + "epoch": 1.5618878005342833, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.067139625549316, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8950170278549194, + "num_tokens": 476731937.0, + "step": 12278 + }, + { + "epoch": 1.5620150108128736, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.075418472290039, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.900518536567688, + "num_tokens": 476771916.0, + "step": 12279 + }, + { + "epoch": 1.5621422210914642, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.14634895324707, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8764055371284485, + "num_tokens": 476808333.0, + "step": 12280 + }, + { + "epoch": 1.5622694313700547, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.05941390991211, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8869754076004028, + "num_tokens": 476850917.0, + "step": 12281 + }, + { + "epoch": 1.5623966416486452, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.13177490234375, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8814318180084229, + "num_tokens": 476885525.0, + "step": 12282 + }, + { + "epoch": 1.5625238519272358, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.057340621948242, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.883532702922821, + "num_tokens": 476927811.0, + "step": 12283 + }, + { + "epoch": 1.5626510622058263, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.134658813476562, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8763200044631958, + "num_tokens": 476966617.0, + "step": 12284 + }, + { + "epoch": 1.5627782724844166, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.242488861083984, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8884812593460083, + "num_tokens": 477002065.0, + "step": 12285 + }, + { + "epoch": 1.5629054827630071, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.040592193603516, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8842333555221558, + "num_tokens": 477041997.0, + "step": 12286 + }, + { + "epoch": 1.5630326930415976, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.171996116638184, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8795071244239807, + "num_tokens": 477080322.0, + "step": 12287 + }, + { + "epoch": 1.5631599033201882, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.126989364624023, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8829463720321655, + "num_tokens": 477118683.0, + "step": 12288 + }, + { + "epoch": 1.5632871135987787, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.144434928894043, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8833913803100586, + "num_tokens": 477153871.0, + "step": 12289 + }, + { + "epoch": 1.5634143238773692, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.065561294555664, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8957300782203674, + "num_tokens": 477191301.0, + "step": 12290 + }, + { + "epoch": 1.5635415341559598, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.079025268554688, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8977814316749573, + "num_tokens": 477228339.0, + "step": 12291 + }, + { + "epoch": 1.5636687444345503, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.096661567687988, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8989701867103577, + "num_tokens": 477268056.0, + "step": 12292 + }, + { + "epoch": 1.5637959547131408, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.063142776489258, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8949223160743713, + "num_tokens": 477301807.0, + "step": 12293 + }, + { + "epoch": 1.5639231649917313, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.161322593688965, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8904662132263184, + "num_tokens": 477336283.0, + "step": 12294 + }, + { + "epoch": 1.5640503752703219, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.017727851867676, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8960033059120178, + "num_tokens": 477380163.0, + "step": 12295 + }, + { + "epoch": 1.5641775855489124, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.00904369354248, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8944523930549622, + "num_tokens": 477420994.0, + "step": 12296 + }, + { + "epoch": 1.564304795827503, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.175409317016602, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.884026050567627, + "num_tokens": 477459433.0, + "step": 12297 + }, + { + "epoch": 1.5644320061060935, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 8.995159149169922, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.9085137248039246, + "num_tokens": 477500547.0, + "step": 12298 + }, + { + "epoch": 1.564559216384684, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.046160697937012, + "learning_rate": 1e-06, + "loss": 0.3868, + "mean_token_accuracy": 0.8699432611465454, + "num_tokens": 477544196.0, + "step": 12299 + }, + { + "epoch": 1.5646864266632745, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.13468074798584, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8960427045822144, + "num_tokens": 477584770.0, + "step": 12300 + }, + { + "epoch": 1.564813636941865, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.041918754577637, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8896642327308655, + "num_tokens": 477623483.0, + "step": 12301 + }, + { + "epoch": 1.5649408472204556, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.121896743774414, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8997623920440674, + "num_tokens": 477659481.0, + "step": 12302 + }, + { + "epoch": 1.5650680574990459, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.233559608459473, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8812592625617981, + "num_tokens": 477698597.0, + "step": 12303 + }, + { + "epoch": 1.5651952677776364, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.107366561889648, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8839936256408691, + "num_tokens": 477733645.0, + "step": 12304 + }, + { + "epoch": 1.565322478056227, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.179924011230469, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8779805898666382, + "num_tokens": 477771582.0, + "step": 12305 + }, + { + "epoch": 1.5654496883348175, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.149611473083496, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8815404176712036, + "num_tokens": 477809615.0, + "step": 12306 + }, + { + "epoch": 1.565576898613408, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.15964126586914, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8766127824783325, + "num_tokens": 477845652.0, + "step": 12307 + }, + { + "epoch": 1.5657041088919985, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.057985305786133, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8891199827194214, + "num_tokens": 477891416.0, + "step": 12308 + }, + { + "epoch": 1.5658313191705888, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.124709129333496, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8880004286766052, + "num_tokens": 477926186.0, + "step": 12309 + }, + { + "epoch": 1.5659585294491793, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.041269302368164, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8923019766807556, + "num_tokens": 477969862.0, + "step": 12310 + }, + { + "epoch": 1.5660857397277699, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 8.96766185760498, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8890236616134644, + "num_tokens": 478014675.0, + "step": 12311 + }, + { + "epoch": 1.5662129500063604, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.07622241973877, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8793950080871582, + "num_tokens": 478056300.0, + "step": 12312 + }, + { + "epoch": 1.566340160284951, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.154065132141113, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8901597261428833, + "num_tokens": 478092781.0, + "step": 12313 + }, + { + "epoch": 1.5664673705635415, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.143728256225586, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8868764638900757, + "num_tokens": 478126969.0, + "step": 12314 + }, + { + "epoch": 1.566594580842132, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.079400062561035, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8817321062088013, + "num_tokens": 478169451.0, + "step": 12315 + }, + { + "epoch": 1.5667217911207225, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 8.998648643493652, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8996450901031494, + "num_tokens": 478211472.0, + "step": 12316 + }, + { + "epoch": 1.566849001399313, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.219256401062012, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8769127726554871, + "num_tokens": 478247943.0, + "step": 12317 + }, + { + "epoch": 1.5669762116779036, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.09426498413086, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8855372667312622, + "num_tokens": 478291045.0, + "step": 12318 + }, + { + "epoch": 1.567103421956494, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.11228084564209, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8811055421829224, + "num_tokens": 478326786.0, + "step": 12319 + }, + { + "epoch": 1.5672306322350846, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.097103118896484, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8948894143104553, + "num_tokens": 478365326.0, + "step": 12320 + }, + { + "epoch": 1.5673578425136752, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.19671630859375, + "learning_rate": 1e-06, + "loss": 0.3733, + "mean_token_accuracy": 0.8729379177093506, + "num_tokens": 478400650.0, + "step": 12321 + }, + { + "epoch": 1.5674850527922657, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.189236640930176, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8806944489479065, + "num_tokens": 478434782.0, + "step": 12322 + }, + { + "epoch": 1.5676122630708562, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.107132911682129, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8916255831718445, + "num_tokens": 478470068.0, + "step": 12323 + }, + { + "epoch": 1.5677394733494467, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.14615249633789, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8715075254440308, + "num_tokens": 478510184.0, + "step": 12324 + }, + { + "epoch": 1.5678666836280373, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.19303035736084, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.876622200012207, + "num_tokens": 478547447.0, + "step": 12325 + }, + { + "epoch": 1.5679938939066278, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.06118392944336, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8891168832778931, + "num_tokens": 478590548.0, + "step": 12326 + }, + { + "epoch": 1.5681211041852183, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.293667793273926, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8860616683959961, + "num_tokens": 478624540.0, + "step": 12327 + }, + { + "epoch": 1.5682483144638086, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.122891426086426, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.885381281375885, + "num_tokens": 478659376.0, + "step": 12328 + }, + { + "epoch": 1.5683755247423992, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.08703899383545, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8963621258735657, + "num_tokens": 478696061.0, + "step": 12329 + }, + { + "epoch": 1.5685027350209897, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.105695724487305, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8964177370071411, + "num_tokens": 478731724.0, + "step": 12330 + }, + { + "epoch": 1.5686299452995802, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.090116500854492, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8826895952224731, + "num_tokens": 478774982.0, + "step": 12331 + }, + { + "epoch": 1.5687571555781707, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.16909122467041, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8803908824920654, + "num_tokens": 478810340.0, + "step": 12332 + }, + { + "epoch": 1.5688843658567613, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.271613121032715, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8878021836280823, + "num_tokens": 478842690.0, + "step": 12333 + }, + { + "epoch": 1.5690115761353516, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.02564811706543, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.9002004861831665, + "num_tokens": 478883602.0, + "step": 12334 + }, + { + "epoch": 1.569138786413942, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.087746620178223, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8937656283378601, + "num_tokens": 478919534.0, + "step": 12335 + }, + { + "epoch": 1.5692659966925326, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.144876480102539, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8871518969535828, + "num_tokens": 478955998.0, + "step": 12336 + }, + { + "epoch": 1.5693932069711232, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.100187301635742, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8916759490966797, + "num_tokens": 478989984.0, + "step": 12337 + }, + { + "epoch": 1.5695204172497137, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.144813537597656, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8850182294845581, + "num_tokens": 479026508.0, + "step": 12338 + }, + { + "epoch": 1.5696476275283042, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.052905082702637, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8994924426078796, + "num_tokens": 479064797.0, + "step": 12339 + }, + { + "epoch": 1.5697748378068948, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.03408145904541, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8895154595375061, + "num_tokens": 479112579.0, + "step": 12340 + }, + { + "epoch": 1.5699020480854853, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.0920991897583, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8837186098098755, + "num_tokens": 479150068.0, + "step": 12341 + }, + { + "epoch": 1.5700292583640758, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.202953338623047, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8847233057022095, + "num_tokens": 479185507.0, + "step": 12342 + }, + { + "epoch": 1.5701564686426663, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 8.998446464538574, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8847123980522156, + "num_tokens": 479229398.0, + "step": 12343 + }, + { + "epoch": 1.5702836789212569, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.152621269226074, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8953363299369812, + "num_tokens": 479267420.0, + "step": 12344 + }, + { + "epoch": 1.5704108891998474, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.166841506958008, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8925454616546631, + "num_tokens": 479304623.0, + "step": 12345 + }, + { + "epoch": 1.570538099478438, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.109031677246094, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.889621376991272, + "num_tokens": 479344373.0, + "step": 12346 + }, + { + "epoch": 1.5706653097570284, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.15703010559082, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.9010449647903442, + "num_tokens": 479378502.0, + "step": 12347 + }, + { + "epoch": 1.570792520035619, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.048595428466797, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8818137049674988, + "num_tokens": 479423343.0, + "step": 12348 + }, + { + "epoch": 1.5709197303142095, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.18953800201416, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8939310908317566, + "num_tokens": 479458565.0, + "step": 12349 + }, + { + "epoch": 1.5710469405928, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.113692283630371, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8893396854400635, + "num_tokens": 479496680.0, + "step": 12350 + }, + { + "epoch": 1.5711741508713906, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.184661865234375, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8981958627700806, + "num_tokens": 479527861.0, + "step": 12351 + }, + { + "epoch": 1.5713013611499809, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.15603256225586, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8964980244636536, + "num_tokens": 479567897.0, + "step": 12352 + }, + { + "epoch": 1.5714285714285714, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.116658210754395, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.896950900554657, + "num_tokens": 479607014.0, + "step": 12353 + }, + { + "epoch": 1.571555781707162, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.011435508728027, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8918556571006775, + "num_tokens": 479648663.0, + "step": 12354 + }, + { + "epoch": 1.5716829919857525, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.22694206237793, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8837598562240601, + "num_tokens": 479682811.0, + "step": 12355 + }, + { + "epoch": 1.571810202264343, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.162915229797363, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8882995843887329, + "num_tokens": 479719214.0, + "step": 12356 + }, + { + "epoch": 1.5719374125429335, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.03469181060791, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.9039335250854492, + "num_tokens": 479759439.0, + "step": 12357 + }, + { + "epoch": 1.5720646228215238, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.188515663146973, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8887337446212769, + "num_tokens": 479793023.0, + "step": 12358 + }, + { + "epoch": 1.5721918331001143, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.124324798583984, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8944125771522522, + "num_tokens": 479832196.0, + "step": 12359 + }, + { + "epoch": 1.5723190433787049, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.03081226348877, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8902279138565063, + "num_tokens": 479873771.0, + "step": 12360 + }, + { + "epoch": 1.5724462536572954, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.079207420349121, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8839958310127258, + "num_tokens": 479914273.0, + "step": 12361 + }, + { + "epoch": 1.572573463935886, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.150201797485352, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8841860294342041, + "num_tokens": 479951012.0, + "step": 12362 + }, + { + "epoch": 1.5727006742144765, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.104479789733887, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8832671642303467, + "num_tokens": 479990582.0, + "step": 12363 + }, + { + "epoch": 1.572827884493067, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.189114570617676, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8849632143974304, + "num_tokens": 480033358.0, + "step": 12364 + }, + { + "epoch": 1.5729550947716575, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.18358039855957, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8939855098724365, + "num_tokens": 480067959.0, + "step": 12365 + }, + { + "epoch": 1.573082305050248, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.215707778930664, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8826748728752136, + "num_tokens": 480103017.0, + "step": 12366 + }, + { + "epoch": 1.5732095153288386, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.181049346923828, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.879499077796936, + "num_tokens": 480142788.0, + "step": 12367 + }, + { + "epoch": 1.573336725607429, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.067548751831055, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8881741762161255, + "num_tokens": 480180933.0, + "step": 12368 + }, + { + "epoch": 1.5734639358860196, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.200005531311035, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8896980285644531, + "num_tokens": 480212272.0, + "step": 12369 + }, + { + "epoch": 1.5735911461646102, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.107307434082031, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.887403666973114, + "num_tokens": 480251734.0, + "step": 12370 + }, + { + "epoch": 1.5737183564432007, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.088897705078125, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8910356163978577, + "num_tokens": 480290749.0, + "step": 12371 + }, + { + "epoch": 1.5738455667217912, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.216585159301758, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8854924440383911, + "num_tokens": 480328483.0, + "step": 12372 + }, + { + "epoch": 1.5739727770003817, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.745887756347656e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.081668853759766, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8943213820457458, + "num_tokens": 480365671.0, + "step": 12373 + }, + { + "epoch": 1.5740999872789723, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.169200897216797, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8879244923591614, + "num_tokens": 480405970.0, + "step": 12374 + }, + { + "epoch": 1.5742271975575628, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.119922637939453, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8862382173538208, + "num_tokens": 480444756.0, + "step": 12375 + }, + { + "epoch": 1.5743544078361533, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.090304374694824, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8931132555007935, + "num_tokens": 480482506.0, + "step": 12376 + }, + { + "epoch": 1.5744816181147436, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.245503425598145, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.886773943901062, + "num_tokens": 480515687.0, + "step": 12377 + }, + { + "epoch": 1.5746088283933342, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.11060905456543, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8858930468559265, + "num_tokens": 480557842.0, + "step": 12378 + }, + { + "epoch": 1.5747360386719247, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.133687019348145, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8833721876144409, + "num_tokens": 480596631.0, + "step": 12379 + }, + { + "epoch": 1.5748632489505152, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.08558177947998, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8792341351509094, + "num_tokens": 480636499.0, + "step": 12380 + }, + { + "epoch": 1.5749904592291057, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.083792686462402, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8979364037513733, + "num_tokens": 480673916.0, + "step": 12381 + }, + { + "epoch": 1.5751176695076963, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.01856803894043, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8961309790611267, + "num_tokens": 480716022.0, + "step": 12382 + }, + { + "epoch": 1.5752448797862866, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.163331985473633, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8988155722618103, + "num_tokens": 480754199.0, + "step": 12383 + }, + { + "epoch": 1.575372090064877, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.085140228271484, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8827553391456604, + "num_tokens": 480791904.0, + "step": 12384 + }, + { + "epoch": 1.5754993003434676, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.24945068359375, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8888294696807861, + "num_tokens": 480831296.0, + "step": 12385 + }, + { + "epoch": 1.5756265106220582, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.071939468383789, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8845239877700806, + "num_tokens": 480871409.0, + "step": 12386 + }, + { + "epoch": 1.5757537209006487, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.110180854797363, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8891937732696533, + "num_tokens": 480905338.0, + "step": 12387 + }, + { + "epoch": 1.5758809311792392, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.150174140930176, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8886058330535889, + "num_tokens": 480942299.0, + "step": 12388 + }, + { + "epoch": 1.5760081414578297, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.032308578491211, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.890250563621521, + "num_tokens": 480981996.0, + "step": 12389 + }, + { + "epoch": 1.5761353517364203, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.165672302246094, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8846556544303894, + "num_tokens": 481021054.0, + "step": 12390 + }, + { + "epoch": 1.5762625620150108, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.023239135742188, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8883826732635498, + "num_tokens": 481063594.0, + "step": 12391 + }, + { + "epoch": 1.5763897722936013, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.15595817565918, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8914487361907959, + "num_tokens": 481101160.0, + "step": 12392 + }, + { + "epoch": 1.5765169825721919, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.057394981384277, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8977851867675781, + "num_tokens": 481144628.0, + "step": 12393 + }, + { + "epoch": 1.5766441928507824, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.078740119934082, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8929044604301453, + "num_tokens": 481184761.0, + "step": 12394 + }, + { + "epoch": 1.576771403129373, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.140098571777344, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8894771933555603, + "num_tokens": 481225614.0, + "step": 12395 + }, + { + "epoch": 1.5768986134079634, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.154991149902344, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8815451860427856, + "num_tokens": 481266705.0, + "step": 12396 + }, + { + "epoch": 1.577025823686554, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.119047164916992, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8945903778076172, + "num_tokens": 481305134.0, + "step": 12397 + }, + { + "epoch": 1.5771530339651445, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.276018142700195, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8929628133773804, + "num_tokens": 481334105.0, + "step": 12398 + }, + { + "epoch": 1.577280244243735, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.104507446289062, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8925765156745911, + "num_tokens": 481368793.0, + "step": 12399 + }, + { + "epoch": 1.5774074545223256, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.08407211303711, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9040847420692444, + "num_tokens": 481403974.0, + "step": 12400 + }, + { + "epoch": 1.5775346648009159, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.138216972351074, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.881723940372467, + "num_tokens": 481444499.0, + "step": 12401 + }, + { + "epoch": 1.5776618750795064, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.130227088928223, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8729169964790344, + "num_tokens": 481487009.0, + "step": 12402 + }, + { + "epoch": 1.577789085358097, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.192567825317383, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8791166543960571, + "num_tokens": 481524193.0, + "step": 12403 + }, + { + "epoch": 1.5779162956366874, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.078259468078613, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.8718867897987366, + "num_tokens": 481568649.0, + "step": 12404 + }, + { + "epoch": 1.578043505915278, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.179001808166504, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8813324570655823, + "num_tokens": 481609969.0, + "step": 12405 + }, + { + "epoch": 1.5781707161938685, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.09442138671875, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8938273787498474, + "num_tokens": 481651322.0, + "step": 12406 + }, + { + "epoch": 1.5782979264724588, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.156806945800781, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8847976326942444, + "num_tokens": 481687641.0, + "step": 12407 + }, + { + "epoch": 1.5784251367510493, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.153839111328125, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8895847797393799, + "num_tokens": 481724553.0, + "step": 12408 + }, + { + "epoch": 1.5785523470296399, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.104591369628906, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8778437376022339, + "num_tokens": 481765554.0, + "step": 12409 + }, + { + "epoch": 1.5786795573082304, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.17742919921875, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8954435586929321, + "num_tokens": 481799841.0, + "step": 12410 + }, + { + "epoch": 1.578806767586821, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.065457344055176, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8812808394432068, + "num_tokens": 481845873.0, + "step": 12411 + }, + { + "epoch": 1.5789339778654115, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.224822044372559, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8997184634208679, + "num_tokens": 481882921.0, + "step": 12412 + }, + { + "epoch": 1.579061188144002, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.175375938415527, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8821388483047485, + "num_tokens": 481918300.0, + "step": 12413 + }, + { + "epoch": 1.5791883984225925, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.142285346984863, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.9021225571632385, + "num_tokens": 481953656.0, + "step": 12414 + }, + { + "epoch": 1.579315608701183, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.18707275390625, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8850168585777283, + "num_tokens": 481985403.0, + "step": 12415 + }, + { + "epoch": 1.5794428189797736, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.188547134399414, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8759907484054565, + "num_tokens": 482028258.0, + "step": 12416 + }, + { + "epoch": 1.579570029258364, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.077543258666992, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8968701958656311, + "num_tokens": 482067774.0, + "step": 12417 + }, + { + "epoch": 1.5796972395369546, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.13961410522461, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.886679470539093, + "num_tokens": 482107549.0, + "step": 12418 + }, + { + "epoch": 1.5798244498155452, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.121988296508789, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8904222249984741, + "num_tokens": 482143305.0, + "step": 12419 + }, + { + "epoch": 1.5799516600941357, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.22552490234375, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.889258623123169, + "num_tokens": 482176968.0, + "step": 12420 + }, + { + "epoch": 1.5800788703727262, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.121450424194336, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8860526084899902, + "num_tokens": 482218386.0, + "step": 12421 + }, + { + "epoch": 1.5802060806513167, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.251043319702148, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8988965153694153, + "num_tokens": 482248794.0, + "step": 12422 + }, + { + "epoch": 1.5803332909299073, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.138084411621094, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8875690698623657, + "num_tokens": 482289038.0, + "step": 12423 + }, + { + "epoch": 1.5804605012084978, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.090377807617188, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8910925388336182, + "num_tokens": 482323568.0, + "step": 12424 + }, + { + "epoch": 1.5805877114870883, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.097270011901855, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8902432918548584, + "num_tokens": 482361978.0, + "step": 12425 + }, + { + "epoch": 1.5807149217656786, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.030831336975098, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8887379169464111, + "num_tokens": 482404516.0, + "step": 12426 + }, + { + "epoch": 1.5808421320442692, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.072922706604004, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8967634439468384, + "num_tokens": 482441494.0, + "step": 12427 + }, + { + "epoch": 1.5809693423228597, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.057262420654297, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8954549431800842, + "num_tokens": 482485266.0, + "step": 12428 + }, + { + "epoch": 1.5810965526014502, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.129594802856445, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8870055675506592, + "num_tokens": 482524275.0, + "step": 12429 + }, + { + "epoch": 1.5812237628800407, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.11728286743164, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.892706036567688, + "num_tokens": 482563167.0, + "step": 12430 + }, + { + "epoch": 1.5813509731586313, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.15827465057373, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8793402910232544, + "num_tokens": 482601946.0, + "step": 12431 + }, + { + "epoch": 1.5814781834372216, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.04496955871582, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8965744376182556, + "num_tokens": 482641148.0, + "step": 12432 + }, + { + "epoch": 1.581605393715812, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.166877746582031, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8842920064926147, + "num_tokens": 482678376.0, + "step": 12433 + }, + { + "epoch": 1.5817326039944026, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.04353141784668, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8949993848800659, + "num_tokens": 482724364.0, + "step": 12434 + }, + { + "epoch": 1.5818598142729932, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.156380653381348, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8952412009239197, + "num_tokens": 482759469.0, + "step": 12435 + }, + { + "epoch": 1.5819870245515837, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.200472831726074, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8858121633529663, + "num_tokens": 482791203.0, + "step": 12436 + }, + { + "epoch": 1.5821142348301742, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.137622833251953, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8838659524917603, + "num_tokens": 482828003.0, + "step": 12437 + }, + { + "epoch": 1.5822414451087647, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.11620044708252, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8856444358825684, + "num_tokens": 482870735.0, + "step": 12438 + }, + { + "epoch": 1.5823686553873553, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.085321426391602, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8939473628997803, + "num_tokens": 482911356.0, + "step": 12439 + }, + { + "epoch": 1.5824958656659458, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.218023300170898, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8804024457931519, + "num_tokens": 482947460.0, + "step": 12440 + }, + { + "epoch": 1.5826230759445363, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.155516624450684, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8780000805854797, + "num_tokens": 482985038.0, + "step": 12441 + }, + { + "epoch": 1.5827502862231269, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.074553489685059, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8866415023803711, + "num_tokens": 483025586.0, + "step": 12442 + }, + { + "epoch": 1.5828774965017174, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.212325096130371, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8755213022232056, + "num_tokens": 483065278.0, + "step": 12443 + }, + { + "epoch": 1.583004706780308, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.111692428588867, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8888789415359497, + "num_tokens": 483107771.0, + "step": 12444 + }, + { + "epoch": 1.5831319170588984, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.130363464355469, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8873090147972107, + "num_tokens": 483147940.0, + "step": 12445 + }, + { + "epoch": 1.583259127337489, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.165167808532715, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.889417827129364, + "num_tokens": 483182977.0, + "step": 12446 + }, + { + "epoch": 1.5833863376160795, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.06560230255127, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8901711702346802, + "num_tokens": 483228163.0, + "step": 12447 + }, + { + "epoch": 1.58351354789467, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.168625831604004, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.888306736946106, + "num_tokens": 483267830.0, + "step": 12448 + }, + { + "epoch": 1.5836407581732606, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.040863990783691, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8907732367515564, + "num_tokens": 483313555.0, + "step": 12449 + }, + { + "epoch": 1.5837679684518509, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.134214401245117, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8921024203300476, + "num_tokens": 483353315.0, + "step": 12450 + }, + { + "epoch": 1.5838951787304414, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.16378402709961, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.882211446762085, + "num_tokens": 483391481.0, + "step": 12451 + }, + { + "epoch": 1.584022389009032, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.21257495880127, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8828699588775635, + "num_tokens": 483425651.0, + "step": 12452 + }, + { + "epoch": 1.5841495992876224, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.103610038757324, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.892881453037262, + "num_tokens": 483466435.0, + "step": 12453 + }, + { + "epoch": 1.584276809566213, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.7697296142578125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.17104721069336, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.877856433391571, + "num_tokens": 483509404.0, + "step": 12454 + }, + { + "epoch": 1.5844040198448035, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.124114036560059, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8831567168235779, + "num_tokens": 483545877.0, + "step": 12455 + }, + { + "epoch": 1.5845312301233938, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.114830017089844, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8904060125350952, + "num_tokens": 483582996.0, + "step": 12456 + }, + { + "epoch": 1.5846584404019843, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.101680755615234, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8899024724960327, + "num_tokens": 483623131.0, + "step": 12457 + }, + { + "epoch": 1.5847856506805749, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.161750793457031, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8853534460067749, + "num_tokens": 483664472.0, + "step": 12458 + }, + { + "epoch": 1.5849128609591654, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.051322937011719, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8903478384017944, + "num_tokens": 483702665.0, + "step": 12459 + }, + { + "epoch": 1.585040071237756, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.14894962310791, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8918249607086182, + "num_tokens": 483746509.0, + "step": 12460 + }, + { + "epoch": 1.5851672815163464, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.14332389831543, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8895047307014465, + "num_tokens": 483784579.0, + "step": 12461 + }, + { + "epoch": 1.585294491794937, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.16257095336914, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8885269165039062, + "num_tokens": 483823360.0, + "step": 12462 + }, + { + "epoch": 1.5854217020735275, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.149937629699707, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8940352201461792, + "num_tokens": 483859202.0, + "step": 12463 + }, + { + "epoch": 1.585548912352118, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.158919334411621, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.880387008190155, + "num_tokens": 483898468.0, + "step": 12464 + }, + { + "epoch": 1.5856761226307086, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.22443675994873, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8793229460716248, + "num_tokens": 483936355.0, + "step": 12465 + }, + { + "epoch": 1.585803332909299, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.090140342712402, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8885993957519531, + "num_tokens": 483976443.0, + "step": 12466 + }, + { + "epoch": 1.5859305431878896, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.20715045928955, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8842717409133911, + "num_tokens": 484010616.0, + "step": 12467 + }, + { + "epoch": 1.5860577534664801, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.157289505004883, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8956932425498962, + "num_tokens": 484046086.0, + "step": 12468 + }, + { + "epoch": 1.5861849637450707, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.041306495666504, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8871487975120544, + "num_tokens": 484091673.0, + "step": 12469 + }, + { + "epoch": 1.5863121740236612, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.234990119934082, + "learning_rate": 1e-06, + "loss": 0.3951, + "mean_token_accuracy": 0.8676866888999939, + "num_tokens": 484130412.0, + "step": 12470 + }, + { + "epoch": 1.5864393843022517, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.140800476074219, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8845376968383789, + "num_tokens": 484165372.0, + "step": 12471 + }, + { + "epoch": 1.5865665945808423, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.044145584106445, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8980097770690918, + "num_tokens": 484202810.0, + "step": 12472 + }, + { + "epoch": 1.5866938048594328, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.129939079284668, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.881294846534729, + "num_tokens": 484241058.0, + "step": 12473 + }, + { + "epoch": 1.5868210151380233, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.173626899719238, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.892480731010437, + "num_tokens": 484271431.0, + "step": 12474 + }, + { + "epoch": 1.5869482254166136, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.06735610961914, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8948116302490234, + "num_tokens": 484314154.0, + "step": 12475 + }, + { + "epoch": 1.5870754356952042, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.164814949035645, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.88525390625, + "num_tokens": 484349873.0, + "step": 12476 + }, + { + "epoch": 1.5872026459737947, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.167424201965332, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.888599693775177, + "num_tokens": 484382757.0, + "step": 12477 + }, + { + "epoch": 1.5873298562523852, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.22303581237793, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8699542284011841, + "num_tokens": 484415082.0, + "step": 12478 + }, + { + "epoch": 1.5874570665309757, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.087890625, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8879678845405579, + "num_tokens": 484452412.0, + "step": 12479 + }, + { + "epoch": 1.5875842768095663, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.175000190734863, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.885541558265686, + "num_tokens": 484491720.0, + "step": 12480 + }, + { + "epoch": 1.5877114870881566, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.084240913391113, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8868144750595093, + "num_tokens": 484531940.0, + "step": 12481 + }, + { + "epoch": 1.587838697366747, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.067560195922852, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8787351250648499, + "num_tokens": 484576191.0, + "step": 12482 + }, + { + "epoch": 1.5879659076453376, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.125049591064453, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.894072949886322, + "num_tokens": 484614586.0, + "step": 12483 + }, + { + "epoch": 1.5880931179239282, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.091431617736816, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8983175754547119, + "num_tokens": 484651919.0, + "step": 12484 + }, + { + "epoch": 1.5882203282025187, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.134673118591309, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8998539447784424, + "num_tokens": 484686297.0, + "step": 12485 + }, + { + "epoch": 1.5883475384811092, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.123056411743164, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8873391151428223, + "num_tokens": 484730584.0, + "step": 12486 + }, + { + "epoch": 1.5884747487596997, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.097503662109375, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8848705291748047, + "num_tokens": 484772781.0, + "step": 12487 + }, + { + "epoch": 1.5886019590382903, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.174015998840332, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8895379304885864, + "num_tokens": 484808186.0, + "step": 12488 + }, + { + "epoch": 1.5887291693168808, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.11876392364502, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8848575353622437, + "num_tokens": 484845862.0, + "step": 12489 + }, + { + "epoch": 1.5888563795954713, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.184735298156738, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8856356143951416, + "num_tokens": 484883886.0, + "step": 12490 + }, + { + "epoch": 1.5889835898740619, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.12057876586914, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8949464559555054, + "num_tokens": 484920542.0, + "step": 12491 + }, + { + "epoch": 1.5891108001526524, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.109430313110352, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8829624056816101, + "num_tokens": 484965692.0, + "step": 12492 + }, + { + "epoch": 1.589238010431243, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.143304824829102, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8715545535087585, + "num_tokens": 485009143.0, + "step": 12493 + }, + { + "epoch": 1.5893652207098334, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.095677375793457, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8940133452415466, + "num_tokens": 485044528.0, + "step": 12494 + }, + { + "epoch": 1.589492430988424, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.218640327453613, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8862234950065613, + "num_tokens": 485080738.0, + "step": 12495 + }, + { + "epoch": 1.5896196412670145, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.101235389709473, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.895789623260498, + "num_tokens": 485116137.0, + "step": 12496 + }, + { + "epoch": 1.589746851545605, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.139677047729492, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8884781002998352, + "num_tokens": 485157573.0, + "step": 12497 + }, + { + "epoch": 1.5898740618241956, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.104372024536133, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8897340297698975, + "num_tokens": 485196770.0, + "step": 12498 + }, + { + "epoch": 1.5900012721027859, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.167436599731445, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.891220211982727, + "num_tokens": 485230054.0, + "step": 12499 + }, + { + "epoch": 1.5901284823813764, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.062450408935547, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8751548528671265, + "num_tokens": 485274676.0, + "step": 12500 + }, + { + "epoch": 1.590255692659967, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.116939544677734, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8967241644859314, + "num_tokens": 485313465.0, + "step": 12501 + }, + { + "epoch": 1.5903829029385574, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.131932258605957, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8863347768783569, + "num_tokens": 485355410.0, + "step": 12502 + }, + { + "epoch": 1.590510113217148, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.142120361328125, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.88775634765625, + "num_tokens": 485396406.0, + "step": 12503 + }, + { + "epoch": 1.5906373234957385, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.113943099975586, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8965632915496826, + "num_tokens": 485433149.0, + "step": 12504 + }, + { + "epoch": 1.5907645337743288, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.157466888427734, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8876878023147583, + "num_tokens": 485472893.0, + "step": 12505 + }, + { + "epoch": 1.5908917440529193, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.12997055053711, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8835750222206116, + "num_tokens": 485517001.0, + "step": 12506 + }, + { + "epoch": 1.5910189543315099, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.150588035583496, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8920266628265381, + "num_tokens": 485556718.0, + "step": 12507 + }, + { + "epoch": 1.5911461646101004, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.161476135253906, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.886501669883728, + "num_tokens": 485595110.0, + "step": 12508 + }, + { + "epoch": 1.591273374888691, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.216304779052734, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8911004066467285, + "num_tokens": 485631203.0, + "step": 12509 + }, + { + "epoch": 1.5914005851672814, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.0718994140625, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8933385610580444, + "num_tokens": 485672476.0, + "step": 12510 + }, + { + "epoch": 1.591527795445872, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.132667541503906, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8937380909919739, + "num_tokens": 485709947.0, + "step": 12511 + }, + { + "epoch": 1.5916550057244625, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.198654174804688, + "learning_rate": 1e-06, + "loss": 0.3879, + "mean_token_accuracy": 0.8649210929870605, + "num_tokens": 485751232.0, + "step": 12512 + }, + { + "epoch": 1.591782216003053, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.120363235473633, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8919726014137268, + "num_tokens": 485787886.0, + "step": 12513 + }, + { + "epoch": 1.5919094262816436, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.136336326599121, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8921087980270386, + "num_tokens": 485826837.0, + "step": 12514 + }, + { + "epoch": 1.592036636560234, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.162917137145996, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8992218971252441, + "num_tokens": 485861830.0, + "step": 12515 + }, + { + "epoch": 1.5921638468388246, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.109354019165039, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8865782022476196, + "num_tokens": 485904739.0, + "step": 12516 + }, + { + "epoch": 1.5922910571174151, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.163684844970703, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8880288600921631, + "num_tokens": 485943163.0, + "step": 12517 + }, + { + "epoch": 1.5924182673960057, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.185311317443848, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8804864287376404, + "num_tokens": 485982132.0, + "step": 12518 + }, + { + "epoch": 1.5925454776745962, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.178388595581055, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8981099128723145, + "num_tokens": 486019223.0, + "step": 12519 + }, + { + "epoch": 1.5926726879531867, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.138705253601074, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8774449825286865, + "num_tokens": 486057771.0, + "step": 12520 + }, + { + "epoch": 1.5927998982317773, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.196414947509766, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8861761093139648, + "num_tokens": 486089143.0, + "step": 12521 + }, + { + "epoch": 1.5929271085103678, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.150276184082031, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8986592292785645, + "num_tokens": 486127917.0, + "step": 12522 + }, + { + "epoch": 1.5930543187889583, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.110361099243164, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8829928636550903, + "num_tokens": 486166302.0, + "step": 12523 + }, + { + "epoch": 1.5931815290675486, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.075399398803711, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8778058886528015, + "num_tokens": 486205970.0, + "step": 12524 + }, + { + "epoch": 1.5933087393461391, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.197232246398926, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8915449380874634, + "num_tokens": 486244072.0, + "step": 12525 + }, + { + "epoch": 1.5934359496247297, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.136991500854492, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8849795460700989, + "num_tokens": 486279588.0, + "step": 12526 + }, + { + "epoch": 1.5935631599033202, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.22748851776123, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.883603036403656, + "num_tokens": 486314898.0, + "step": 12527 + }, + { + "epoch": 1.5936903701819107, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.165966987609863, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8806806802749634, + "num_tokens": 486352084.0, + "step": 12528 + }, + { + "epoch": 1.5938175804605013, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.122047424316406, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8943458795547485, + "num_tokens": 486387365.0, + "step": 12529 + }, + { + "epoch": 1.5939447907390916, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.817413330078125e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.109447479248047, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8833902478218079, + "num_tokens": 486428764.0, + "step": 12530 + }, + { + "epoch": 1.594072001017682, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.110000610351562, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8889535665512085, + "num_tokens": 486471035.0, + "step": 12531 + }, + { + "epoch": 1.5941992112962726, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.144654273986816, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8820825219154358, + "num_tokens": 486512817.0, + "step": 12532 + }, + { + "epoch": 1.5943264215748632, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.238677024841309, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8987532258033752, + "num_tokens": 486545118.0, + "step": 12533 + }, + { + "epoch": 1.5944536318534537, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.193741798400879, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8826220035552979, + "num_tokens": 486578553.0, + "step": 12534 + }, + { + "epoch": 1.5945808421320442, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.104606628417969, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8874891400337219, + "num_tokens": 486619763.0, + "step": 12535 + }, + { + "epoch": 1.5947080524106347, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.164870262145996, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8953053951263428, + "num_tokens": 486661691.0, + "step": 12536 + }, + { + "epoch": 1.5948352626892253, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.083486557006836, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8936820030212402, + "num_tokens": 486700335.0, + "step": 12537 + }, + { + "epoch": 1.5949624729678158, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.257238388061523, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8911851644515991, + "num_tokens": 486735940.0, + "step": 12538 + }, + { + "epoch": 1.5950896832464063, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.202797889709473, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.887679934501648, + "num_tokens": 486771383.0, + "step": 12539 + }, + { + "epoch": 1.5952168935249968, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.201170921325684, + "learning_rate": 1e-06, + "loss": 0.3761, + "mean_token_accuracy": 0.8723618984222412, + "num_tokens": 486807390.0, + "step": 12540 + }, + { + "epoch": 1.5953441038035874, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.139660835266113, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.890844464302063, + "num_tokens": 486851858.0, + "step": 12541 + }, + { + "epoch": 1.595471314082178, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.168044090270996, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.883330762386322, + "num_tokens": 486891775.0, + "step": 12542 + }, + { + "epoch": 1.5955985243607684, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.098807334899902, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8849017024040222, + "num_tokens": 486932587.0, + "step": 12543 + }, + { + "epoch": 1.595725734639359, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.131950378417969, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8912327289581299, + "num_tokens": 486971811.0, + "step": 12544 + }, + { + "epoch": 1.5958529449179495, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.157964706420898, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.885018527507782, + "num_tokens": 487009908.0, + "step": 12545 + }, + { + "epoch": 1.59598015519654, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.1108980178833, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9101922512054443, + "num_tokens": 487044583.0, + "step": 12546 + }, + { + "epoch": 1.5961073654751305, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.128127098083496, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8905898332595825, + "num_tokens": 487086154.0, + "step": 12547 + }, + { + "epoch": 1.5962345757537209, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.12518310546875, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8944166898727417, + "num_tokens": 487123887.0, + "step": 12548 + }, + { + "epoch": 1.5963617860323114, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.191851615905762, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8870459198951721, + "num_tokens": 487156759.0, + "step": 12549 + }, + { + "epoch": 1.596488996310902, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.102450370788574, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.898439884185791, + "num_tokens": 487194596.0, + "step": 12550 + }, + { + "epoch": 1.5966162065894924, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.265671730041504, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.877373218536377, + "num_tokens": 487237029.0, + "step": 12551 + }, + { + "epoch": 1.596743416868083, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.072749137878418, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8966798782348633, + "num_tokens": 487276842.0, + "step": 12552 + }, + { + "epoch": 1.5968706271466735, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.074640274047852, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8860821723937988, + "num_tokens": 487320050.0, + "step": 12553 + }, + { + "epoch": 1.5969978374252638, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.13171672821045, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8833085894584656, + "num_tokens": 487362434.0, + "step": 12554 + }, + { + "epoch": 1.5971250477038543, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.13442611694336, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8836954832077026, + "num_tokens": 487400516.0, + "step": 12555 + }, + { + "epoch": 1.5972522579824449, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.180736541748047, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8862422108650208, + "num_tokens": 487436479.0, + "step": 12556 + }, + { + "epoch": 1.5973794682610354, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.216217041015625, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8710169792175293, + "num_tokens": 487472090.0, + "step": 12557 + }, + { + "epoch": 1.597506678539626, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.130237579345703, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8907586336135864, + "num_tokens": 487512003.0, + "step": 12558 + }, + { + "epoch": 1.5976338888182164, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.193836212158203, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8881356716156006, + "num_tokens": 487546437.0, + "step": 12559 + }, + { + "epoch": 1.597761099096807, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.174674987792969, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8900002241134644, + "num_tokens": 487580318.0, + "step": 12560 + }, + { + "epoch": 1.5978883093753975, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.011961936950684, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8997285962104797, + "num_tokens": 487620676.0, + "step": 12561 + }, + { + "epoch": 1.598015519653988, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.096074104309082, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8849762678146362, + "num_tokens": 487660652.0, + "step": 12562 + }, + { + "epoch": 1.5981427299325786, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.089936256408691, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8868430852890015, + "num_tokens": 487700597.0, + "step": 12563 + }, + { + "epoch": 1.598269940211169, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.066545486450195, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8937957286834717, + "num_tokens": 487746344.0, + "step": 12564 + }, + { + "epoch": 1.5983971504897596, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.12767219543457, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.889168918132782, + "num_tokens": 487783715.0, + "step": 12565 + }, + { + "epoch": 1.5985243607683501, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.168396949768066, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8893648386001587, + "num_tokens": 487818496.0, + "step": 12566 + }, + { + "epoch": 1.5986515710469407, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.841255187988281e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 8.989513397216797, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8974000215530396, + "num_tokens": 487861567.0, + "step": 12567 + }, + { + "epoch": 1.5987787813255312, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.10290241241455, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8967416286468506, + "num_tokens": 487897470.0, + "step": 12568 + }, + { + "epoch": 1.5989059916041217, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.070433616638184, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8865655660629272, + "num_tokens": 487940053.0, + "step": 12569 + }, + { + "epoch": 1.5990332018827123, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.20753002166748, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.875458836555481, + "num_tokens": 487976514.0, + "step": 12570 + }, + { + "epoch": 1.5991604121613028, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.120185852050781, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8798165321350098, + "num_tokens": 488022344.0, + "step": 12571 + }, + { + "epoch": 1.5992876224398933, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.052947998046875, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8860722184181213, + "num_tokens": 488062631.0, + "step": 12572 + }, + { + "epoch": 1.5994148327184836, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.162083625793457, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8983016014099121, + "num_tokens": 488100118.0, + "step": 12573 + }, + { + "epoch": 1.5995420429970741, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.071724891662598, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8835737705230713, + "num_tokens": 488142240.0, + "step": 12574 + }, + { + "epoch": 1.5996692532756647, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.138143539428711, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8885406255722046, + "num_tokens": 488183799.0, + "step": 12575 + }, + { + "epoch": 1.5997964635542552, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.148054122924805, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8846539855003357, + "num_tokens": 488223132.0, + "step": 12576 + }, + { + "epoch": 1.5999236738328457, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.252632141113281, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8741258978843689, + "num_tokens": 488259492.0, + "step": 12577 + }, + { + "epoch": 1.6000508841114363, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.087873458862305, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8963807821273804, + "num_tokens": 488295121.0, + "step": 12578 + }, + { + "epoch": 1.6001780943900266, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.158291816711426, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8848103284835815, + "num_tokens": 488331485.0, + "step": 12579 + }, + { + "epoch": 1.600305304668617, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.123873710632324, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8805204629898071, + "num_tokens": 488369309.0, + "step": 12580 + }, + { + "epoch": 1.6004325149472076, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.14286994934082, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8890469074249268, + "num_tokens": 488408542.0, + "step": 12581 + }, + { + "epoch": 1.6005597252257981, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.1627836227417, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8840134143829346, + "num_tokens": 488444939.0, + "step": 12582 + }, + { + "epoch": 1.6006869355043887, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.048734664916992, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8888529539108276, + "num_tokens": 488490779.0, + "step": 12583 + }, + { + "epoch": 1.6008141457829792, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.120256423950195, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8921823501586914, + "num_tokens": 488530261.0, + "step": 12584 + }, + { + "epoch": 1.6009413560615697, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.145439147949219, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8897574543952942, + "num_tokens": 488567352.0, + "step": 12585 + }, + { + "epoch": 1.6010685663401603, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.195521354675293, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8827363848686218, + "num_tokens": 488599441.0, + "step": 12586 + }, + { + "epoch": 1.6011957766187508, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.224774360656738, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8866702318191528, + "num_tokens": 488633196.0, + "step": 12587 + }, + { + "epoch": 1.6013229868973413, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.129420280456543, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8967537879943848, + "num_tokens": 488667861.0, + "step": 12588 + }, + { + "epoch": 1.6014501971759318, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.127859115600586, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8984347581863403, + "num_tokens": 488703541.0, + "step": 12589 + }, + { + "epoch": 1.6015774074545224, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.107223510742188, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9032818078994751, + "num_tokens": 488739889.0, + "step": 12590 + }, + { + "epoch": 1.601704617733113, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.229244232177734, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.877493143081665, + "num_tokens": 488777543.0, + "step": 12591 + }, + { + "epoch": 1.6018318280117034, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.178237915039062, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8885141015052795, + "num_tokens": 488811855.0, + "step": 12592 + }, + { + "epoch": 1.601959038290294, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.134788513183594, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8956996202468872, + "num_tokens": 488853864.0, + "step": 12593 + }, + { + "epoch": 1.6020862485688845, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.060876846313477, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8962140083312988, + "num_tokens": 488896850.0, + "step": 12594 + }, + { + "epoch": 1.602213458847475, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.268145561218262, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8771562576293945, + "num_tokens": 488934687.0, + "step": 12595 + }, + { + "epoch": 1.6023406691260655, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.207192420959473, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.879240095615387, + "num_tokens": 488974036.0, + "step": 12596 + }, + { + "epoch": 1.6024678794046558, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.182621002197266, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8900574445724487, + "num_tokens": 489009301.0, + "step": 12597 + }, + { + "epoch": 1.6025950896832464, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.133625984191895, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8894485235214233, + "num_tokens": 489045940.0, + "step": 12598 + }, + { + "epoch": 1.602722299961837, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.250346183776855, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.9013264179229736, + "num_tokens": 489081314.0, + "step": 12599 + }, + { + "epoch": 1.6028495102404274, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.150598526000977, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8879880309104919, + "num_tokens": 489117877.0, + "step": 12600 + }, + { + "epoch": 1.602976720519018, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.217273712158203, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.893068790435791, + "num_tokens": 489153078.0, + "step": 12601 + }, + { + "epoch": 1.6031039307976085, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.099601745605469, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8891916275024414, + "num_tokens": 489193779.0, + "step": 12602 + }, + { + "epoch": 1.6032311410761988, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.145257949829102, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8774814605712891, + "num_tokens": 489230483.0, + "step": 12603 + }, + { + "epoch": 1.6033583513547893, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.102368354797363, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.896300196647644, + "num_tokens": 489266498.0, + "step": 12604 + }, + { + "epoch": 1.6034855616333799, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.081032752990723, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8963388204574585, + "num_tokens": 489306664.0, + "step": 12605 + }, + { + "epoch": 1.6036127719119704, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.098288536071777, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9023765921592712, + "num_tokens": 489340412.0, + "step": 12606 + }, + { + "epoch": 1.603739982190561, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.196617126464844, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.885523796081543, + "num_tokens": 489378086.0, + "step": 12607 + }, + { + "epoch": 1.6038671924691514, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.133268356323242, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8857467174530029, + "num_tokens": 489417701.0, + "step": 12608 + }, + { + "epoch": 1.603994402747742, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.08431625366211, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.9053740501403809, + "num_tokens": 489458771.0, + "step": 12609 + }, + { + "epoch": 1.6041216130263325, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.240118026733398, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8875014781951904, + "num_tokens": 489500280.0, + "step": 12610 + }, + { + "epoch": 1.604248823304923, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.115628242492676, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8895148038864136, + "num_tokens": 489535321.0, + "step": 12611 + }, + { + "epoch": 1.6043760335835135, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.223349571228027, + "learning_rate": 1e-06, + "loss": 0.3924, + "mean_token_accuracy": 0.86813884973526, + "num_tokens": 489575039.0, + "step": 12612 + }, + { + "epoch": 1.604503243862104, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.116464614868164, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8946545124053955, + "num_tokens": 489610798.0, + "step": 12613 + }, + { + "epoch": 1.6046304541406946, + "ewc_loss": 0.01611328125, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.1850004196167, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8742461204528809, + "num_tokens": 489646109.0, + "step": 12614 + }, + { + "epoch": 1.6047576644192851, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.081734657287598, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8878936767578125, + "num_tokens": 489686946.0, + "step": 12615 + }, + { + "epoch": 1.6048848746978757, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.266270637512207, + "learning_rate": 1e-06, + "loss": 0.3978, + "mean_token_accuracy": 0.8674349784851074, + "num_tokens": 489725524.0, + "step": 12616 + }, + { + "epoch": 1.6050120849764662, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.153623580932617, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8874995112419128, + "num_tokens": 489762473.0, + "step": 12617 + }, + { + "epoch": 1.6051392952550567, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.116724014282227, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8878023028373718, + "num_tokens": 489804817.0, + "step": 12618 + }, + { + "epoch": 1.6052665055336472, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.135215759277344, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8923351168632507, + "num_tokens": 489843408.0, + "step": 12619 + }, + { + "epoch": 1.6053937158122378, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.156254768371582, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.880799412727356, + "num_tokens": 489880484.0, + "step": 12620 + }, + { + "epoch": 1.6055209260908283, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.215761184692383, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.891973614692688, + "num_tokens": 489913156.0, + "step": 12621 + }, + { + "epoch": 1.6056481363694186, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.180699348449707, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8885194063186646, + "num_tokens": 489951761.0, + "step": 12622 + }, + { + "epoch": 1.6057753466480091, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.093558311462402, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.9002243280410767, + "num_tokens": 489988545.0, + "step": 12623 + }, + { + "epoch": 1.6059025569265997, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.150710105895996, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8856754302978516, + "num_tokens": 490032540.0, + "step": 12624 + }, + { + "epoch": 1.6060297672051902, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.125818252563477, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8747604489326477, + "num_tokens": 490077796.0, + "step": 12625 + }, + { + "epoch": 1.6061569774837807, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.155525207519531, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8984998464584351, + "num_tokens": 490115662.0, + "step": 12626 + }, + { + "epoch": 1.6062841877623713, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.179287910461426, + "learning_rate": 1e-06, + "loss": 0.3938, + "mean_token_accuracy": 0.8660596013069153, + "num_tokens": 490158799.0, + "step": 12627 + }, + { + "epoch": 1.6064113980409616, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.123984336853027, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8782651424407959, + "num_tokens": 490204748.0, + "step": 12628 + }, + { + "epoch": 1.606538608319552, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.169464111328125, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8799715042114258, + "num_tokens": 490244232.0, + "step": 12629 + }, + { + "epoch": 1.6066658185981426, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.148797988891602, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8788398504257202, + "num_tokens": 490285614.0, + "step": 12630 + }, + { + "epoch": 1.6067930288767331, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.163692474365234, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8936840891838074, + "num_tokens": 490321036.0, + "step": 12631 + }, + { + "epoch": 1.6069202391553237, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.367157936096191, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8887962698936462, + "num_tokens": 490353073.0, + "step": 12632 + }, + { + "epoch": 1.6070474494339142, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.22637939453125, + "learning_rate": 1e-06, + "loss": 0.3769, + "mean_token_accuracy": 0.8723052740097046, + "num_tokens": 490389139.0, + "step": 12633 + }, + { + "epoch": 1.6071746597125047, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.123588562011719, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.900079071521759, + "num_tokens": 490423099.0, + "step": 12634 + }, + { + "epoch": 1.6073018699910953, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.051125526428223, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8986842632293701, + "num_tokens": 490463024.0, + "step": 12635 + }, + { + "epoch": 1.6074290802696858, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.14354133605957, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.884243369102478, + "num_tokens": 490505947.0, + "step": 12636 + }, + { + "epoch": 1.6075562905482763, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.116691589355469, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9061930179595947, + "num_tokens": 490540552.0, + "step": 12637 + }, + { + "epoch": 1.6076835008268668, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.150565147399902, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8888247013092041, + "num_tokens": 490578390.0, + "step": 12638 + }, + { + "epoch": 1.6078107111054574, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.18551254272461, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8840818405151367, + "num_tokens": 490614403.0, + "step": 12639 + }, + { + "epoch": 1.607937921384048, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.134190559387207, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8857970833778381, + "num_tokens": 490654417.0, + "step": 12640 + }, + { + "epoch": 1.6080651316626384, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.140697479248047, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.9016902446746826, + "num_tokens": 490687575.0, + "step": 12641 + }, + { + "epoch": 1.608192341941229, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.229813575744629, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8936673402786255, + "num_tokens": 490726236.0, + "step": 12642 + }, + { + "epoch": 1.6083195522198195, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.17613410949707, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8918793201446533, + "num_tokens": 490760684.0, + "step": 12643 + }, + { + "epoch": 1.60844676249841, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.8650970458984375e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.20393180847168, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.887994647026062, + "num_tokens": 490798050.0, + "step": 12644 + }, + { + "epoch": 1.6085739727770005, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.132877349853516, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8751997947692871, + "num_tokens": 490839028.0, + "step": 12645 + }, + { + "epoch": 1.6087011830555908, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.258282661437988, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8775975704193115, + "num_tokens": 490875491.0, + "step": 12646 + }, + { + "epoch": 1.6088283933341814, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.245469093322754, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8835122585296631, + "num_tokens": 490909509.0, + "step": 12647 + }, + { + "epoch": 1.608955603612772, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.115717887878418, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8876745104789734, + "num_tokens": 490948621.0, + "step": 12648 + }, + { + "epoch": 1.6090828138913624, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.154274940490723, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8828409314155579, + "num_tokens": 490986978.0, + "step": 12649 + }, + { + "epoch": 1.609210024169953, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.121459007263184, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8870518803596497, + "num_tokens": 491023249.0, + "step": 12650 + }, + { + "epoch": 1.6093372344485435, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.228334426879883, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8776519894599915, + "num_tokens": 491064294.0, + "step": 12651 + }, + { + "epoch": 1.6094644447271338, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.148762702941895, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8811599016189575, + "num_tokens": 491103081.0, + "step": 12652 + }, + { + "epoch": 1.6095916550057243, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.15407657623291, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8848508596420288, + "num_tokens": 491142873.0, + "step": 12653 + }, + { + "epoch": 1.6097188652843148, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.099812507629395, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.886987566947937, + "num_tokens": 491183307.0, + "step": 12654 + }, + { + "epoch": 1.6098460755629054, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.120590209960938, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8969622850418091, + "num_tokens": 491220582.0, + "step": 12655 + }, + { + "epoch": 1.609973285841496, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.123706817626953, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8890214562416077, + "num_tokens": 491262303.0, + "step": 12656 + }, + { + "epoch": 1.6101004961200864, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.112386703491211, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8969155550003052, + "num_tokens": 491300688.0, + "step": 12657 + }, + { + "epoch": 1.610227706398677, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.18991756439209, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8916375041007996, + "num_tokens": 491337159.0, + "step": 12658 + }, + { + "epoch": 1.6103549166772675, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.069550514221191, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8808895349502563, + "num_tokens": 491381194.0, + "step": 12659 + }, + { + "epoch": 1.610482126955858, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.179483413696289, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8937867879867554, + "num_tokens": 491416750.0, + "step": 12660 + }, + { + "epoch": 1.6106093372344485, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.202088356018066, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8850066065788269, + "num_tokens": 491454596.0, + "step": 12661 + }, + { + "epoch": 1.610736547513039, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.215703964233398, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8696118593215942, + "num_tokens": 491493672.0, + "step": 12662 + }, + { + "epoch": 1.6108637577916296, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.10764217376709, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.888602614402771, + "num_tokens": 491533748.0, + "step": 12663 + }, + { + "epoch": 1.6109909680702201, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.27283763885498, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8785353899002075, + "num_tokens": 491564767.0, + "step": 12664 + }, + { + "epoch": 1.6111181783488107, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.069707870483398, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8967723846435547, + "num_tokens": 491602089.0, + "step": 12665 + }, + { + "epoch": 1.6112453886274012, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.129192352294922, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.889144778251648, + "num_tokens": 491641214.0, + "step": 12666 + }, + { + "epoch": 1.6113725989059917, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.17840576171875, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8911941051483154, + "num_tokens": 491675230.0, + "step": 12667 + }, + { + "epoch": 1.6114998091845822, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.085980415344238, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8983538150787354, + "num_tokens": 491715819.0, + "step": 12668 + }, + { + "epoch": 1.6116270194631728, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.080216407775879, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8989072442054749, + "num_tokens": 491758848.0, + "step": 12669 + }, + { + "epoch": 1.6117542297417633, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.230708122253418, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8862217664718628, + "num_tokens": 491790439.0, + "step": 12670 + }, + { + "epoch": 1.6118814400203536, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.160199165344238, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8795241117477417, + "num_tokens": 491830169.0, + "step": 12671 + }, + { + "epoch": 1.6120086502989441, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.175549507141113, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8942976593971252, + "num_tokens": 491868032.0, + "step": 12672 + }, + { + "epoch": 1.6121358605775347, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.11198902130127, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8891409039497375, + "num_tokens": 491907841.0, + "step": 12673 + }, + { + "epoch": 1.6122630708561252, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.189693450927734, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8807768821716309, + "num_tokens": 491948060.0, + "step": 12674 + }, + { + "epoch": 1.6123902811347157, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.094907760620117, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8867181539535522, + "num_tokens": 491992305.0, + "step": 12675 + }, + { + "epoch": 1.6125174914133062, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.116439819335938, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8873693346977234, + "num_tokens": 492039363.0, + "step": 12676 + }, + { + "epoch": 1.6126447016918966, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.098735809326172, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8886074423789978, + "num_tokens": 492082857.0, + "step": 12677 + }, + { + "epoch": 1.612771911970487, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.201464653015137, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8892694115638733, + "num_tokens": 492118420.0, + "step": 12678 + }, + { + "epoch": 1.6128991222490776, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.1541166305542, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8834784030914307, + "num_tokens": 492159072.0, + "step": 12679 + }, + { + "epoch": 1.6130263325276681, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.163030624389648, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8969149589538574, + "num_tokens": 492194619.0, + "step": 12680 + }, + { + "epoch": 1.6131535428062587, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.16020393371582, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8900546431541443, + "num_tokens": 492230579.0, + "step": 12681 + }, + { + "epoch": 1.6132807530848492, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.129591941833496, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.880963921546936, + "num_tokens": 492273747.0, + "step": 12682 + }, + { + "epoch": 1.6134079633634397, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.109190940856934, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8927841782569885, + "num_tokens": 492323239.0, + "step": 12683 + }, + { + "epoch": 1.6135351736420303, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.280326843261719, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8856809139251709, + "num_tokens": 492359726.0, + "step": 12684 + }, + { + "epoch": 1.6136623839206208, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.082863807678223, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8971399068832397, + "num_tokens": 492403841.0, + "step": 12685 + }, + { + "epoch": 1.6137895941992113, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.156752586364746, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9033762216567993, + "num_tokens": 492441199.0, + "step": 12686 + }, + { + "epoch": 1.6139168044778018, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.27226734161377, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.883082389831543, + "num_tokens": 492477187.0, + "step": 12687 + }, + { + "epoch": 1.6140440147563924, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.198948860168457, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8861294984817505, + "num_tokens": 492509895.0, + "step": 12688 + }, + { + "epoch": 1.614171225034983, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.131694793701172, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8952391147613525, + "num_tokens": 492553300.0, + "step": 12689 + }, + { + "epoch": 1.6142984353135734, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.174845695495605, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8797544240951538, + "num_tokens": 492594009.0, + "step": 12690 + }, + { + "epoch": 1.614425645592164, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.186443328857422, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8988724946975708, + "num_tokens": 492629852.0, + "step": 12691 + }, + { + "epoch": 1.6145528558707545, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.200957298278809, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8819224834442139, + "num_tokens": 492669215.0, + "step": 12692 + }, + { + "epoch": 1.614680066149345, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.257075309753418, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.9012969732284546, + "num_tokens": 492698870.0, + "step": 12693 + }, + { + "epoch": 1.6148072764279355, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.137222290039062, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8857055306434631, + "num_tokens": 492740278.0, + "step": 12694 + }, + { + "epoch": 1.6149344867065258, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.108832359313965, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8922116756439209, + "num_tokens": 492783270.0, + "step": 12695 + }, + { + "epoch": 1.6150616969851164, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.11375904083252, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.887698769569397, + "num_tokens": 492830083.0, + "step": 12696 + }, + { + "epoch": 1.615188907263707, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.080962181091309, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8940293788909912, + "num_tokens": 492870451.0, + "step": 12697 + }, + { + "epoch": 1.6153161175422974, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.17278003692627, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8868869543075562, + "num_tokens": 492910215.0, + "step": 12698 + }, + { + "epoch": 1.615443327820888, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.264348030090332, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8902817368507385, + "num_tokens": 492945054.0, + "step": 12699 + }, + { + "epoch": 1.6155705380994785, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.096670150756836, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8925727009773254, + "num_tokens": 492990920.0, + "step": 12700 + }, + { + "epoch": 1.6156977483780688, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.229904174804688, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8915029764175415, + "num_tokens": 493024974.0, + "step": 12701 + }, + { + "epoch": 1.6158249586566593, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.0844144821167, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8947948813438416, + "num_tokens": 493071004.0, + "step": 12702 + }, + { + "epoch": 1.6159521689352498, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.204712867736816, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8768587112426758, + "num_tokens": 493110213.0, + "step": 12703 + }, + { + "epoch": 1.6160793792138404, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.114006996154785, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8979275822639465, + "num_tokens": 493147905.0, + "step": 12704 + }, + { + "epoch": 1.616206589492431, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.18234634399414, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.901421070098877, + "num_tokens": 493184087.0, + "step": 12705 + }, + { + "epoch": 1.6163337997710214, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.183650970458984, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8947319984436035, + "num_tokens": 493223547.0, + "step": 12706 + }, + { + "epoch": 1.616461010049612, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.110011100769043, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8955350518226624, + "num_tokens": 493260732.0, + "step": 12707 + }, + { + "epoch": 1.6165882203282025, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.241190910339355, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.9005281925201416, + "num_tokens": 493299690.0, + "step": 12708 + }, + { + "epoch": 1.616715430606793, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.128255844116211, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8847490549087524, + "num_tokens": 493337978.0, + "step": 12709 + }, + { + "epoch": 1.6168426408853835, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.203803062438965, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8864823579788208, + "num_tokens": 493371938.0, + "step": 12710 + }, + { + "epoch": 1.616969851163974, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.083940505981445, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8959336280822754, + "num_tokens": 493410218.0, + "step": 12711 + }, + { + "epoch": 1.6170970614425646, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.185897827148438, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.897081732749939, + "num_tokens": 493443706.0, + "step": 12712 + }, + { + "epoch": 1.6172242717211551, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.219704627990723, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8706474304199219, + "num_tokens": 493482395.0, + "step": 12713 + }, + { + "epoch": 1.6173514819997457, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.182516098022461, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8901021480560303, + "num_tokens": 493524079.0, + "step": 12714 + }, + { + "epoch": 1.6174786922783362, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.098258018493652, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8987303972244263, + "num_tokens": 493568257.0, + "step": 12715 + }, + { + "epoch": 1.6176059025569267, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.131175994873047, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8864015340805054, + "num_tokens": 493608108.0, + "step": 12716 + }, + { + "epoch": 1.6177331128355172, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.1901273727417, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8856931328773499, + "num_tokens": 493647584.0, + "step": 12717 + }, + { + "epoch": 1.6178603231141078, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.145776748657227, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8981401920318604, + "num_tokens": 493687614.0, + "step": 12718 + }, + { + "epoch": 1.6179875333926983, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.241232872009277, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8889758586883545, + "num_tokens": 493723811.0, + "step": 12719 + }, + { + "epoch": 1.6181147436712886, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.156805038452148, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.894134521484375, + "num_tokens": 493762870.0, + "step": 12720 + }, + { + "epoch": 1.6182419539498791, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.165860176086426, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8924974799156189, + "num_tokens": 493797863.0, + "step": 12721 + }, + { + "epoch": 1.6183691642284697, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.161453247070312, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8881210088729858, + "num_tokens": 493836404.0, + "step": 12722 + }, + { + "epoch": 1.6184963745070602, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.115625381469727, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8856945037841797, + "num_tokens": 493877625.0, + "step": 12723 + }, + { + "epoch": 1.6186235847856507, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.164606094360352, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8963257074356079, + "num_tokens": 493910623.0, + "step": 12724 + }, + { + "epoch": 1.6187507950642412, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.042081832885742, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9036005735397339, + "num_tokens": 493956363.0, + "step": 12725 + }, + { + "epoch": 1.6188780053428315, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.19629955291748, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8906810879707336, + "num_tokens": 493989469.0, + "step": 12726 + }, + { + "epoch": 1.619005215621422, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.254055976867676, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8924858570098877, + "num_tokens": 494023278.0, + "step": 12727 + }, + { + "epoch": 1.6191324259000126, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.037720680236816, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8968005776405334, + "num_tokens": 494065683.0, + "step": 12728 + }, + { + "epoch": 1.6192596361786031, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.28873062133789, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8820288181304932, + "num_tokens": 494099726.0, + "step": 12729 + }, + { + "epoch": 1.6193868464571937, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.183952331542969, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8841971158981323, + "num_tokens": 494137243.0, + "step": 12730 + }, + { + "epoch": 1.6195140567357842, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.153009414672852, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8909111022949219, + "num_tokens": 494176186.0, + "step": 12731 + }, + { + "epoch": 1.6196412670143747, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.22465705871582, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.9021217226982117, + "num_tokens": 494210010.0, + "step": 12732 + }, + { + "epoch": 1.6197684772929652, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.130035400390625, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8911324739456177, + "num_tokens": 494246987.0, + "step": 12733 + }, + { + "epoch": 1.6198956875715558, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.125075340270996, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8958046436309814, + "num_tokens": 494285121.0, + "step": 12734 + }, + { + "epoch": 1.6200228978501463, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.075860977172852, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8988944292068481, + "num_tokens": 494323600.0, + "step": 12735 + }, + { + "epoch": 1.6201501081287368, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.145279884338379, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8880053162574768, + "num_tokens": 494363439.0, + "step": 12736 + }, + { + "epoch": 1.6202773184073274, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.126038551330566, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8886160254478455, + "num_tokens": 494403249.0, + "step": 12737 + }, + { + "epoch": 1.6204045286859179, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.465261459350586, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8860025405883789, + "num_tokens": 494440635.0, + "step": 12738 + }, + { + "epoch": 1.6205317389645084, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.067656517028809, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8836299777030945, + "num_tokens": 494482246.0, + "step": 12739 + }, + { + "epoch": 1.620658949243099, + "ewc_loss": 0.0162353515625, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.06207275390625, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.884354829788208, + "num_tokens": 494529783.0, + "step": 12740 + }, + { + "epoch": 1.6207861595216895, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.152652740478516, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8894846439361572, + "num_tokens": 494562802.0, + "step": 12741 + }, + { + "epoch": 1.62091336980028, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.888938903808594e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.175654411315918, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8903960585594177, + "num_tokens": 494596979.0, + "step": 12742 + }, + { + "epoch": 1.6210405800788705, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.188679695129395, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8790940046310425, + "num_tokens": 494634612.0, + "step": 12743 + }, + { + "epoch": 1.6211677903574608, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.17929744720459, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.882593035697937, + "num_tokens": 494672111.0, + "step": 12744 + }, + { + "epoch": 1.6212950006360514, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.248478889465332, + "learning_rate": 1e-06, + "loss": 0.3654, + "mean_token_accuracy": 0.874850332736969, + "num_tokens": 494702955.0, + "step": 12745 + }, + { + "epoch": 1.621422210914642, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.054546356201172, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8977352380752563, + "num_tokens": 494747428.0, + "step": 12746 + }, + { + "epoch": 1.6215494211932324, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.111831665039062, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8865143060684204, + "num_tokens": 494790020.0, + "step": 12747 + }, + { + "epoch": 1.621676631471823, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.133196830749512, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8931293487548828, + "num_tokens": 494828377.0, + "step": 12748 + }, + { + "epoch": 1.6218038417504135, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.199204444885254, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8890777826309204, + "num_tokens": 494864173.0, + "step": 12749 + }, + { + "epoch": 1.6219310520290038, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.091333389282227, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.878275990486145, + "num_tokens": 494908511.0, + "step": 12750 + }, + { + "epoch": 1.6220582623075943, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.213805198669434, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8792693018913269, + "num_tokens": 494944268.0, + "step": 12751 + }, + { + "epoch": 1.6221854725861848, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.160283088684082, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8934508562088013, + "num_tokens": 494980416.0, + "step": 12752 + }, + { + "epoch": 1.6223126828647754, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.055871963500977, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8911039233207703, + "num_tokens": 495023042.0, + "step": 12753 + }, + { + "epoch": 1.622439893143366, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.206209182739258, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8952778577804565, + "num_tokens": 495060170.0, + "step": 12754 + }, + { + "epoch": 1.6225671034219564, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.175350189208984, + "learning_rate": 1e-06, + "loss": 0.3724, + "mean_token_accuracy": 0.870821475982666, + "num_tokens": 495097868.0, + "step": 12755 + }, + { + "epoch": 1.622694313700547, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.174386024475098, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.9007744789123535, + "num_tokens": 495134087.0, + "step": 12756 + }, + { + "epoch": 1.6228215239791375, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.22165298461914, + "learning_rate": 1e-06, + "loss": 0.3895, + "mean_token_accuracy": 0.872194230556488, + "num_tokens": 495173807.0, + "step": 12757 + }, + { + "epoch": 1.622948734257728, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.142454147338867, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8951618671417236, + "num_tokens": 495211560.0, + "step": 12758 + }, + { + "epoch": 1.6230759445363185, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.223292350769043, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8836565613746643, + "num_tokens": 495244472.0, + "step": 12759 + }, + { + "epoch": 1.623203154814909, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.23891830444336, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8819079399108887, + "num_tokens": 495278262.0, + "step": 12760 + }, + { + "epoch": 1.6233303650934996, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.122915267944336, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8763384222984314, + "num_tokens": 495317786.0, + "step": 12761 + }, + { + "epoch": 1.6234575753720901, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.022607803344727, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.897042989730835, + "num_tokens": 495360108.0, + "step": 12762 + }, + { + "epoch": 1.6235847856506807, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.192215919494629, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8937518000602722, + "num_tokens": 495397330.0, + "step": 12763 + }, + { + "epoch": 1.6237119959292712, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.12061882019043, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8787871599197388, + "num_tokens": 495437850.0, + "step": 12764 + }, + { + "epoch": 1.6238392062078617, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.103721618652344, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.9032570123672485, + "num_tokens": 495478794.0, + "step": 12765 + }, + { + "epoch": 1.6239664164864522, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.22960090637207, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8710619211196899, + "num_tokens": 495511139.0, + "step": 12766 + }, + { + "epoch": 1.6240936267650428, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.127103805541992, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.901250958442688, + "num_tokens": 495547221.0, + "step": 12767 + }, + { + "epoch": 1.6242208370436333, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.057852745056152, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8861013054847717, + "num_tokens": 495590187.0, + "step": 12768 + }, + { + "epoch": 1.6243480473222236, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.221449851989746, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8970207571983337, + "num_tokens": 495622271.0, + "step": 12769 + }, + { + "epoch": 1.6244752576008141, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.099093437194824, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8911792039871216, + "num_tokens": 495664962.0, + "step": 12770 + }, + { + "epoch": 1.6246024678794047, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.132619857788086, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8906373381614685, + "num_tokens": 495702583.0, + "step": 12771 + }, + { + "epoch": 1.6247296781579952, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.2014741897583, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8787130117416382, + "num_tokens": 495742362.0, + "step": 12772 + }, + { + "epoch": 1.6248568884365857, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.150583267211914, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.877348005771637, + "num_tokens": 495786733.0, + "step": 12773 + }, + { + "epoch": 1.6249840987151762, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.085770606994629, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8927288055419922, + "num_tokens": 495828407.0, + "step": 12774 + }, + { + "epoch": 1.6251113089937665, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.224713325500488, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8902150392532349, + "num_tokens": 495859425.0, + "step": 12775 + }, + { + "epoch": 1.625238519272357, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.164992332458496, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8763366937637329, + "num_tokens": 495896716.0, + "step": 12776 + }, + { + "epoch": 1.6253657295509476, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.113466262817383, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8893280029296875, + "num_tokens": 495937805.0, + "step": 12777 + }, + { + "epoch": 1.6254929398295381, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.153422355651855, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.88961261510849, + "num_tokens": 495974641.0, + "step": 12778 + }, + { + "epoch": 1.6256201501081287, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.133127212524414, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8967618942260742, + "num_tokens": 496016013.0, + "step": 12779 + }, + { + "epoch": 1.6257473603867192, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.192171096801758, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8911863565444946, + "num_tokens": 496048302.0, + "step": 12780 + }, + { + "epoch": 1.6258745706653097, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.152225494384766, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8878568410873413, + "num_tokens": 496087144.0, + "step": 12781 + }, + { + "epoch": 1.6260017809439002, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.147005081176758, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.894188404083252, + "num_tokens": 496125479.0, + "step": 12782 + }, + { + "epoch": 1.6261289912224908, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.160966873168945, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8926964998245239, + "num_tokens": 496163023.0, + "step": 12783 + }, + { + "epoch": 1.6262562015010813, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.060294151306152, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8886843323707581, + "num_tokens": 496208480.0, + "step": 12784 + }, + { + "epoch": 1.6263834117796718, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.088837623596191, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8860926628112793, + "num_tokens": 496250096.0, + "step": 12785 + }, + { + "epoch": 1.6265106220582624, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.216024398803711, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8882755041122437, + "num_tokens": 496285900.0, + "step": 12786 + }, + { + "epoch": 1.6266378323368529, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.084160804748535, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.89421147108078, + "num_tokens": 496322408.0, + "step": 12787 + }, + { + "epoch": 1.6267650426154434, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.2150239944458, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8827899694442749, + "num_tokens": 496358594.0, + "step": 12788 + }, + { + "epoch": 1.626892252894034, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.026949882507324, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8817059993743896, + "num_tokens": 496405053.0, + "step": 12789 + }, + { + "epoch": 1.6270194631726245, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.19968318939209, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8825274705886841, + "num_tokens": 496441688.0, + "step": 12790 + }, + { + "epoch": 1.627146673451215, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.220732688903809, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8806461095809937, + "num_tokens": 496477550.0, + "step": 12791 + }, + { + "epoch": 1.6272738837298055, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.109431266784668, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8900540471076965, + "num_tokens": 496518321.0, + "step": 12792 + }, + { + "epoch": 1.6274010940083958, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.163677215576172, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8942751884460449, + "num_tokens": 496553839.0, + "step": 12793 + }, + { + "epoch": 1.6275283042869864, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.186681747436523, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.890313982963562, + "num_tokens": 496593381.0, + "step": 12794 + }, + { + "epoch": 1.6276555145655769, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.112743377685547, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8955248594284058, + "num_tokens": 496631081.0, + "step": 12795 + }, + { + "epoch": 1.6277827248441674, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.201366424560547, + "learning_rate": 1e-06, + "loss": 0.3597, + "mean_token_accuracy": 0.8741164207458496, + "num_tokens": 496669036.0, + "step": 12796 + }, + { + "epoch": 1.627909935122758, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.158929824829102, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8883981704711914, + "num_tokens": 496706834.0, + "step": 12797 + }, + { + "epoch": 1.6280371454013485, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.179177284240723, + "learning_rate": 1e-06, + "loss": 0.3736, + "mean_token_accuracy": 0.8722692728042603, + "num_tokens": 496747953.0, + "step": 12798 + }, + { + "epoch": 1.6281643556799388, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.261558532714844, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.895531177520752, + "num_tokens": 496783722.0, + "step": 12799 + }, + { + "epoch": 1.6282915659585293, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.112767219543457, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8898054361343384, + "num_tokens": 496825525.0, + "step": 12800 + }, + { + "epoch": 1.6284187762371198, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.145723342895508, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8943184614181519, + "num_tokens": 496869137.0, + "step": 12801 + }, + { + "epoch": 1.6285459865157104, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.152791976928711, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.89808189868927, + "num_tokens": 496911082.0, + "step": 12802 + }, + { + "epoch": 1.628673196794301, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.184906959533691, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8881567716598511, + "num_tokens": 496951545.0, + "step": 12803 + }, + { + "epoch": 1.6288004070728914, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.197678565979004, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8912367820739746, + "num_tokens": 496989861.0, + "step": 12804 + }, + { + "epoch": 1.628927617351482, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.155324935913086, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8890637159347534, + "num_tokens": 497035379.0, + "step": 12805 + }, + { + "epoch": 1.6290548276300725, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.232134819030762, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8868755102157593, + "num_tokens": 497073198.0, + "step": 12806 + }, + { + "epoch": 1.629182037908663, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.171504020690918, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.902285099029541, + "num_tokens": 497115066.0, + "step": 12807 + }, + { + "epoch": 1.6293092481872535, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.176795959472656, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8931496143341064, + "num_tokens": 497152940.0, + "step": 12808 + }, + { + "epoch": 1.629436458465844, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.237966537475586, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8973020911216736, + "num_tokens": 497183046.0, + "step": 12809 + }, + { + "epoch": 1.6295636687444346, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.102469444274902, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8910051584243774, + "num_tokens": 497220644.0, + "step": 12810 + }, + { + "epoch": 1.6296908790230251, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.179771423339844, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8958719968795776, + "num_tokens": 497257984.0, + "step": 12811 + }, + { + "epoch": 1.6298180893016156, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.195403099060059, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8851776719093323, + "num_tokens": 497292242.0, + "step": 12812 + }, + { + "epoch": 1.6299452995802062, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.118284225463867, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8911503553390503, + "num_tokens": 497330877.0, + "step": 12813 + }, + { + "epoch": 1.6300725098587967, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.187252044677734, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8774381875991821, + "num_tokens": 497371796.0, + "step": 12814 + }, + { + "epoch": 1.6301997201373872, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.172207832336426, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8897725343704224, + "num_tokens": 497411546.0, + "step": 12815 + }, + { + "epoch": 1.6303269304159778, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.129820823669434, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8849552869796753, + "num_tokens": 497449663.0, + "step": 12816 + }, + { + "epoch": 1.630454140694568, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.164449691772461, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8978805541992188, + "num_tokens": 497486910.0, + "step": 12817 + }, + { + "epoch": 1.6305813509731586, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.156862258911133, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8751392364501953, + "num_tokens": 497531504.0, + "step": 12818 + }, + { + "epoch": 1.6307085612517491, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.175148963928223, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.878201425075531, + "num_tokens": 497575150.0, + "step": 12819 + }, + { + "epoch": 1.6308357715303397, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.246514320373535, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8843411803245544, + "num_tokens": 497610352.0, + "step": 12820 + }, + { + "epoch": 1.6309629818089302, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.083328247070312, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8920803666114807, + "num_tokens": 497653764.0, + "step": 12821 + }, + { + "epoch": 1.6310901920875207, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.044504165649414, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.891063392162323, + "num_tokens": 497701571.0, + "step": 12822 + }, + { + "epoch": 1.6312174023661112, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.147320747375488, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8989566564559937, + "num_tokens": 497742522.0, + "step": 12823 + }, + { + "epoch": 1.6313446126447015, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.182053565979004, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.887862503528595, + "num_tokens": 497780213.0, + "step": 12824 + }, + { + "epoch": 1.631471822923292, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.20324420928955, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8803091049194336, + "num_tokens": 497818677.0, + "step": 12825 + }, + { + "epoch": 1.6315990332018826, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.247681617736816, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.8696056008338928, + "num_tokens": 497857607.0, + "step": 12826 + }, + { + "epoch": 1.6317262434804731, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.143685340881348, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8892690539360046, + "num_tokens": 497894911.0, + "step": 12827 + }, + { + "epoch": 1.6318534537590637, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.1876859664917, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8817518949508667, + "num_tokens": 497935171.0, + "step": 12828 + }, + { + "epoch": 1.6319806640376542, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.25035285949707, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8686540126800537, + "num_tokens": 497971488.0, + "step": 12829 + }, + { + "epoch": 1.6321078743162447, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.071185111999512, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8938939571380615, + "num_tokens": 498016414.0, + "step": 12830 + }, + { + "epoch": 1.6322350845948352, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.184636116027832, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8803282380104065, + "num_tokens": 498057844.0, + "step": 12831 + }, + { + "epoch": 1.6323622948734258, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.239272117614746, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8879646062850952, + "num_tokens": 498093214.0, + "step": 12832 + }, + { + "epoch": 1.6324895051520163, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.220162391662598, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8878359794616699, + "num_tokens": 498130076.0, + "step": 12833 + }, + { + "epoch": 1.6326167154306068, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.218809127807617, + "learning_rate": 1e-06, + "loss": 0.4112, + "mean_token_accuracy": 0.8622859120368958, + "num_tokens": 498173992.0, + "step": 12834 + }, + { + "epoch": 1.6327439257091974, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.144641876220703, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8892228603363037, + "num_tokens": 498212673.0, + "step": 12835 + }, + { + "epoch": 1.6328711359877879, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.188899040222168, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8851709961891174, + "num_tokens": 498248001.0, + "step": 12836 + }, + { + "epoch": 1.6329983462663784, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.113579750061035, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8953511714935303, + "num_tokens": 498288368.0, + "step": 12837 + }, + { + "epoch": 1.633125556544969, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.15125846862793, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8923298120498657, + "num_tokens": 498328150.0, + "step": 12838 + }, + { + "epoch": 1.6332527668235595, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.159761428833008, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8950177431106567, + "num_tokens": 498366404.0, + "step": 12839 + }, + { + "epoch": 1.63337997710215, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.129966735839844, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8961331248283386, + "num_tokens": 498405513.0, + "step": 12840 + }, + { + "epoch": 1.6335071873807405, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.211037635803223, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.889156699180603, + "num_tokens": 498440998.0, + "step": 12841 + }, + { + "epoch": 1.6336343976593308, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.91278076171875e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.15167236328125, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9078684449195862, + "num_tokens": 498477279.0, + "step": 12842 + }, + { + "epoch": 1.6337616079379214, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.183979988098145, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8905938863754272, + "num_tokens": 498515065.0, + "step": 12843 + }, + { + "epoch": 1.6338888182165119, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.196839332580566, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8967682123184204, + "num_tokens": 498552456.0, + "step": 12844 + }, + { + "epoch": 1.6340160284951024, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.123517990112305, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8935019969940186, + "num_tokens": 498590155.0, + "step": 12845 + }, + { + "epoch": 1.634143238773693, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.067607879638672, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8987077474594116, + "num_tokens": 498635495.0, + "step": 12846 + }, + { + "epoch": 1.6342704490522835, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.272439956665039, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8908050656318665, + "num_tokens": 498668520.0, + "step": 12847 + }, + { + "epoch": 1.6343976593308738, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.264593124389648, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.8809047937393188, + "num_tokens": 498703571.0, + "step": 12848 + }, + { + "epoch": 1.6345248696094643, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.12539291381836, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8981504440307617, + "num_tokens": 498742572.0, + "step": 12849 + }, + { + "epoch": 1.6346520798880548, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.107039451599121, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8910225033760071, + "num_tokens": 498783095.0, + "step": 12850 + }, + { + "epoch": 1.6347792901666454, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.19627571105957, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8862433433532715, + "num_tokens": 498823140.0, + "step": 12851 + }, + { + "epoch": 1.6349065004452359, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.209818840026855, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8989849090576172, + "num_tokens": 498860290.0, + "step": 12852 + }, + { + "epoch": 1.6350337107238264, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.064802169799805, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8997870683670044, + "num_tokens": 498900290.0, + "step": 12853 + }, + { + "epoch": 1.635160921002417, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.171683311462402, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8833916187286377, + "num_tokens": 498937673.0, + "step": 12854 + }, + { + "epoch": 1.6352881312810075, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.271803855895996, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8899979591369629, + "num_tokens": 498971612.0, + "step": 12855 + }, + { + "epoch": 1.635415341559598, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.185312271118164, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8793388605117798, + "num_tokens": 499009725.0, + "step": 12856 + }, + { + "epoch": 1.6355425518381885, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.157693862915039, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9010996222496033, + "num_tokens": 499042227.0, + "step": 12857 + }, + { + "epoch": 1.635669762116779, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.255885124206543, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.8769979476928711, + "num_tokens": 499081831.0, + "step": 12858 + }, + { + "epoch": 1.6357969723953696, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.185921669006348, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8796554207801819, + "num_tokens": 499118166.0, + "step": 12859 + }, + { + "epoch": 1.6359241826739601, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.082317352294922, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8970017433166504, + "num_tokens": 499159565.0, + "step": 12860 + }, + { + "epoch": 1.6360513929525506, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.145576477050781, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8795509934425354, + "num_tokens": 499200700.0, + "step": 12861 + }, + { + "epoch": 1.6361786032311412, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.143109321594238, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8949233293533325, + "num_tokens": 499238772.0, + "step": 12862 + }, + { + "epoch": 1.6363058135097317, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.145437240600586, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8907703757286072, + "num_tokens": 499276687.0, + "step": 12863 + }, + { + "epoch": 1.6364330237883222, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.194947242736816, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8847560882568359, + "num_tokens": 499318162.0, + "step": 12864 + }, + { + "epoch": 1.6365602340669128, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.175888061523438, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8738412857055664, + "num_tokens": 499358805.0, + "step": 12865 + }, + { + "epoch": 1.636687444345503, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.206147193908691, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8842421770095825, + "num_tokens": 499399001.0, + "step": 12866 + }, + { + "epoch": 1.6368146546240936, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.131202697753906, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8951736092567444, + "num_tokens": 499442246.0, + "step": 12867 + }, + { + "epoch": 1.6369418649026841, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.144450187683105, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8871630430221558, + "num_tokens": 499484508.0, + "step": 12868 + }, + { + "epoch": 1.6370690751812746, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.352871894836426, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8819184899330139, + "num_tokens": 499516414.0, + "step": 12869 + }, + { + "epoch": 1.6371962854598652, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.192821502685547, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8825300931930542, + "num_tokens": 499558977.0, + "step": 12870 + }, + { + "epoch": 1.6373234957384557, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.936622619628906e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.283148765563965, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8795909881591797, + "num_tokens": 499593407.0, + "step": 12871 + }, + { + "epoch": 1.6374507060170462, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.156490325927734, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8849600553512573, + "num_tokens": 499632722.0, + "step": 12872 + }, + { + "epoch": 1.6375779162956365, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.189550399780273, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8863012790679932, + "num_tokens": 499667989.0, + "step": 12873 + }, + { + "epoch": 1.637705126574227, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.226410865783691, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9023088216781616, + "num_tokens": 499700259.0, + "step": 12874 + }, + { + "epoch": 1.6378323368528176, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.098237037658691, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8896023035049438, + "num_tokens": 499739242.0, + "step": 12875 + }, + { + "epoch": 1.6379595471314081, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.141167640686035, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8890897631645203, + "num_tokens": 499780646.0, + "step": 12876 + }, + { + "epoch": 1.6380867574099987, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.146005630493164, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.881591260433197, + "num_tokens": 499819871.0, + "step": 12877 + }, + { + "epoch": 1.6382139676885892, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.16833782196045, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8838866949081421, + "num_tokens": 499859762.0, + "step": 12878 + }, + { + "epoch": 1.6383411779671797, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.158616065979004, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.889238178730011, + "num_tokens": 499896557.0, + "step": 12879 + }, + { + "epoch": 1.6384683882457702, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.152571678161621, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8858470916748047, + "num_tokens": 499937854.0, + "step": 12880 + }, + { + "epoch": 1.6385955985243608, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.150102615356445, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.899661123752594, + "num_tokens": 499974183.0, + "step": 12881 + }, + { + "epoch": 1.6387228088029513, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.168525695800781, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8942890167236328, + "num_tokens": 500013224.0, + "step": 12882 + }, + { + "epoch": 1.6388500190815418, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.211614608764648, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8825494050979614, + "num_tokens": 500052644.0, + "step": 12883 + }, + { + "epoch": 1.6389772293601323, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.154191970825195, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8880560398101807, + "num_tokens": 500089276.0, + "step": 12884 + }, + { + "epoch": 1.6391044396387229, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.087983131408691, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8924276828765869, + "num_tokens": 500132314.0, + "step": 12885 + }, + { + "epoch": 1.6392316499173134, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.129093170166016, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8881300687789917, + "num_tokens": 500173049.0, + "step": 12886 + }, + { + "epoch": 1.639358860195904, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.10280990600586, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8828014731407166, + "num_tokens": 500220339.0, + "step": 12887 + }, + { + "epoch": 1.6394860704744945, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.152734756469727, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8869247436523438, + "num_tokens": 500260803.0, + "step": 12888 + }, + { + "epoch": 1.639613280753085, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.151388168334961, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8922505378723145, + "num_tokens": 500302127.0, + "step": 12889 + }, + { + "epoch": 1.6397404910316755, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.158696174621582, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9056822061538696, + "num_tokens": 500340030.0, + "step": 12890 + }, + { + "epoch": 1.6398677013102658, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.177001953125, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8824601173400879, + "num_tokens": 500384897.0, + "step": 12891 + }, + { + "epoch": 1.6399949115888564, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.298861503601074, + "learning_rate": 1e-06, + "loss": 0.399, + "mean_token_accuracy": 0.8682690858840942, + "num_tokens": 500419407.0, + "step": 12892 + }, + { + "epoch": 1.6401221218674469, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.137600898742676, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8855992555618286, + "num_tokens": 500459275.0, + "step": 12893 + }, + { + "epoch": 1.6402493321460374, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.185650825500488, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8955117464065552, + "num_tokens": 500492884.0, + "step": 12894 + }, + { + "epoch": 1.640376542424628, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.225680351257324, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8874449729919434, + "num_tokens": 500528859.0, + "step": 12895 + }, + { + "epoch": 1.6405037527032185, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.10435676574707, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8823798894882202, + "num_tokens": 500573251.0, + "step": 12896 + }, + { + "epoch": 1.6406309629818088, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.161340713500977, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8811887502670288, + "num_tokens": 500613769.0, + "step": 12897 + }, + { + "epoch": 1.6407581732603993, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.254934310913086, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8841103911399841, + "num_tokens": 500647584.0, + "step": 12898 + }, + { + "epoch": 1.6408853835389898, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.139386177062988, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8872702121734619, + "num_tokens": 500689380.0, + "step": 12899 + }, + { + "epoch": 1.6410125938175804, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.170795440673828, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8862529993057251, + "num_tokens": 500730017.0, + "step": 12900 + }, + { + "epoch": 1.6411398040961709, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.22512435913086, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8910799026489258, + "num_tokens": 500766420.0, + "step": 12901 + }, + { + "epoch": 1.6412670143747614, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.262887001037598, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8872459530830383, + "num_tokens": 500802214.0, + "step": 12902 + }, + { + "epoch": 1.641394224653352, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.110294342041016, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.9002149105072021, + "num_tokens": 500844542.0, + "step": 12903 + }, + { + "epoch": 1.6415214349319425, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.196147918701172, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8813893795013428, + "num_tokens": 500885968.0, + "step": 12904 + }, + { + "epoch": 1.641648645210533, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.272848129272461, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8914451599121094, + "num_tokens": 500917772.0, + "step": 12905 + }, + { + "epoch": 1.6417758554891235, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.244038581848145, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8901669383049011, + "num_tokens": 500950648.0, + "step": 12906 + }, + { + "epoch": 1.641903065767714, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.131726264953613, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8878017663955688, + "num_tokens": 500992038.0, + "step": 12907 + }, + { + "epoch": 1.6420302760463046, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.203073501586914, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.878786563873291, + "num_tokens": 501030537.0, + "step": 12908 + }, + { + "epoch": 1.642157486324895, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.194669723510742, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.891098141670227, + "num_tokens": 501065339.0, + "step": 12909 + }, + { + "epoch": 1.6422846966034856, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.154840469360352, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.9084770083427429, + "num_tokens": 501097053.0, + "step": 12910 + }, + { + "epoch": 1.6424119068820762, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.126463890075684, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8949106335639954, + "num_tokens": 501136449.0, + "step": 12911 + }, + { + "epoch": 1.6425391171606667, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.162420272827148, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8950733542442322, + "num_tokens": 501171341.0, + "step": 12912 + }, + { + "epoch": 1.6426663274392572, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.18306827545166, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.885124683380127, + "num_tokens": 501217020.0, + "step": 12913 + }, + { + "epoch": 1.6427935377178478, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.162271499633789, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8870378732681274, + "num_tokens": 501252932.0, + "step": 12914 + }, + { + "epoch": 1.642920747996438, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.107810974121094, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8933496475219727, + "num_tokens": 501294712.0, + "step": 12915 + }, + { + "epoch": 1.6430479582750286, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.194117546081543, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8894073367118835, + "num_tokens": 501334042.0, + "step": 12916 + }, + { + "epoch": 1.6431751685536191, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.147804260253906, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8832558393478394, + "num_tokens": 501373295.0, + "step": 12917 + }, + { + "epoch": 1.6433023788322096, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.424099922180176, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8759041428565979, + "num_tokens": 501400599.0, + "step": 12918 + }, + { + "epoch": 1.6434295891108002, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.186657905578613, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8911772966384888, + "num_tokens": 501433520.0, + "step": 12919 + }, + { + "epoch": 1.6435567993893907, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.148765563964844, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.887114405632019, + "num_tokens": 501473239.0, + "step": 12920 + }, + { + "epoch": 1.6436840096679812, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.234108924865723, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8970866203308105, + "num_tokens": 501507007.0, + "step": 12921 + }, + { + "epoch": 1.6438112199465715, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.160507202148438, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8875554203987122, + "num_tokens": 501547678.0, + "step": 12922 + }, + { + "epoch": 1.643938430225162, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.19340991973877, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.878208577632904, + "num_tokens": 501591817.0, + "step": 12923 + }, + { + "epoch": 1.6440656405037526, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.16588020324707, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8835001587867737, + "num_tokens": 501625514.0, + "step": 12924 + }, + { + "epoch": 1.6441928507823431, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.088491439819336, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.9031640291213989, + "num_tokens": 501670572.0, + "step": 12925 + }, + { + "epoch": 1.6443200610609336, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.126051902770996, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8949622511863708, + "num_tokens": 501706899.0, + "step": 12926 + }, + { + "epoch": 1.6444472713395242, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.13669204711914, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9084392189979553, + "num_tokens": 501746466.0, + "step": 12927 + }, + { + "epoch": 1.6445744816181147, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.171300888061523, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8932704925537109, + "num_tokens": 501784334.0, + "step": 12928 + }, + { + "epoch": 1.6447016918967052, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.201947212219238, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8895007371902466, + "num_tokens": 501822076.0, + "step": 12929 + }, + { + "epoch": 1.6448289021752958, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.248503684997559, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8735221028327942, + "num_tokens": 501862076.0, + "step": 12930 + }, + { + "epoch": 1.6449561124538863, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.141490936279297, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8895277380943298, + "num_tokens": 501901716.0, + "step": 12931 + }, + { + "epoch": 1.6450833227324768, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.102575302124023, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8918951749801636, + "num_tokens": 501943855.0, + "step": 12932 + }, + { + "epoch": 1.6452105330110673, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.152432441711426, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8777800798416138, + "num_tokens": 501992128.0, + "step": 12933 + }, + { + "epoch": 1.6453377432896579, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.201709747314453, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8930577635765076, + "num_tokens": 502027132.0, + "step": 12934 + }, + { + "epoch": 1.6454649535682484, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.191496849060059, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8893284797668457, + "num_tokens": 502063169.0, + "step": 12935 + }, + { + "epoch": 1.645592163846839, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.0935697555542, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8907078504562378, + "num_tokens": 502103684.0, + "step": 12936 + }, + { + "epoch": 1.6457193741254295, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.183631896972656, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8881136178970337, + "num_tokens": 502139894.0, + "step": 12937 + }, + { + "epoch": 1.64584658440402, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.26059341430664, + "learning_rate": 1e-06, + "loss": 0.4002, + "mean_token_accuracy": 0.8620054721832275, + "num_tokens": 502174898.0, + "step": 12938 + }, + { + "epoch": 1.6459737946826105, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.167510986328125, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8943310379981995, + "num_tokens": 502210244.0, + "step": 12939 + }, + { + "epoch": 1.6461010049612008, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.068341255187988, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8871020078659058, + "num_tokens": 502254357.0, + "step": 12940 + }, + { + "epoch": 1.6462282152397913, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.153059005737305, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8868957757949829, + "num_tokens": 502297611.0, + "step": 12941 + }, + { + "epoch": 1.6463554255183819, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.085749626159668, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8948330879211426, + "num_tokens": 502345031.0, + "step": 12942 + }, + { + "epoch": 1.6464826357969724, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.115448951721191, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8933876752853394, + "num_tokens": 502384725.0, + "step": 12943 + }, + { + "epoch": 1.646609846075563, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.286059379577637, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8771489858627319, + "num_tokens": 502421840.0, + "step": 12944 + }, + { + "epoch": 1.6467370563541535, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.07451343536377, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.9004175662994385, + "num_tokens": 502462831.0, + "step": 12945 + }, + { + "epoch": 1.6468642666327438, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.158834457397461, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8913103938102722, + "num_tokens": 502504068.0, + "step": 12946 + }, + { + "epoch": 1.6469914769113343, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.466095924377441, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8951867818832397, + "num_tokens": 502548588.0, + "step": 12947 + }, + { + "epoch": 1.6471186871899248, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.124208450317383, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8914052248001099, + "num_tokens": 502585411.0, + "step": 12948 + }, + { + "epoch": 1.6472458974685154, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.234609603881836, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8885607719421387, + "num_tokens": 502616720.0, + "step": 12949 + }, + { + "epoch": 1.6473731077471059, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.184539794921875, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8806648254394531, + "num_tokens": 502657160.0, + "step": 12950 + }, + { + "epoch": 1.6475003180256964, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.178694725036621, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8908781409263611, + "num_tokens": 502689187.0, + "step": 12951 + }, + { + "epoch": 1.647627528304287, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.195948600769043, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.885720431804657, + "num_tokens": 502728687.0, + "step": 12952 + }, + { + "epoch": 1.6477547385828775, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.02487850189209, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8926231861114502, + "num_tokens": 502769885.0, + "step": 12953 + }, + { + "epoch": 1.647881948861468, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.255145072937012, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8748384714126587, + "num_tokens": 502806707.0, + "step": 12954 + }, + { + "epoch": 1.6480091591400585, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.098289489746094, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8922841548919678, + "num_tokens": 502842965.0, + "step": 12955 + }, + { + "epoch": 1.648136369418649, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.195354461669922, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8893483877182007, + "num_tokens": 502878133.0, + "step": 12956 + }, + { + "epoch": 1.6482635796972396, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.123313903808594, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8988230228424072, + "num_tokens": 502915187.0, + "step": 12957 + }, + { + "epoch": 1.64839078997583, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 5.984306335449219e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.089030265808105, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8876656293869019, + "num_tokens": 502956103.0, + "step": 12958 + }, + { + "epoch": 1.6485180002544206, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.159943580627441, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8920785188674927, + "num_tokens": 502992281.0, + "step": 12959 + }, + { + "epoch": 1.6486452105330112, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.045778274536133, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8875815868377686, + "num_tokens": 503038209.0, + "step": 12960 + }, + { + "epoch": 1.6487724208116017, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.139321327209473, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8800457119941711, + "num_tokens": 503084131.0, + "step": 12961 + }, + { + "epoch": 1.6488996310901922, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.316646575927734, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8784735202789307, + "num_tokens": 503115701.0, + "step": 12962 + }, + { + "epoch": 1.6490268413687827, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.176855087280273, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8901728987693787, + "num_tokens": 503151883.0, + "step": 12963 + }, + { + "epoch": 1.649154051647373, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.102401733398438, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8893181085586548, + "num_tokens": 503199930.0, + "step": 12964 + }, + { + "epoch": 1.6492812619259636, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.193127632141113, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8723219633102417, + "num_tokens": 503240095.0, + "step": 12965 + }, + { + "epoch": 1.649408472204554, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.170546531677246, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8859438896179199, + "num_tokens": 503275607.0, + "step": 12966 + }, + { + "epoch": 1.6495356824831446, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.128908157348633, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.892120361328125, + "num_tokens": 503317522.0, + "step": 12967 + }, + { + "epoch": 1.6496628927617352, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.223697662353516, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8848307132720947, + "num_tokens": 503352332.0, + "step": 12968 + }, + { + "epoch": 1.6497901030403257, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.14869499206543, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8968600034713745, + "num_tokens": 503390354.0, + "step": 12969 + }, + { + "epoch": 1.6499173133189162, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.189103126525879, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8996695280075073, + "num_tokens": 503427984.0, + "step": 12970 + }, + { + "epoch": 1.6500445235975065, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.168596267700195, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8801307678222656, + "num_tokens": 503465958.0, + "step": 12971 + }, + { + "epoch": 1.650171733876097, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.262472152709961, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8815563917160034, + "num_tokens": 503500009.0, + "step": 12972 + }, + { + "epoch": 1.6502989441546876, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.122708320617676, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8856184482574463, + "num_tokens": 503537859.0, + "step": 12973 + }, + { + "epoch": 1.6504261544332781, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.104743957519531, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8904799222946167, + "num_tokens": 503581660.0, + "step": 12974 + }, + { + "epoch": 1.6505533647118686, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.2439603805542, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8844202160835266, + "num_tokens": 503618617.0, + "step": 12975 + }, + { + "epoch": 1.6506805749904592, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.123462677001953, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8861044645309448, + "num_tokens": 503659646.0, + "step": 12976 + }, + { + "epoch": 1.6508077852690497, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.193004608154297, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8881065845489502, + "num_tokens": 503699834.0, + "step": 12977 + }, + { + "epoch": 1.6509349955476402, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.27408504486084, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8842513561248779, + "num_tokens": 503732616.0, + "step": 12978 + }, + { + "epoch": 1.6510622058262308, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.281621932983398, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.885708212852478, + "num_tokens": 503771039.0, + "step": 12979 + }, + { + "epoch": 1.6511894161048213, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.17657470703125, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8970319628715515, + "num_tokens": 503809002.0, + "step": 12980 + }, + { + "epoch": 1.6513166263834118, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.267770767211914, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8859687447547913, + "num_tokens": 503840845.0, + "step": 12981 + }, + { + "epoch": 1.6514438366620023, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.164972305297852, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8927492499351501, + "num_tokens": 503882839.0, + "step": 12982 + }, + { + "epoch": 1.6515710469405929, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.160188674926758, + "learning_rate": 1e-06, + "loss": 0.3475, + "mean_token_accuracy": 0.8801772594451904, + "num_tokens": 503922112.0, + "step": 12983 + }, + { + "epoch": 1.6516982572191834, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.28515625, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8946236371994019, + "num_tokens": 503960113.0, + "step": 12984 + }, + { + "epoch": 1.651825467497774, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.099339485168457, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8949307203292847, + "num_tokens": 504003407.0, + "step": 12985 + }, + { + "epoch": 1.6519526777763645, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.185040473937988, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8790892362594604, + "num_tokens": 504046741.0, + "step": 12986 + }, + { + "epoch": 1.652079888054955, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.224614143371582, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8902357220649719, + "num_tokens": 504084514.0, + "step": 12987 + }, + { + "epoch": 1.6522070983335455, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.095861434936523, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8936202526092529, + "num_tokens": 504123746.0, + "step": 12988 + }, + { + "epoch": 1.6523343086121358, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.123075485229492, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8857206702232361, + "num_tokens": 504163194.0, + "step": 12989 + }, + { + "epoch": 1.6524615188907263, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.170393943786621, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8791244626045227, + "num_tokens": 504203853.0, + "step": 12990 + }, + { + "epoch": 1.6525887291693169, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.191215515136719, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8843168616294861, + "num_tokens": 504241148.0, + "step": 12991 + }, + { + "epoch": 1.6527159394479074, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.130430221557617, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8972321152687073, + "num_tokens": 504276689.0, + "step": 12992 + }, + { + "epoch": 1.652843149726498, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.229558944702148, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8801032304763794, + "num_tokens": 504314590.0, + "step": 12993 + }, + { + "epoch": 1.6529703600050885, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.168012619018555, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8922939300537109, + "num_tokens": 504350518.0, + "step": 12994 + }, + { + "epoch": 1.6530975702836788, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.187833786010742, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8743211627006531, + "num_tokens": 504386944.0, + "step": 12995 + }, + { + "epoch": 1.6532247805622693, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.226770401000977, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8894839286804199, + "num_tokens": 504421347.0, + "step": 12996 + }, + { + "epoch": 1.6533519908408598, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.095513343811035, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8914154767990112, + "num_tokens": 504460516.0, + "step": 12997 + }, + { + "epoch": 1.6534792011194503, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.174586296081543, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.875140905380249, + "num_tokens": 504501332.0, + "step": 12998 + }, + { + "epoch": 1.6536064113980409, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.051517486572266, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9049710035324097, + "num_tokens": 504545996.0, + "step": 12999 + }, + { + "epoch": 1.6537336216766314, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.031990051269531e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.267520904541016, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8781049847602844, + "num_tokens": 504584444.0, + "step": 13000 + }, + { + "epoch": 1.653860831955222, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.198993682861328, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8797351121902466, + "num_tokens": 504621047.0, + "step": 13001 + }, + { + "epoch": 1.6539880422338125, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.12879753112793, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8918954133987427, + "num_tokens": 504665941.0, + "step": 13002 + }, + { + "epoch": 1.654115252512403, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.211319923400879, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8851798176765442, + "num_tokens": 504703244.0, + "step": 13003 + }, + { + "epoch": 1.6542424627909935, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.212417602539062, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8785769939422607, + "num_tokens": 504740602.0, + "step": 13004 + }, + { + "epoch": 1.654369673069584, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.122751235961914, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8857118487358093, + "num_tokens": 504779775.0, + "step": 13005 + }, + { + "epoch": 1.6544968833481746, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.302788734436035, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8825260400772095, + "num_tokens": 504815505.0, + "step": 13006 + }, + { + "epoch": 1.654624093626765, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.065776824951172, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8948729038238525, + "num_tokens": 504857714.0, + "step": 13007 + }, + { + "epoch": 1.6547513039053556, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.178442001342773, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8838282227516174, + "num_tokens": 504898566.0, + "step": 13008 + }, + { + "epoch": 1.6548785141839462, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.245427131652832, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8775118589401245, + "num_tokens": 504934578.0, + "step": 13009 + }, + { + "epoch": 1.6550057244625367, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.16690731048584, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8761728405952454, + "num_tokens": 504972809.0, + "step": 13010 + }, + { + "epoch": 1.6551329347411272, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.218245506286621, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8810288906097412, + "num_tokens": 505014582.0, + "step": 13011 + }, + { + "epoch": 1.6552601450197177, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.071954727172852, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8990229368209839, + "num_tokens": 505053927.0, + "step": 13012 + }, + { + "epoch": 1.655387355298308, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.162841796875, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8869622945785522, + "num_tokens": 505090768.0, + "step": 13013 + }, + { + "epoch": 1.6555145655768986, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.17794418334961, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8868977427482605, + "num_tokens": 505126240.0, + "step": 13014 + }, + { + "epoch": 1.655641775855489, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.057415962219238, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8986087441444397, + "num_tokens": 505169692.0, + "step": 13015 + }, + { + "epoch": 1.6557689861340796, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.390071868896484, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8869611024856567, + "num_tokens": 505200457.0, + "step": 13016 + }, + { + "epoch": 1.6558961964126702, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.240983009338379, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8846423625946045, + "num_tokens": 505235485.0, + "step": 13017 + }, + { + "epoch": 1.6560234066912607, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.0556058883667, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8869494199752808, + "num_tokens": 505279111.0, + "step": 13018 + }, + { + "epoch": 1.6561506169698512, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.345686912536621, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8988771438598633, + "num_tokens": 505308189.0, + "step": 13019 + }, + { + "epoch": 1.6562778272484415, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.322014808654785, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8724839091300964, + "num_tokens": 505341615.0, + "step": 13020 + }, + { + "epoch": 1.656405037527032, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.109487533569336, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.9002135396003723, + "num_tokens": 505383001.0, + "step": 13021 + }, + { + "epoch": 1.6565322478056226, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.242229461669922, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8846626281738281, + "num_tokens": 505414204.0, + "step": 13022 + }, + { + "epoch": 1.656659458084213, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.199435234069824, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8855544328689575, + "num_tokens": 505452184.0, + "step": 13023 + }, + { + "epoch": 1.6567866683628036, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.264456748962402, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8816330432891846, + "num_tokens": 505488806.0, + "step": 13024 + }, + { + "epoch": 1.6569138786413942, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.231512069702148, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8992886543273926, + "num_tokens": 505519505.0, + "step": 13025 + }, + { + "epoch": 1.6570410889199847, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.209895133972168, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8915401697158813, + "num_tokens": 505554988.0, + "step": 13026 + }, + { + "epoch": 1.6571682991985752, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.334261894226074, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8875783681869507, + "num_tokens": 505584962.0, + "step": 13027 + }, + { + "epoch": 1.6572955094771658, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.137002944946289, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8913982510566711, + "num_tokens": 505630854.0, + "step": 13028 + }, + { + "epoch": 1.6574227197557563, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.095970153808594, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8988543748855591, + "num_tokens": 505667668.0, + "step": 13029 + }, + { + "epoch": 1.6575499300343468, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.245545387268066, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8789244294166565, + "num_tokens": 505704526.0, + "step": 13030 + }, + { + "epoch": 1.6576771403129373, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.14035701751709, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8906934261322021, + "num_tokens": 505750341.0, + "step": 13031 + }, + { + "epoch": 1.6578043505915279, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.136210441589355, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8854572772979736, + "num_tokens": 505791633.0, + "step": 13032 + }, + { + "epoch": 1.6579315608701184, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.340447425842285, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8925097584724426, + "num_tokens": 505827113.0, + "step": 13033 + }, + { + "epoch": 1.658058771148709, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.120498657226562, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8983014225959778, + "num_tokens": 505867772.0, + "step": 13034 + }, + { + "epoch": 1.6581859814272994, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.330209732055664, + "learning_rate": 1e-06, + "loss": 0.4069, + "mean_token_accuracy": 0.8627792596817017, + "num_tokens": 505906221.0, + "step": 13035 + }, + { + "epoch": 1.65831319170589, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.171468734741211, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8853319883346558, + "num_tokens": 505947480.0, + "step": 13036 + }, + { + "epoch": 1.6584404019844805, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.232831954956055, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.884263813495636, + "num_tokens": 505985631.0, + "step": 13037 + }, + { + "epoch": 1.6585676122630708, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.213014602661133, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.9030032157897949, + "num_tokens": 506022442.0, + "step": 13038 + }, + { + "epoch": 1.6586948225416613, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.185856819152832, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8918246030807495, + "num_tokens": 506060556.0, + "step": 13039 + }, + { + "epoch": 1.6588220328202519, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.216448783874512, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8798757195472717, + "num_tokens": 506100400.0, + "step": 13040 + }, + { + "epoch": 1.6589492430988424, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.286109924316406, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.882905125617981, + "num_tokens": 506135737.0, + "step": 13041 + }, + { + "epoch": 1.659076453377433, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.206666946411133, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8859959840774536, + "num_tokens": 506171888.0, + "step": 13042 + }, + { + "epoch": 1.6592036636560235, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.20785140991211, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8898269534111023, + "num_tokens": 506212650.0, + "step": 13043 + }, + { + "epoch": 1.6593308739346138, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.206969261169434, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8848462104797363, + "num_tokens": 506254814.0, + "step": 13044 + }, + { + "epoch": 1.6594580842132043, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.187549591064453, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8883222341537476, + "num_tokens": 506296151.0, + "step": 13045 + }, + { + "epoch": 1.6595852944917948, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.235381126403809, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.882987380027771, + "num_tokens": 506335620.0, + "step": 13046 + }, + { + "epoch": 1.6597125047703853, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.19579792022705, + "learning_rate": 1e-06, + "loss": 0.3939, + "mean_token_accuracy": 0.8671772480010986, + "num_tokens": 506377838.0, + "step": 13047 + }, + { + "epoch": 1.6598397150489759, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.126765251159668, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8964812755584717, + "num_tokens": 506422419.0, + "step": 13048 + }, + { + "epoch": 1.6599669253275664, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.233831405639648, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8988121747970581, + "num_tokens": 506455843.0, + "step": 13049 + }, + { + "epoch": 1.660094135606157, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.2584867477417, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8817095160484314, + "num_tokens": 506491680.0, + "step": 13050 + }, + { + "epoch": 1.6602213458847475, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.141940116882324, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8910744190216064, + "num_tokens": 506535011.0, + "step": 13051 + }, + { + "epoch": 1.660348556163338, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.299859046936035, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8823896646499634, + "num_tokens": 506567363.0, + "step": 13052 + }, + { + "epoch": 1.6604757664419285, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.173397064208984, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8973701596260071, + "num_tokens": 506605994.0, + "step": 13053 + }, + { + "epoch": 1.660602976720519, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.26512622833252, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.87929368019104, + "num_tokens": 506645528.0, + "step": 13054 + }, + { + "epoch": 1.6607301869991096, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.1970853805542, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8896430730819702, + "num_tokens": 506678941.0, + "step": 13055 + }, + { + "epoch": 1.6608573972777, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.176793098449707, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.888967752456665, + "num_tokens": 506717745.0, + "step": 13056 + }, + { + "epoch": 1.6609846075562906, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.167033195495605, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8785133361816406, + "num_tokens": 506758843.0, + "step": 13057 + }, + { + "epoch": 1.6611118178348812, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.225866317749023, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8741985559463501, + "num_tokens": 506799994.0, + "step": 13058 + }, + { + "epoch": 1.6612390281134717, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.225381851196289, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8858059644699097, + "num_tokens": 506836822.0, + "step": 13059 + }, + { + "epoch": 1.6613662383920622, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.179473876953125, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8856906890869141, + "num_tokens": 506878001.0, + "step": 13060 + }, + { + "epoch": 1.6614934486706527, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.342185020446777, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8711415529251099, + "num_tokens": 506916752.0, + "step": 13061 + }, + { + "epoch": 1.661620658949243, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.187860488891602, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8866719007492065, + "num_tokens": 506950135.0, + "step": 13062 + }, + { + "epoch": 1.6617478692278336, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.191664695739746, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8890107870101929, + "num_tokens": 506988655.0, + "step": 13063 + }, + { + "epoch": 1.661875079506424, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.150911331176758, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8791441917419434, + "num_tokens": 507034737.0, + "step": 13064 + }, + { + "epoch": 1.6620022897850146, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.179428100585938, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8837451338768005, + "num_tokens": 507076621.0, + "step": 13065 + }, + { + "epoch": 1.6621295000636052, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.192302703857422, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8793171644210815, + "num_tokens": 507118510.0, + "step": 13066 + }, + { + "epoch": 1.6622567103421957, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.254034042358398, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8793978691101074, + "num_tokens": 507157178.0, + "step": 13067 + }, + { + "epoch": 1.662383920620786, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.274680137634277, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8879395723342896, + "num_tokens": 507190086.0, + "step": 13068 + }, + { + "epoch": 1.6625111308993765, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.194441795349121, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8836060762405396, + "num_tokens": 507232166.0, + "step": 13069 + }, + { + "epoch": 1.662638341177967, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.217768669128418, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8746832609176636, + "num_tokens": 507272682.0, + "step": 13070 + }, + { + "epoch": 1.6627655514565576, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.176856994628906, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8833457231521606, + "num_tokens": 507314683.0, + "step": 13071 + }, + { + "epoch": 1.662892761735148, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.213622093200684, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.905524730682373, + "num_tokens": 507350865.0, + "step": 13072 + }, + { + "epoch": 1.6630199720137386, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.082550048828125, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.9011484980583191, + "num_tokens": 507394427.0, + "step": 13073 + }, + { + "epoch": 1.6631471822923292, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.170016288757324, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8885628581047058, + "num_tokens": 507439057.0, + "step": 13074 + }, + { + "epoch": 1.6632743925709197, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.257303237915039, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8776639103889465, + "num_tokens": 507480456.0, + "step": 13075 + }, + { + "epoch": 1.6634016028495102, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.173192024230957, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8972983956336975, + "num_tokens": 507518237.0, + "step": 13076 + }, + { + "epoch": 1.6635288131281007, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.127129554748535, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.887111485004425, + "num_tokens": 507559958.0, + "step": 13077 + }, + { + "epoch": 1.6636560234066913, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.236263275146484, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8877270221710205, + "num_tokens": 507597496.0, + "step": 13078 + }, + { + "epoch": 1.6637832336852818, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.24292278289795, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8824957013130188, + "num_tokens": 507634473.0, + "step": 13079 + }, + { + "epoch": 1.6639104439638723, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.233673095703125, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8913297057151794, + "num_tokens": 507669395.0, + "step": 13080 + }, + { + "epoch": 1.6640376542424629, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.149483680725098, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8870459198951721, + "num_tokens": 507710805.0, + "step": 13081 + }, + { + "epoch": 1.6641648645210534, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.221653938293457, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8781569004058838, + "num_tokens": 507751947.0, + "step": 13082 + }, + { + "epoch": 1.664292074799644, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.196088790893555, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8928848505020142, + "num_tokens": 507792654.0, + "step": 13083 + }, + { + "epoch": 1.6644192850782344, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.156806945800781, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.8717252016067505, + "num_tokens": 507836811.0, + "step": 13084 + }, + { + "epoch": 1.664546495356825, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.24626636505127, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8798988461494446, + "num_tokens": 507875473.0, + "step": 13085 + }, + { + "epoch": 1.6646737056354155, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.09162712097168, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8953968286514282, + "num_tokens": 507917885.0, + "step": 13086 + }, + { + "epoch": 1.6648009159140058, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.22425651550293, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8968825936317444, + "num_tokens": 507953550.0, + "step": 13087 + }, + { + "epoch": 1.6649281261925963, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.204475402832031, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8820667266845703, + "num_tokens": 507992230.0, + "step": 13088 + }, + { + "epoch": 1.6650553364711869, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.132630348205566, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8998744487762451, + "num_tokens": 508028674.0, + "step": 13089 + }, + { + "epoch": 1.6651825467497774, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.182053565979004, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8937771320343018, + "num_tokens": 508064888.0, + "step": 13090 + }, + { + "epoch": 1.665309757028368, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.249319076538086, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8934871554374695, + "num_tokens": 508098933.0, + "step": 13091 + }, + { + "epoch": 1.6654369673069584, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.246689796447754, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8930445313453674, + "num_tokens": 508134875.0, + "step": 13092 + }, + { + "epoch": 1.6655641775855488, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.214693069458008, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8837414979934692, + "num_tokens": 508173116.0, + "step": 13093 + }, + { + "epoch": 1.6656913878641393, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.156432151794434, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.9006229043006897, + "num_tokens": 508209242.0, + "step": 13094 + }, + { + "epoch": 1.6658185981427298, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.272539138793945, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8969149589538574, + "num_tokens": 508242186.0, + "step": 13095 + }, + { + "epoch": 1.6659458084213203, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.121332168579102, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.893143355846405, + "num_tokens": 508285628.0, + "step": 13096 + }, + { + "epoch": 1.6660730186999109, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.162135124206543, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8816715478897095, + "num_tokens": 508332714.0, + "step": 13097 + }, + { + "epoch": 1.6662002289785014, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.232454299926758, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8914187550544739, + "num_tokens": 508367855.0, + "step": 13098 + }, + { + "epoch": 1.666327439257092, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.271504402160645, + "learning_rate": 1e-06, + "loss": 0.3745, + "mean_token_accuracy": 0.8737988471984863, + "num_tokens": 508407400.0, + "step": 13099 + }, + { + "epoch": 1.6664546495356825, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.243260383605957, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8858979940414429, + "num_tokens": 508445503.0, + "step": 13100 + }, + { + "epoch": 1.666581859814273, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.209957122802734, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8728588819503784, + "num_tokens": 508483657.0, + "step": 13101 + }, + { + "epoch": 1.6667090700928635, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.234024047851562, + "learning_rate": 1e-06, + "loss": 0.3854, + "mean_token_accuracy": 0.8703213930130005, + "num_tokens": 508525724.0, + "step": 13102 + }, + { + "epoch": 1.666836280371454, + "ewc_loss": 0.016357421875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0311603546142578e-05, + "grad_norm": 9.166072845458984, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8852146863937378, + "num_tokens": 508561823.0, + "step": 13103 + }, + { + "epoch": 1.6669634906500446, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.182799339294434, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8786158561706543, + "num_tokens": 508598815.0, + "step": 13104 + }, + { + "epoch": 1.667090700928635, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.148606300354004, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8889358043670654, + "num_tokens": 508640359.0, + "step": 13105 + }, + { + "epoch": 1.6672179112072256, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.249712944030762, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.889584481716156, + "num_tokens": 508672822.0, + "step": 13106 + }, + { + "epoch": 1.6673451214858162, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.037527084350586, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.9037872552871704, + "num_tokens": 508716046.0, + "step": 13107 + }, + { + "epoch": 1.6674723317644067, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.263333320617676, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8914557695388794, + "num_tokens": 508749496.0, + "step": 13108 + }, + { + "epoch": 1.6675995420429972, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.226408004760742, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.884016215801239, + "num_tokens": 508784870.0, + "step": 13109 + }, + { + "epoch": 1.6677267523215877, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.149593353271484, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8995062112808228, + "num_tokens": 508820808.0, + "step": 13110 + }, + { + "epoch": 1.667853962600178, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.140204429626465, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.889025092124939, + "num_tokens": 508857430.0, + "step": 13111 + }, + { + "epoch": 1.6679811728787686, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.2736177444458, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.8790088891983032, + "num_tokens": 508897005.0, + "step": 13112 + }, + { + "epoch": 1.668108383157359, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.135032653808594, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8869764804840088, + "num_tokens": 508937513.0, + "step": 13113 + }, + { + "epoch": 1.6682355934359496, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.154460906982422, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8746768236160278, + "num_tokens": 508979627.0, + "step": 13114 + }, + { + "epoch": 1.6683628037145402, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.203230857849121, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.884864330291748, + "num_tokens": 509016147.0, + "step": 13115 + }, + { + "epoch": 1.6684900139931307, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.244681358337402, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8881181478500366, + "num_tokens": 509051187.0, + "step": 13116 + }, + { + "epoch": 1.668617224271721, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.239029884338379, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8861364126205444, + "num_tokens": 509086534.0, + "step": 13117 + }, + { + "epoch": 1.6687444345503115, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.089235305786133, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8894100189208984, + "num_tokens": 509131235.0, + "step": 13118 + }, + { + "epoch": 1.668871644828902, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.143010139465332, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8925586938858032, + "num_tokens": 509174024.0, + "step": 13119 + }, + { + "epoch": 1.6689988551074926, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.21539306640625, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8968523740768433, + "num_tokens": 509206698.0, + "step": 13120 + }, + { + "epoch": 1.669126065386083, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.197161674499512, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8879936337471008, + "num_tokens": 509241252.0, + "step": 13121 + }, + { + "epoch": 1.6692532756646736, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.181025505065918, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8954128623008728, + "num_tokens": 509278978.0, + "step": 13122 + }, + { + "epoch": 1.6693804859432642, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.225882530212402, + "learning_rate": 1e-06, + "loss": 0.4044, + "mean_token_accuracy": 0.8639023303985596, + "num_tokens": 509317967.0, + "step": 13123 + }, + { + "epoch": 1.6695076962218547, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.119376182556152, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8998172283172607, + "num_tokens": 509360366.0, + "step": 13124 + }, + { + "epoch": 1.6696349065004452, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.070629119873047, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8898212909698486, + "num_tokens": 509400400.0, + "step": 13125 + }, + { + "epoch": 1.6697621167790357, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.240447998046875, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8825522065162659, + "num_tokens": 509437602.0, + "step": 13126 + }, + { + "epoch": 1.6698893270576263, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.167827606201172, + "learning_rate": 1e-06, + "loss": 0.3835, + "mean_token_accuracy": 0.8727445602416992, + "num_tokens": 509477860.0, + "step": 13127 + }, + { + "epoch": 1.6700165373362168, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.168384552001953, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8866361379623413, + "num_tokens": 509516716.0, + "step": 13128 + }, + { + "epoch": 1.6701437476148073, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.173644065856934, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8732717633247375, + "num_tokens": 509559297.0, + "step": 13129 + }, + { + "epoch": 1.6702709578933979, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.169400215148926, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8888342380523682, + "num_tokens": 509596168.0, + "step": 13130 + }, + { + "epoch": 1.6703981681719884, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.254487037658691, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8809698820114136, + "num_tokens": 509633994.0, + "step": 13131 + }, + { + "epoch": 1.670525378450579, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.087148666381836, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8946458101272583, + "num_tokens": 509676908.0, + "step": 13132 + }, + { + "epoch": 1.6706525887291694, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.213582992553711, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8815938830375671, + "num_tokens": 509720616.0, + "step": 13133 + }, + { + "epoch": 1.67077979900776, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.180913925170898, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8896826505661011, + "num_tokens": 509760577.0, + "step": 13134 + }, + { + "epoch": 1.6709070092863505, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.235262870788574, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8940561413764954, + "num_tokens": 509797032.0, + "step": 13135 + }, + { + "epoch": 1.6710342195649408, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.307451248168945, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8848868608474731, + "num_tokens": 509835682.0, + "step": 13136 + }, + { + "epoch": 1.6711614298435313, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.262924194335938, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8822399973869324, + "num_tokens": 509874004.0, + "step": 13137 + }, + { + "epoch": 1.6712886401221219, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.237908363342285, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8933968544006348, + "num_tokens": 509910486.0, + "step": 13138 + }, + { + "epoch": 1.6714158504007124, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.208516120910645, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8849760293960571, + "num_tokens": 509952500.0, + "step": 13139 + }, + { + "epoch": 1.671543060679303, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.266761779785156, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8822291493415833, + "num_tokens": 509989070.0, + "step": 13140 + }, + { + "epoch": 1.6716702709578934, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.2234468460083, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8727589845657349, + "num_tokens": 510032137.0, + "step": 13141 + }, + { + "epoch": 1.6717974812364838, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.245983123779297, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8950467109680176, + "num_tokens": 510069317.0, + "step": 13142 + }, + { + "epoch": 1.6719246915150743, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.173383712768555, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8802822828292847, + "num_tokens": 510111082.0, + "step": 13143 + }, + { + "epoch": 1.6720519017936648, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.250533103942871, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8897230625152588, + "num_tokens": 510150902.0, + "step": 13144 + }, + { + "epoch": 1.6721791120722553, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.268304824829102, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8782984018325806, + "num_tokens": 510189557.0, + "step": 13145 + }, + { + "epoch": 1.6723063223508459, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.187379837036133, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8912737369537354, + "num_tokens": 510230613.0, + "step": 13146 + }, + { + "epoch": 1.6724335326294364, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.248968124389648, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8897510766983032, + "num_tokens": 510273462.0, + "step": 13147 + }, + { + "epoch": 1.672560742908027, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.175129890441895, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8969389796257019, + "num_tokens": 510315062.0, + "step": 13148 + }, + { + "epoch": 1.6726879531866174, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.27691650390625, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8878645896911621, + "num_tokens": 510356576.0, + "step": 13149 + }, + { + "epoch": 1.672815163465208, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.196534156799316, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8960975408554077, + "num_tokens": 510395501.0, + "step": 13150 + }, + { + "epoch": 1.6729423737437985, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.19173812866211, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8800968527793884, + "num_tokens": 510439614.0, + "step": 13151 + }, + { + "epoch": 1.673069584022389, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.302715301513672, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.88404381275177, + "num_tokens": 510475443.0, + "step": 13152 + }, + { + "epoch": 1.6731967943009796, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.281045913696289, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8879901170730591, + "num_tokens": 510515685.0, + "step": 13153 + }, + { + "epoch": 1.67332400457957, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.146204948425293, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8796947598457336, + "num_tokens": 510559319.0, + "step": 13154 + }, + { + "epoch": 1.6734512148581606, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.268959045410156, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8962076902389526, + "num_tokens": 510599025.0, + "step": 13155 + }, + { + "epoch": 1.6735784251367511, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.241656303405762, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8861749172210693, + "num_tokens": 510638474.0, + "step": 13156 + }, + { + "epoch": 1.6737056354153417, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.191304206848145, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8805282711982727, + "num_tokens": 510680572.0, + "step": 13157 + }, + { + "epoch": 1.6738328456939322, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.36771011352539, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8855431079864502, + "num_tokens": 510711755.0, + "step": 13158 + }, + { + "epoch": 1.6739600559725227, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.235522270202637, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8823861479759216, + "num_tokens": 510746503.0, + "step": 13159 + }, + { + "epoch": 1.674087266251113, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.120676040649414, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8796479105949402, + "num_tokens": 510794335.0, + "step": 13160 + }, + { + "epoch": 1.6742144765297036, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.188026428222656, + "learning_rate": 1e-06, + "loss": 0.368, + "mean_token_accuracy": 0.874008059501648, + "num_tokens": 510840199.0, + "step": 13161 + }, + { + "epoch": 1.674341686808294, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.367817878723145, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8826779723167419, + "num_tokens": 510877019.0, + "step": 13162 + }, + { + "epoch": 1.6744688970868846, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.21818733215332, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8926610946655273, + "num_tokens": 510923872.0, + "step": 13163 + }, + { + "epoch": 1.6745961073654752, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.17255687713623, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.886137068271637, + "num_tokens": 510964277.0, + "step": 13164 + }, + { + "epoch": 1.6747233176440657, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.292728424072266, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8868215084075928, + "num_tokens": 510999833.0, + "step": 13165 + }, + { + "epoch": 1.674850527922656, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.215457916259766, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.8734651207923889, + "num_tokens": 511039547.0, + "step": 13166 + }, + { + "epoch": 1.6749777382012465, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.293111801147461, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.87884920835495, + "num_tokens": 511075392.0, + "step": 13167 + }, + { + "epoch": 1.675104948479837, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.283595085144043, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8856920003890991, + "num_tokens": 511112916.0, + "step": 13168 + }, + { + "epoch": 1.6752321587584276, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.160185813903809, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8863993287086487, + "num_tokens": 511156005.0, + "step": 13169 + }, + { + "epoch": 1.675359369037018, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.218181610107422, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8789744973182678, + "num_tokens": 511197764.0, + "step": 13170 + }, + { + "epoch": 1.6754865793156086, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.186151504516602, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8910011053085327, + "num_tokens": 511237976.0, + "step": 13171 + }, + { + "epoch": 1.6756137895941992, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.159280776977539, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8873505592346191, + "num_tokens": 511283195.0, + "step": 13172 + }, + { + "epoch": 1.6757409998727897, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.079673767089844e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.240068435668945, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8862356543540955, + "num_tokens": 511324459.0, + "step": 13173 + }, + { + "epoch": 1.6758682101513802, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.176642417907715, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.9019240140914917, + "num_tokens": 511361382.0, + "step": 13174 + }, + { + "epoch": 1.6759954204299707, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.22318172454834, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8816949129104614, + "num_tokens": 511404433.0, + "step": 13175 + }, + { + "epoch": 1.6761226307085613, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.275758743286133, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8816584348678589, + "num_tokens": 511442352.0, + "step": 13176 + }, + { + "epoch": 1.6762498409871518, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.237582206726074, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8946364521980286, + "num_tokens": 511477749.0, + "step": 13177 + }, + { + "epoch": 1.6763770512657423, + "ewc_loss": 0.0164794921875, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.246830940246582, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8727191686630249, + "num_tokens": 511519040.0, + "step": 13178 + }, + { + "epoch": 1.6765042615443329, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.132662773132324, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8893368244171143, + "num_tokens": 511560340.0, + "step": 13179 + }, + { + "epoch": 1.6766314718229234, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.244668006896973, + "learning_rate": 1e-06, + "loss": 0.3689, + "mean_token_accuracy": 0.8768734931945801, + "num_tokens": 511603043.0, + "step": 13180 + }, + { + "epoch": 1.676758682101514, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.260165214538574, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8795626163482666, + "num_tokens": 511635815.0, + "step": 13181 + }, + { + "epoch": 1.6768858923801044, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.287327766418457, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.885662317276001, + "num_tokens": 511666526.0, + "step": 13182 + }, + { + "epoch": 1.677013102658695, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.162211418151855, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.903380274772644, + "num_tokens": 511701849.0, + "step": 13183 + }, + { + "epoch": 1.6771403129372855, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.104986190795898, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.9046223163604736, + "num_tokens": 511745322.0, + "step": 13184 + }, + { + "epoch": 1.6772675232158758, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.16850757598877, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.885765016078949, + "num_tokens": 511784923.0, + "step": 13185 + }, + { + "epoch": 1.6773947334944663, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.344971656799316, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8795546889305115, + "num_tokens": 511824498.0, + "step": 13186 + }, + { + "epoch": 1.6775219437730569, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.163298606872559, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.896693766117096, + "num_tokens": 511864605.0, + "step": 13187 + }, + { + "epoch": 1.6776491540516474, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.113170623779297, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8855603933334351, + "num_tokens": 511914128.0, + "step": 13188 + }, + { + "epoch": 1.677776364330238, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.283472061157227, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8957061767578125, + "num_tokens": 511948964.0, + "step": 13189 + }, + { + "epoch": 1.6779035746088284, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.157684326171875, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8794537782669067, + "num_tokens": 511992052.0, + "step": 13190 + }, + { + "epoch": 1.6780307848874187, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.320686340332031, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8804341554641724, + "num_tokens": 512028942.0, + "step": 13191 + }, + { + "epoch": 1.6781579951660093, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.194755554199219, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8841204643249512, + "num_tokens": 512069496.0, + "step": 13192 + }, + { + "epoch": 1.6782852054445998, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.168608665466309, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8906362652778625, + "num_tokens": 512111417.0, + "step": 13193 + }, + { + "epoch": 1.6784124157231903, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.154606819152832, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8819484710693359, + "num_tokens": 512154248.0, + "step": 13194 + }, + { + "epoch": 1.6785396260017809, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.219292640686035, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8748675584793091, + "num_tokens": 512195589.0, + "step": 13195 + }, + { + "epoch": 1.6786668362803714, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.228059768676758, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8970667719841003, + "num_tokens": 512232138.0, + "step": 13196 + }, + { + "epoch": 1.678794046558962, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.275165557861328, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8858387470245361, + "num_tokens": 512268002.0, + "step": 13197 + }, + { + "epoch": 1.6789212568375524, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.263883590698242, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.886311948299408, + "num_tokens": 512301014.0, + "step": 13198 + }, + { + "epoch": 1.679048467116143, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.132977485656738, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8865763545036316, + "num_tokens": 512339848.0, + "step": 13199 + }, + { + "epoch": 1.6791756773947335, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.1563720703125, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8998890519142151, + "num_tokens": 512381755.0, + "step": 13200 + }, + { + "epoch": 1.679302887673324, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.151987075805664, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.893447995185852, + "num_tokens": 512423416.0, + "step": 13201 + }, + { + "epoch": 1.6794300979519146, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.138169288635254, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8889323472976685, + "num_tokens": 512467621.0, + "step": 13202 + }, + { + "epoch": 1.679557308230505, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.31990909576416, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8853238821029663, + "num_tokens": 512500024.0, + "step": 13203 + }, + { + "epoch": 1.6796845185090956, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.383365631103516, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8830184936523438, + "num_tokens": 512529464.0, + "step": 13204 + }, + { + "epoch": 1.6798117287876861, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.079413414001465, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.9003089070320129, + "num_tokens": 512575252.0, + "step": 13205 + }, + { + "epoch": 1.6799389390662767, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.17628002166748, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8893819451332092, + "num_tokens": 512617685.0, + "step": 13206 + }, + { + "epoch": 1.6800661493448672, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.22276496887207, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8931741118431091, + "num_tokens": 512653298.0, + "step": 13207 + }, + { + "epoch": 1.6801933596234577, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.153648376464844, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8825627565383911, + "num_tokens": 512691677.0, + "step": 13208 + }, + { + "epoch": 1.680320569902048, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.19294548034668, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8833328485488892, + "num_tokens": 512730901.0, + "step": 13209 + }, + { + "epoch": 1.6804477801806386, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.15438461303711, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8877239227294922, + "num_tokens": 512772202.0, + "step": 13210 + }, + { + "epoch": 1.680574990459229, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.238110542297363, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8839009404182434, + "num_tokens": 512810296.0, + "step": 13211 + }, + { + "epoch": 1.6807022007378196, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.1244535446167, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8889469504356384, + "num_tokens": 512852184.0, + "step": 13212 + }, + { + "epoch": 1.6808294110164101, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.256346702575684, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8871262073516846, + "num_tokens": 512885524.0, + "step": 13213 + }, + { + "epoch": 1.6809566212950007, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.245706558227539, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.875037431716919, + "num_tokens": 512927091.0, + "step": 13214 + }, + { + "epoch": 1.681083831573591, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.19844913482666, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.884788990020752, + "num_tokens": 512965144.0, + "step": 13215 + }, + { + "epoch": 1.6812110418521815, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.219494819641113, + "learning_rate": 1e-06, + "loss": 0.3752, + "mean_token_accuracy": 0.8747183680534363, + "num_tokens": 513006492.0, + "step": 13216 + }, + { + "epoch": 1.681338252130772, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.233057975769043, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8888617157936096, + "num_tokens": 513044850.0, + "step": 13217 + }, + { + "epoch": 1.6814654624093626, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.297472953796387, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8905666470527649, + "num_tokens": 513078640.0, + "step": 13218 + }, + { + "epoch": 1.681592672687953, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.306471824645996, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8876861333847046, + "num_tokens": 513113624.0, + "step": 13219 + }, + { + "epoch": 1.6817198829665436, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.27406120300293, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8840334415435791, + "num_tokens": 513151532.0, + "step": 13220 + }, + { + "epoch": 1.6818470932451342, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.324629783630371, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.875373363494873, + "num_tokens": 513189543.0, + "step": 13221 + }, + { + "epoch": 1.6819743035237247, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.2135009765625, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8873528242111206, + "num_tokens": 513230728.0, + "step": 13222 + }, + { + "epoch": 1.6821015138023152, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.175158500671387, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9038100242614746, + "num_tokens": 513268680.0, + "step": 13223 + }, + { + "epoch": 1.6822287240809057, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.22472095489502, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.886203408241272, + "num_tokens": 513308838.0, + "step": 13224 + }, + { + "epoch": 1.6823559343594963, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.209121704101562, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8886429071426392, + "num_tokens": 513347711.0, + "step": 13225 + }, + { + "epoch": 1.6824831446380868, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.163564682006836, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8821706771850586, + "num_tokens": 513395105.0, + "step": 13226 + }, + { + "epoch": 1.6826103549166773, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.153226852416992, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8866509199142456, + "num_tokens": 513433000.0, + "step": 13227 + }, + { + "epoch": 1.6827375651952678, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.141545295715332, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8933438658714294, + "num_tokens": 513473911.0, + "step": 13228 + }, + { + "epoch": 1.6828647754738584, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.245424270629883, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.878768265247345, + "num_tokens": 513514990.0, + "step": 13229 + }, + { + "epoch": 1.682991985752449, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.170966148376465, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8949577808380127, + "num_tokens": 513554869.0, + "step": 13230 + }, + { + "epoch": 1.6831191960310394, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.286218643188477, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8923565745353699, + "num_tokens": 513589217.0, + "step": 13231 + }, + { + "epoch": 1.68324640630963, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.261255264282227, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8898550271987915, + "num_tokens": 513623815.0, + "step": 13232 + }, + { + "epoch": 1.6833736165882205, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.167009353637695, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8908975124359131, + "num_tokens": 513663740.0, + "step": 13233 + }, + { + "epoch": 1.6835008268668108, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.187044143676758, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8879401683807373, + "num_tokens": 513706902.0, + "step": 13234 + }, + { + "epoch": 1.6836280371454013, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.167729377746582, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.874803900718689, + "num_tokens": 513748851.0, + "step": 13235 + }, + { + "epoch": 1.6837552474239919, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.16707706451416, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8904334306716919, + "num_tokens": 513788340.0, + "step": 13236 + }, + { + "epoch": 1.6838824577025824, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.314382553100586, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.881183385848999, + "num_tokens": 513821888.0, + "step": 13237 + }, + { + "epoch": 1.684009667981173, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.19583511352539, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.8708497285842896, + "num_tokens": 513863712.0, + "step": 13238 + }, + { + "epoch": 1.6841368782597634, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.207836151123047, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8885654211044312, + "num_tokens": 513902203.0, + "step": 13239 + }, + { + "epoch": 1.6842640885383537, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.223563194274902, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8787891864776611, + "num_tokens": 513945396.0, + "step": 13240 + }, + { + "epoch": 1.6843912988169443, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.170001029968262, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8913392424583435, + "num_tokens": 513983465.0, + "step": 13241 + }, + { + "epoch": 1.6845185090955348, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.142449378967285, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8839420080184937, + "num_tokens": 514025208.0, + "step": 13242 + }, + { + "epoch": 1.6846457193741253, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.103515625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.293834686279297, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8895403742790222, + "num_tokens": 514060370.0, + "step": 13243 + }, + { + "epoch": 1.6847729296527159, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.143019676208496, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8813164234161377, + "num_tokens": 514109068.0, + "step": 13244 + }, + { + "epoch": 1.6849001399313064, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.126399040222168, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8998692631721497, + "num_tokens": 514147869.0, + "step": 13245 + }, + { + "epoch": 1.685027350209897, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.227108001708984, + "learning_rate": 1e-06, + "loss": 0.3475, + "mean_token_accuracy": 0.8813075423240662, + "num_tokens": 514187899.0, + "step": 13246 + }, + { + "epoch": 1.6851545604884874, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.145231246948242, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.883506178855896, + "num_tokens": 514230637.0, + "step": 13247 + }, + { + "epoch": 1.685281770767078, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.162701606750488, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9036908745765686, + "num_tokens": 514270464.0, + "step": 13248 + }, + { + "epoch": 1.6854089810456685, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.196516036987305, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.906012237071991, + "num_tokens": 514308438.0, + "step": 13249 + }, + { + "epoch": 1.685536191324259, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.17327880859375, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.892462432384491, + "num_tokens": 514345561.0, + "step": 13250 + }, + { + "epoch": 1.6856634016028496, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.205550193786621, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8924580812454224, + "num_tokens": 514384073.0, + "step": 13251 + }, + { + "epoch": 1.68579061188144, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.143949508666992, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.885196328163147, + "num_tokens": 514428141.0, + "step": 13252 + }, + { + "epoch": 1.6859178221600306, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.337471961975098, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8890384435653687, + "num_tokens": 514465307.0, + "step": 13253 + }, + { + "epoch": 1.6860450324386211, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.25348949432373, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.9033665657043457, + "num_tokens": 514500956.0, + "step": 13254 + }, + { + "epoch": 1.6861722427172117, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.167173385620117, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9013625979423523, + "num_tokens": 514536881.0, + "step": 13255 + }, + { + "epoch": 1.6862994529958022, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.242860794067383, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.896070659160614, + "num_tokens": 514571114.0, + "step": 13256 + }, + { + "epoch": 1.6864266632743927, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.206829071044922, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8913482427597046, + "num_tokens": 514608365.0, + "step": 13257 + }, + { + "epoch": 1.686553873552983, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.199002265930176, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.8714407682418823, + "num_tokens": 514647497.0, + "step": 13258 + }, + { + "epoch": 1.6866810838315736, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.265007972717285, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.877367377281189, + "num_tokens": 514687047.0, + "step": 13259 + }, + { + "epoch": 1.686808294110164, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.324594497680664, + "learning_rate": 1e-06, + "loss": 0.4435, + "mean_token_accuracy": 0.8595527410507202, + "num_tokens": 514724989.0, + "step": 13260 + }, + { + "epoch": 1.6869355043887546, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.152777671813965, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8967976570129395, + "num_tokens": 514763337.0, + "step": 13261 + }, + { + "epoch": 1.6870627146673451, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.199275016784668, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8788778185844421, + "num_tokens": 514804346.0, + "step": 13262 + }, + { + "epoch": 1.6871899249459357, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.138495445251465, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8920358419418335, + "num_tokens": 514843933.0, + "step": 13263 + }, + { + "epoch": 1.687317135224526, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.186662673950195, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8889853954315186, + "num_tokens": 514880779.0, + "step": 13264 + }, + { + "epoch": 1.6874443455031165, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.19017505645752, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8917831182479858, + "num_tokens": 514922231.0, + "step": 13265 + }, + { + "epoch": 1.687571555781707, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.150779724121094, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8944094777107239, + "num_tokens": 514966308.0, + "step": 13266 + }, + { + "epoch": 1.6876987660602976, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.382723808288574, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8921316266059875, + "num_tokens": 514998711.0, + "step": 13267 + }, + { + "epoch": 1.687825976338888, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.368268966674805, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8797769546508789, + "num_tokens": 515032974.0, + "step": 13268 + }, + { + "epoch": 1.6879531866174786, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.222746849060059, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8742278814315796, + "num_tokens": 515070508.0, + "step": 13269 + }, + { + "epoch": 1.6880803968960691, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.24416446685791, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8882819414138794, + "num_tokens": 515107846.0, + "step": 13270 + }, + { + "epoch": 1.6882076071746597, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.08578872680664, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.9030104875564575, + "num_tokens": 515146132.0, + "step": 13271 + }, + { + "epoch": 1.6883348174532502, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.259385108947754, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8997198343276978, + "num_tokens": 515179395.0, + "step": 13272 + }, + { + "epoch": 1.6884620277318407, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.236031532287598, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8803116083145142, + "num_tokens": 515215344.0, + "step": 13273 + }, + { + "epoch": 1.6885892380104313, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.196666717529297, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.901494562625885, + "num_tokens": 515250249.0, + "step": 13274 + }, + { + "epoch": 1.6887164482890218, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.17403793334961, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8978201150894165, + "num_tokens": 515288205.0, + "step": 13275 + }, + { + "epoch": 1.6888436585676123, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.1805419921875, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8893527388572693, + "num_tokens": 515328294.0, + "step": 13276 + }, + { + "epoch": 1.6889708688462028, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.253908157348633, + "learning_rate": 1e-06, + "loss": 0.3811, + "mean_token_accuracy": 0.871119499206543, + "num_tokens": 515364286.0, + "step": 13277 + }, + { + "epoch": 1.6890980791247934, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.217381477355957, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8865618705749512, + "num_tokens": 515401372.0, + "step": 13278 + }, + { + "epoch": 1.689225289403384, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.21652889251709, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8757530450820923, + "num_tokens": 515438931.0, + "step": 13279 + }, + { + "epoch": 1.6893524996819744, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.268428802490234, + "learning_rate": 1e-06, + "loss": 0.379, + "mean_token_accuracy": 0.8704092502593994, + "num_tokens": 515477227.0, + "step": 13280 + }, + { + "epoch": 1.689479709960565, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.238516807556152, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8936525583267212, + "num_tokens": 515515961.0, + "step": 13281 + }, + { + "epoch": 1.6896069202391555, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.180618286132812, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8909833431243896, + "num_tokens": 515553154.0, + "step": 13282 + }, + { + "epoch": 1.6897341305177458, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.159988403320312, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8749662637710571, + "num_tokens": 515600135.0, + "step": 13283 + }, + { + "epoch": 1.6898613407963363, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.116018295288086, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8939681053161621, + "num_tokens": 515639063.0, + "step": 13284 + }, + { + "epoch": 1.6899885510749268, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.228726387023926, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8870294690132141, + "num_tokens": 515683467.0, + "step": 13285 + }, + { + "epoch": 1.6901157613535174, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.205293655395508, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.895605206489563, + "num_tokens": 515718862.0, + "step": 13286 + }, + { + "epoch": 1.690242971632108, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.242509841918945, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8920949697494507, + "num_tokens": 515753825.0, + "step": 13287 + }, + { + "epoch": 1.6903701819106984, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.152175903320312, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8933644890785217, + "num_tokens": 515795198.0, + "step": 13288 + }, + { + "epoch": 1.6904973921892887, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.21359634399414, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8891831636428833, + "num_tokens": 515835930.0, + "step": 13289 + }, + { + "epoch": 1.6906246024678793, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.185928344726562, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8899174332618713, + "num_tokens": 515874596.0, + "step": 13290 + }, + { + "epoch": 1.6907518127464698, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.259939193725586, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8832634091377258, + "num_tokens": 515911414.0, + "step": 13291 + }, + { + "epoch": 1.6908790230250603, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.098888397216797, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8923678398132324, + "num_tokens": 515954317.0, + "step": 13292 + }, + { + "epoch": 1.6910062333036509, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.234774589538574, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8995359539985657, + "num_tokens": 515991243.0, + "step": 13293 + }, + { + "epoch": 1.6911334435822414, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.223445892333984, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8825861811637878, + "num_tokens": 516031582.0, + "step": 13294 + }, + { + "epoch": 1.691260653860832, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.136369705200195, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8894776105880737, + "num_tokens": 516073526.0, + "step": 13295 + }, + { + "epoch": 1.6913878641394224, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.293964385986328, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8766450881958008, + "num_tokens": 516113444.0, + "step": 13296 + }, + { + "epoch": 1.691515074418013, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.131233215332031, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.885887861251831, + "num_tokens": 516159745.0, + "step": 13297 + }, + { + "epoch": 1.6916422846966035, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.172758102416992, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8935869932174683, + "num_tokens": 516199049.0, + "step": 13298 + }, + { + "epoch": 1.691769494975194, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.193093299865723, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8867003917694092, + "num_tokens": 516238905.0, + "step": 13299 + }, + { + "epoch": 1.6918967052537845, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.112796783447266, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8898568153381348, + "num_tokens": 516283519.0, + "step": 13300 + }, + { + "epoch": 1.692023915532375, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.171526908874512, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.89857017993927, + "num_tokens": 516325541.0, + "step": 13301 + }, + { + "epoch": 1.6921511258109656, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.199532508850098, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8921862244606018, + "num_tokens": 516364220.0, + "step": 13302 + }, + { + "epoch": 1.6922783360895561, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.135109901428223, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8938495516777039, + "num_tokens": 516404424.0, + "step": 13303 + }, + { + "epoch": 1.6924055463681467, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.18120288848877, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8928970098495483, + "num_tokens": 516446981.0, + "step": 13304 + }, + { + "epoch": 1.6925327566467372, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.234125137329102, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8889166116714478, + "num_tokens": 516481679.0, + "step": 13305 + }, + { + "epoch": 1.6926599669253277, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.182303428649902, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9079379439353943, + "num_tokens": 516519668.0, + "step": 13306 + }, + { + "epoch": 1.692787177203918, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.24625015258789, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8926745057106018, + "num_tokens": 516556855.0, + "step": 13307 + }, + { + "epoch": 1.6929143874825086, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.220767974853516, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8865379691123962, + "num_tokens": 516593810.0, + "step": 13308 + }, + { + "epoch": 1.693041597761099, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.169296264648438, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.894415557384491, + "num_tokens": 516631867.0, + "step": 13309 + }, + { + "epoch": 1.6931688080396896, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.25747299194336, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8767999410629272, + "num_tokens": 516670764.0, + "step": 13310 + }, + { + "epoch": 1.6932960183182801, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.205358505249023, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8882304430007935, + "num_tokens": 516708469.0, + "step": 13311 + }, + { + "epoch": 1.6934232285968707, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.201722145080566, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8914084434509277, + "num_tokens": 516745047.0, + "step": 13312 + }, + { + "epoch": 1.693550438875461, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.116792678833008, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8789434432983398, + "num_tokens": 516791233.0, + "step": 13313 + }, + { + "epoch": 1.6936776491540515, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.191795349121094, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8791152834892273, + "num_tokens": 516833975.0, + "step": 13314 + }, + { + "epoch": 1.693804859432642, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.183706283569336, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8954268097877502, + "num_tokens": 516869411.0, + "step": 13315 + }, + { + "epoch": 1.6939320697112326, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.121148109436035, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8821383714675903, + "num_tokens": 516913103.0, + "step": 13316 + }, + { + "epoch": 1.694059279989823, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.2407808303833, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.881432294845581, + "num_tokens": 516949313.0, + "step": 13317 + }, + { + "epoch": 1.6941864902684136, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.316826820373535, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8859730362892151, + "num_tokens": 516982106.0, + "step": 13318 + }, + { + "epoch": 1.6943137005470041, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.186845779418945, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8847519755363464, + "num_tokens": 517022096.0, + "step": 13319 + }, + { + "epoch": 1.6944409108255947, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.167724609375, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8834342956542969, + "num_tokens": 517062341.0, + "step": 13320 + }, + { + "epoch": 1.6945681211041852, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.430954933166504, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8856456875801086, + "num_tokens": 517088209.0, + "step": 13321 + }, + { + "epoch": 1.6946953313827757, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.17744255065918, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8790858387947083, + "num_tokens": 517128950.0, + "step": 13322 + }, + { + "epoch": 1.6948225416613663, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.125680923461914, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.9043958187103271, + "num_tokens": 517171582.0, + "step": 13323 + }, + { + "epoch": 1.6949497519399568, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.13247013092041, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.8783009052276611, + "num_tokens": 517217938.0, + "step": 13324 + }, + { + "epoch": 1.6950769622185473, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.298934936523438, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.883648693561554, + "num_tokens": 517253873.0, + "step": 13325 + }, + { + "epoch": 1.6952041724971378, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.22111988067627, + "learning_rate": 1e-06, + "loss": 0.3816, + "mean_token_accuracy": 0.871840238571167, + "num_tokens": 517295334.0, + "step": 13326 + }, + { + "epoch": 1.6953313827757284, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.31983470916748, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8899915814399719, + "num_tokens": 517329923.0, + "step": 13327 + }, + { + "epoch": 1.695458593054319, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.110610008239746, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8896998167037964, + "num_tokens": 517374864.0, + "step": 13328 + }, + { + "epoch": 1.6955858033329094, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.258530616760254, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8831671476364136, + "num_tokens": 517415105.0, + "step": 13329 + }, + { + "epoch": 1.6957130136115, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.2659273147583, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8805567622184753, + "num_tokens": 517448688.0, + "step": 13330 + }, + { + "epoch": 1.6958402238900905, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.17138957977295, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8993761539459229, + "num_tokens": 517486702.0, + "step": 13331 + }, + { + "epoch": 1.6959674341686808, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.2422456741333, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8941426277160645, + "num_tokens": 517530543.0, + "step": 13332 + }, + { + "epoch": 1.6960946444472713, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.205255508422852, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8983359336853027, + "num_tokens": 517569407.0, + "step": 13333 + }, + { + "epoch": 1.6962218547258618, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.213872909545898, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.893600583076477, + "num_tokens": 517609613.0, + "step": 13334 + }, + { + "epoch": 1.6963490650044524, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.187423706054688, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8940731883049011, + "num_tokens": 517647043.0, + "step": 13335 + }, + { + "epoch": 1.696476275283043, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.233473777770996, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.893535315990448, + "num_tokens": 517686707.0, + "step": 13336 + }, + { + "epoch": 1.6966034855616334, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.32746696472168, + "learning_rate": 1e-06, + "loss": 0.3664, + "mean_token_accuracy": 0.8803805708885193, + "num_tokens": 517720125.0, + "step": 13337 + }, + { + "epoch": 1.6967306958402237, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.258702278137207, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.9000765085220337, + "num_tokens": 517751814.0, + "step": 13338 + }, + { + "epoch": 1.6968579061188143, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.123275756835938, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8787051439285278, + "num_tokens": 517795957.0, + "step": 13339 + }, + { + "epoch": 1.6969851163974048, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.294770240783691, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8795264959335327, + "num_tokens": 517834072.0, + "step": 13340 + }, + { + "epoch": 1.6971123266759953, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.232349395751953, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.888495683670044, + "num_tokens": 517871002.0, + "step": 13341 + }, + { + "epoch": 1.6972395369545858, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.260915756225586, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8854320645332336, + "num_tokens": 517906165.0, + "step": 13342 + }, + { + "epoch": 1.6973667472331764, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.219727516174316, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8883919715881348, + "num_tokens": 517942688.0, + "step": 13343 + }, + { + "epoch": 1.697493957511767, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.230646133422852, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8871995210647583, + "num_tokens": 517984335.0, + "step": 13344 + }, + { + "epoch": 1.6976211677903574, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.264687538146973, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8939007520675659, + "num_tokens": 518019057.0, + "step": 13345 + }, + { + "epoch": 1.697748378068948, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.28443431854248, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8946982622146606, + "num_tokens": 518052669.0, + "step": 13346 + }, + { + "epoch": 1.6978755883475385, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.155226707458496, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8940328359603882, + "num_tokens": 518094275.0, + "step": 13347 + }, + { + "epoch": 1.698002798626129, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.169197082519531, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.896259069442749, + "num_tokens": 518134298.0, + "step": 13348 + }, + { + "epoch": 1.6981300089047195, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.25698471069336, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8779456615447998, + "num_tokens": 518173991.0, + "step": 13349 + }, + { + "epoch": 1.69825721918331, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.19231128692627, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.897355318069458, + "num_tokens": 518210096.0, + "step": 13350 + }, + { + "epoch": 1.6983844294619006, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.246700286865234, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8894577026367188, + "num_tokens": 518244107.0, + "step": 13351 + }, + { + "epoch": 1.6985116397404911, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.151199340820312e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.151598930358887, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9032011032104492, + "num_tokens": 518280565.0, + "step": 13352 + }, + { + "epoch": 1.6986388500190817, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.23852252960205, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8838841915130615, + "num_tokens": 518317974.0, + "step": 13353 + }, + { + "epoch": 1.6987660602976722, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.264004707336426, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8807448744773865, + "num_tokens": 518357271.0, + "step": 13354 + }, + { + "epoch": 1.6988932705762627, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.218241691589355, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8906388282775879, + "num_tokens": 518393290.0, + "step": 13355 + }, + { + "epoch": 1.699020480854853, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.161584854125977, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8893760442733765, + "num_tokens": 518436820.0, + "step": 13356 + }, + { + "epoch": 1.6991476911334435, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.244617462158203, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8815308213233948, + "num_tokens": 518482510.0, + "step": 13357 + }, + { + "epoch": 1.699274901412034, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.366769790649414, + "learning_rate": 1e-06, + "loss": 0.3939, + "mean_token_accuracy": 0.8678007125854492, + "num_tokens": 518518433.0, + "step": 13358 + }, + { + "epoch": 1.6994021116906246, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.260553359985352, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8861804008483887, + "num_tokens": 518554466.0, + "step": 13359 + }, + { + "epoch": 1.6995293219692151, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.186813354492188, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8842998743057251, + "num_tokens": 518595853.0, + "step": 13360 + }, + { + "epoch": 1.6996565322478057, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.2920503616333, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8869715332984924, + "num_tokens": 518631273.0, + "step": 13361 + }, + { + "epoch": 1.699783742526396, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.188078880310059, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.883619487285614, + "num_tokens": 518667881.0, + "step": 13362 + }, + { + "epoch": 1.6999109528049865, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.226066589355469, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.89694744348526, + "num_tokens": 518704159.0, + "step": 13363 + }, + { + "epoch": 1.700038163083577, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.231354713439941, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.896854817867279, + "num_tokens": 518739342.0, + "step": 13364 + }, + { + "epoch": 1.7001653733621676, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.260210037231445, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8862779140472412, + "num_tokens": 518778323.0, + "step": 13365 + }, + { + "epoch": 1.700292583640758, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.326403617858887, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8784456253051758, + "num_tokens": 518818501.0, + "step": 13366 + }, + { + "epoch": 1.7004197939193486, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.26955795288086, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8907533884048462, + "num_tokens": 518855586.0, + "step": 13367 + }, + { + "epoch": 1.7005470041979391, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.203752517700195, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8983777761459351, + "num_tokens": 518890438.0, + "step": 13368 + }, + { + "epoch": 1.7006742144765297, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.284944534301758, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8890677094459534, + "num_tokens": 518930849.0, + "step": 13369 + }, + { + "epoch": 1.7008014247551202, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.211332321166992, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8833221197128296, + "num_tokens": 518972014.0, + "step": 13370 + }, + { + "epoch": 1.7009286350337107, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.242810249328613, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8867319822311401, + "num_tokens": 519009601.0, + "step": 13371 + }, + { + "epoch": 1.7010558453123013, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.23559284210205, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8933894038200378, + "num_tokens": 519046553.0, + "step": 13372 + }, + { + "epoch": 1.7011830555908918, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.290587425231934, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8962980508804321, + "num_tokens": 519080720.0, + "step": 13373 + }, + { + "epoch": 1.7013102658694823, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.292731285095215, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8770188093185425, + "num_tokens": 519117739.0, + "step": 13374 + }, + { + "epoch": 1.7014374761480728, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.2044038772583, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8914394974708557, + "num_tokens": 519158371.0, + "step": 13375 + }, + { + "epoch": 1.7015646864266634, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.22714614868164, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8881969451904297, + "num_tokens": 519196836.0, + "step": 13376 + }, + { + "epoch": 1.701691896705254, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.177946090698242, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9042994379997253, + "num_tokens": 519239205.0, + "step": 13377 + }, + { + "epoch": 1.7018191069838444, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.224316596984863, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8908733129501343, + "num_tokens": 519282918.0, + "step": 13378 + }, + { + "epoch": 1.701946317262435, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.215888023376465, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.9002450704574585, + "num_tokens": 519324070.0, + "step": 13379 + }, + { + "epoch": 1.7020735275410255, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.265974998474121, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8737130165100098, + "num_tokens": 519361164.0, + "step": 13380 + }, + { + "epoch": 1.7022007378196158, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.1946439743042, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8948647975921631, + "num_tokens": 519400685.0, + "step": 13381 + }, + { + "epoch": 1.7023279480982063, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.269502639770508, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8910307884216309, + "num_tokens": 519442024.0, + "step": 13382 + }, + { + "epoch": 1.7024551583767968, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.19420051574707, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8879024982452393, + "num_tokens": 519481590.0, + "step": 13383 + }, + { + "epoch": 1.7025823686553874, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.206693649291992, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8975948691368103, + "num_tokens": 519517032.0, + "step": 13384 + }, + { + "epoch": 1.702709578933978, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.309179306030273, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8875899314880371, + "num_tokens": 519552599.0, + "step": 13385 + }, + { + "epoch": 1.7028367892125684, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.257575035095215, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8988416790962219, + "num_tokens": 519589720.0, + "step": 13386 + }, + { + "epoch": 1.7029639994911587, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.139840126037598, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.886676549911499, + "num_tokens": 519636619.0, + "step": 13387 + }, + { + "epoch": 1.7030912097697493, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.183478355407715, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.9031007289886475, + "num_tokens": 519678707.0, + "step": 13388 + }, + { + "epoch": 1.7032184200483398, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.18436050415039, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8975516557693481, + "num_tokens": 519717553.0, + "step": 13389 + }, + { + "epoch": 1.7033456303269303, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.24375057220459, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8846138119697571, + "num_tokens": 519759029.0, + "step": 13390 + }, + { + "epoch": 1.7034728406055208, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.214445114135742, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8821564316749573, + "num_tokens": 519800237.0, + "step": 13391 + }, + { + "epoch": 1.7036000508841114, + "ewc_loss": 0.0166015625, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0371208190917969e-05, + "grad_norm": 9.331645011901855, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8725898861885071, + "num_tokens": 519835251.0, + "step": 13392 + }, + { + "epoch": 1.703727261162702, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.196625709533691, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8980687856674194, + "num_tokens": 519872201.0, + "step": 13393 + }, + { + "epoch": 1.7038544714412924, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.148702621459961, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8827850818634033, + "num_tokens": 519913685.0, + "step": 13394 + }, + { + "epoch": 1.703981681719883, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.323122024536133, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8853879570960999, + "num_tokens": 519950428.0, + "step": 13395 + }, + { + "epoch": 1.7041088919984735, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.243500709533691, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8941505551338196, + "num_tokens": 519983999.0, + "step": 13396 + }, + { + "epoch": 1.704236102277064, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.15846061706543, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8908337950706482, + "num_tokens": 520020619.0, + "step": 13397 + }, + { + "epoch": 1.7043633125556545, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.198883056640625e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.248466491699219, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8884165287017822, + "num_tokens": 520054212.0, + "step": 13398 + }, + { + "epoch": 1.704490522834245, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.195089340209961, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8939356803894043, + "num_tokens": 520095542.0, + "step": 13399 + }, + { + "epoch": 1.7046177331128356, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.279582023620605, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8788312077522278, + "num_tokens": 520134203.0, + "step": 13400 + }, + { + "epoch": 1.7047449433914261, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.257023811340332, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8888962268829346, + "num_tokens": 520172114.0, + "step": 13401 + }, + { + "epoch": 1.7048721536700167, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.333067893981934, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.882021963596344, + "num_tokens": 520205766.0, + "step": 13402 + }, + { + "epoch": 1.7049993639486072, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.24481201171875, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8914151191711426, + "num_tokens": 520241401.0, + "step": 13403 + }, + { + "epoch": 1.7051265742271977, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.147693634033203, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8856592178344727, + "num_tokens": 520283474.0, + "step": 13404 + }, + { + "epoch": 1.705253784505788, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.26451587677002, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8955202698707581, + "num_tokens": 520321371.0, + "step": 13405 + }, + { + "epoch": 1.7053809947843785, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.269697189331055, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.894864559173584, + "num_tokens": 520355226.0, + "step": 13406 + }, + { + "epoch": 1.705508205062969, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.184183120727539, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9060640335083008, + "num_tokens": 520393630.0, + "step": 13407 + }, + { + "epoch": 1.7056354153415596, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.318262100219727, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8906370997428894, + "num_tokens": 520427258.0, + "step": 13408 + }, + { + "epoch": 1.7057626256201501, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.174347877502441, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8810360431671143, + "num_tokens": 520469390.0, + "step": 13409 + }, + { + "epoch": 1.7058898358987407, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.251056671142578, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8870230913162231, + "num_tokens": 520507130.0, + "step": 13410 + }, + { + "epoch": 1.706017046177331, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.193739891052246, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8933131694793701, + "num_tokens": 520550331.0, + "step": 13411 + }, + { + "epoch": 1.7061442564559215, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.227657318115234, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8874576091766357, + "num_tokens": 520591425.0, + "step": 13412 + }, + { + "epoch": 1.706271466734512, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.30582046508789, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8755282163619995, + "num_tokens": 520628472.0, + "step": 13413 + }, + { + "epoch": 1.7063986770131025, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.224736213684082, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8910649418830872, + "num_tokens": 520663180.0, + "step": 13414 + }, + { + "epoch": 1.706525887291693, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.221049308776855, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8884790539741516, + "num_tokens": 520703518.0, + "step": 13415 + }, + { + "epoch": 1.7066530975702836, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.335770606994629, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8815451860427856, + "num_tokens": 520737858.0, + "step": 13416 + }, + { + "epoch": 1.7067803078488741, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.149774551391602, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8957970142364502, + "num_tokens": 520772326.0, + "step": 13417 + }, + { + "epoch": 1.7069075181274647, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.227177619934082, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8924793601036072, + "num_tokens": 520808436.0, + "step": 13418 + }, + { + "epoch": 1.7070347284060552, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.225197792053223, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8849242925643921, + "num_tokens": 520847594.0, + "step": 13419 + }, + { + "epoch": 1.7071619386846457, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.18680477142334, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8929774165153503, + "num_tokens": 520886623.0, + "step": 13420 + }, + { + "epoch": 1.7072891489632362, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.234786987304688, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.885435938835144, + "num_tokens": 520926048.0, + "step": 13421 + }, + { + "epoch": 1.7074163592418268, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.639424324035645, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8861854076385498, + "num_tokens": 520966436.0, + "step": 13422 + }, + { + "epoch": 1.7075435695204173, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.203495025634766, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.893856942653656, + "num_tokens": 521003458.0, + "step": 13423 + }, + { + "epoch": 1.7076707797990078, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.101888656616211, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8980629444122314, + "num_tokens": 521042290.0, + "step": 13424 + }, + { + "epoch": 1.7077979900775984, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.305292129516602, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8901928067207336, + "num_tokens": 521078193.0, + "step": 13425 + }, + { + "epoch": 1.7079252003561889, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.177169799804688, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.880480170249939, + "num_tokens": 521112896.0, + "step": 13426 + }, + { + "epoch": 1.7080524106347794, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.237696647644043, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.883675217628479, + "num_tokens": 521149048.0, + "step": 13427 + }, + { + "epoch": 1.70817962091337, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.256462097167969, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8766286373138428, + "num_tokens": 521186972.0, + "step": 13428 + }, + { + "epoch": 1.7083068311919605, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.288330078125, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8910945653915405, + "num_tokens": 521220381.0, + "step": 13429 + }, + { + "epoch": 1.7084340414705508, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.255163192749023, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8750134706497192, + "num_tokens": 521257521.0, + "step": 13430 + }, + { + "epoch": 1.7085612517491413, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.300127983093262, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8888100385665894, + "num_tokens": 521295209.0, + "step": 13431 + }, + { + "epoch": 1.7086884620277318, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.255001068115234, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8967480063438416, + "num_tokens": 521331338.0, + "step": 13432 + }, + { + "epoch": 1.7088156723063224, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.172786712646484, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8978849649429321, + "num_tokens": 521370379.0, + "step": 13433 + }, + { + "epoch": 1.708942882584913, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.31021785736084, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8870783448219299, + "num_tokens": 521401839.0, + "step": 13434 + }, + { + "epoch": 1.7090700928635034, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.220775604248047, + "learning_rate": 1e-06, + "loss": 0.3985, + "mean_token_accuracy": 0.868619441986084, + "num_tokens": 521438771.0, + "step": 13435 + }, + { + "epoch": 1.7091973031420937, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.18752670288086, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8878418207168579, + "num_tokens": 521478110.0, + "step": 13436 + }, + { + "epoch": 1.7093245134206843, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.229640007019043, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8849694132804871, + "num_tokens": 521515539.0, + "step": 13437 + }, + { + "epoch": 1.7094517236992748, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.25124454498291, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8848249316215515, + "num_tokens": 521552855.0, + "step": 13438 + }, + { + "epoch": 1.7095789339778653, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.238690376281738, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8719561100006104, + "num_tokens": 521594031.0, + "step": 13439 + }, + { + "epoch": 1.7097061442564558, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.143260955810547, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8770270347595215, + "num_tokens": 521633888.0, + "step": 13440 + }, + { + "epoch": 1.7098333545350464, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.187017440795898, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8831343650817871, + "num_tokens": 521678632.0, + "step": 13441 + }, + { + "epoch": 1.709960564813637, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.105691909790039, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8974415063858032, + "num_tokens": 521716670.0, + "step": 13442 + }, + { + "epoch": 1.7100877750922274, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.33781623840332, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8725628852844238, + "num_tokens": 521752483.0, + "step": 13443 + }, + { + "epoch": 1.710214985370818, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.253990173339844, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8922138214111328, + "num_tokens": 521789127.0, + "step": 13444 + }, + { + "epoch": 1.7103421956494085, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.112157821655273, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8881667852401733, + "num_tokens": 521830783.0, + "step": 13445 + }, + { + "epoch": 1.710469405927999, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.21107292175293, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.89537513256073, + "num_tokens": 521867451.0, + "step": 13446 + }, + { + "epoch": 1.7105966162065895, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.229841232299805, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.89217209815979, + "num_tokens": 521907842.0, + "step": 13447 + }, + { + "epoch": 1.71072382648518, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.141674995422363, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8896734714508057, + "num_tokens": 521949066.0, + "step": 13448 + }, + { + "epoch": 1.7108510367637706, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.225095748901367, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8929404616355896, + "num_tokens": 521990330.0, + "step": 13449 + }, + { + "epoch": 1.7109782470423611, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.14828872680664, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8981224298477173, + "num_tokens": 522030246.0, + "step": 13450 + }, + { + "epoch": 1.7111054573209517, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.296464920043945, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8805501461029053, + "num_tokens": 522068793.0, + "step": 13451 + }, + { + "epoch": 1.7112326675995422, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.249106407165527, + "learning_rate": 1e-06, + "loss": 0.3976, + "mean_token_accuracy": 0.8643309473991394, + "num_tokens": 522110054.0, + "step": 13452 + }, + { + "epoch": 1.7113598778781327, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.166769981384277, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8915492296218872, + "num_tokens": 522150013.0, + "step": 13453 + }, + { + "epoch": 1.711487088156723, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.200972557067871, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8858996629714966, + "num_tokens": 522192228.0, + "step": 13454 + }, + { + "epoch": 1.7116142984353135, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.256821632385254, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.888542652130127, + "num_tokens": 522231738.0, + "step": 13455 + }, + { + "epoch": 1.711741508713904, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.200934410095215, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8908874988555908, + "num_tokens": 522270266.0, + "step": 13456 + }, + { + "epoch": 1.7118687189924946, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.434964179992676, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8784689903259277, + "num_tokens": 522300159.0, + "step": 13457 + }, + { + "epoch": 1.7119959292710851, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.183289527893066, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8771454691886902, + "num_tokens": 522342658.0, + "step": 13458 + }, + { + "epoch": 1.7121231395496757, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.20134449005127, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8844454884529114, + "num_tokens": 522380078.0, + "step": 13459 + }, + { + "epoch": 1.712250349828266, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.27739143371582, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8855805397033691, + "num_tokens": 522417865.0, + "step": 13460 + }, + { + "epoch": 1.7123775601068565, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.15846061706543, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8919740319252014, + "num_tokens": 522455678.0, + "step": 13461 + }, + { + "epoch": 1.712504770385447, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.215003967285156, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8831115365028381, + "num_tokens": 522500059.0, + "step": 13462 + }, + { + "epoch": 1.7126319806640375, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.365530014038086, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8810855150222778, + "num_tokens": 522532887.0, + "step": 13463 + }, + { + "epoch": 1.712759190942628, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.190777778625488, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8796707391738892, + "num_tokens": 522576631.0, + "step": 13464 + }, + { + "epoch": 1.7128864012212186, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.272010803222656, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8848165273666382, + "num_tokens": 522620168.0, + "step": 13465 + }, + { + "epoch": 1.7130136114998091, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.180002212524414, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8944127559661865, + "num_tokens": 522660454.0, + "step": 13466 + }, + { + "epoch": 1.7131408217783997, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.202363014221191, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8812383413314819, + "num_tokens": 522700214.0, + "step": 13467 + }, + { + "epoch": 1.7132680320569902, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.37686824798584, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.883224606513977, + "num_tokens": 522738112.0, + "step": 13468 + }, + { + "epoch": 1.7133952423355807, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.297368049621582, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8887513875961304, + "num_tokens": 522781228.0, + "step": 13469 + }, + { + "epoch": 1.7135224526141712, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.192577362060547, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8832955360412598, + "num_tokens": 522823373.0, + "step": 13470 + }, + { + "epoch": 1.7136496628927618, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.151006698608398, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8855989575386047, + "num_tokens": 522866827.0, + "step": 13471 + }, + { + "epoch": 1.7137768731713523, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.252472877502441, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8761455416679382, + "num_tokens": 522905624.0, + "step": 13472 + }, + { + "epoch": 1.7139040834499428, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.128351211547852, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8921391367912292, + "num_tokens": 522947266.0, + "step": 13473 + }, + { + "epoch": 1.7140312937285334, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.258657455444336, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8914563059806824, + "num_tokens": 522983202.0, + "step": 13474 + }, + { + "epoch": 1.7141585040071239, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.292817115783691, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8727496862411499, + "num_tokens": 523021109.0, + "step": 13475 + }, + { + "epoch": 1.7142857142857144, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.228676795959473, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8792828321456909, + "num_tokens": 523056360.0, + "step": 13476 + }, + { + "epoch": 1.714412924564305, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.202909469604492, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8829482197761536, + "num_tokens": 523099536.0, + "step": 13477 + }, + { + "epoch": 1.7145401348428955, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.250163078308105, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8912571668624878, + "num_tokens": 523135280.0, + "step": 13478 + }, + { + "epoch": 1.7146673451214858, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.205206871032715, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8793367147445679, + "num_tokens": 523176607.0, + "step": 13479 + }, + { + "epoch": 1.7147945554000763, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.292780876159668, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.883682370185852, + "num_tokens": 523214872.0, + "step": 13480 + }, + { + "epoch": 1.7149217656786668, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.194269180297852, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8949741721153259, + "num_tokens": 523250631.0, + "step": 13481 + }, + { + "epoch": 1.7150489759572574, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.282208442687988, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8827745318412781, + "num_tokens": 523284378.0, + "step": 13482 + }, + { + "epoch": 1.7151761862358479, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.191122055053711, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8874454498291016, + "num_tokens": 523325659.0, + "step": 13483 + }, + { + "epoch": 1.7153033965144384, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.178593635559082, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8806889057159424, + "num_tokens": 523368833.0, + "step": 13484 + }, + { + "epoch": 1.7154306067930287, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.31831169128418, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8867356777191162, + "num_tokens": 523400486.0, + "step": 13485 + }, + { + "epoch": 1.7155578170716193, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.202676773071289, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8935517072677612, + "num_tokens": 523439199.0, + "step": 13486 + }, + { + "epoch": 1.7156850273502098, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.18996524810791, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8745498657226562, + "num_tokens": 523482690.0, + "step": 13487 + }, + { + "epoch": 1.7158122376288003, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.20810604095459, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8967305421829224, + "num_tokens": 523526495.0, + "step": 13488 + }, + { + "epoch": 1.7159394479073908, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.175642013549805, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8887401819229126, + "num_tokens": 523567474.0, + "step": 13489 + }, + { + "epoch": 1.7160666581859814, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.279788970947266, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8774802684783936, + "num_tokens": 523607749.0, + "step": 13490 + }, + { + "epoch": 1.716193868464572, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.221855163574219, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8860362768173218, + "num_tokens": 523648321.0, + "step": 13491 + }, + { + "epoch": 1.7163210787431624, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.22110652923584, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8851122260093689, + "num_tokens": 523688975.0, + "step": 13492 + }, + { + "epoch": 1.716448289021753, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.228092193603516, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8925715684890747, + "num_tokens": 523729146.0, + "step": 13493 + }, + { + "epoch": 1.7165754993003435, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.183137893676758, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8926843404769897, + "num_tokens": 523769856.0, + "step": 13494 + }, + { + "epoch": 1.716702709578934, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.20048713684082, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8942097425460815, + "num_tokens": 523807268.0, + "step": 13495 + }, + { + "epoch": 1.7168299198575245, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.244192123413086, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.882394552230835, + "num_tokens": 523854603.0, + "step": 13496 + }, + { + "epoch": 1.716957130136115, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.219738960266113, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8982182145118713, + "num_tokens": 523896161.0, + "step": 13497 + }, + { + "epoch": 1.7170843404147056, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.170990943908691, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8933897018432617, + "num_tokens": 523934404.0, + "step": 13498 + }, + { + "epoch": 1.7172115506932961, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.293785095214844, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8799164891242981, + "num_tokens": 523972207.0, + "step": 13499 + }, + { + "epoch": 1.7173387609718866, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.19967269897461, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.9013223052024841, + "num_tokens": 524009088.0, + "step": 13500 + }, + { + "epoch": 1.7174659712504772, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.263436317443848, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8821076154708862, + "num_tokens": 524050637.0, + "step": 13501 + }, + { + "epoch": 1.7175931815290677, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.30316162109375, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8865175843238831, + "num_tokens": 524086292.0, + "step": 13502 + }, + { + "epoch": 1.717720391807658, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.151017189025879, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8823398351669312, + "num_tokens": 524130546.0, + "step": 13503 + }, + { + "epoch": 1.7178476020862485, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.043081283569336e-05, + "grad_norm": 9.20939826965332, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8896709680557251, + "num_tokens": 524168138.0, + "step": 13504 + }, + { + "epoch": 1.717974812364839, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.22814655303955, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.885645866394043, + "num_tokens": 524209222.0, + "step": 13505 + }, + { + "epoch": 1.7181020226434296, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.278855323791504, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8768140077590942, + "num_tokens": 524244122.0, + "step": 13506 + }, + { + "epoch": 1.7182292329220201, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.216025352478027, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8883381485939026, + "num_tokens": 524282014.0, + "step": 13507 + }, + { + "epoch": 1.7183564432006107, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.244077682495117, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8823307752609253, + "num_tokens": 524323868.0, + "step": 13508 + }, + { + "epoch": 1.718483653479201, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.297802925109863, + "learning_rate": 1e-06, + "loss": 0.399, + "mean_token_accuracy": 0.86456298828125, + "num_tokens": 524359809.0, + "step": 13509 + }, + { + "epoch": 1.7186108637577915, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.230046272277832, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8919811248779297, + "num_tokens": 524398290.0, + "step": 13510 + }, + { + "epoch": 1.718738074036382, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.206070899963379, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8847030997276306, + "num_tokens": 524434247.0, + "step": 13511 + }, + { + "epoch": 1.7188652843149725, + "ewc_loss": 0.0167236328125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.388691902160645, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.895880937576294, + "num_tokens": 524464547.0, + "step": 13512 + }, + { + "epoch": 1.718992494593563, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.320302963256836, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8917281627655029, + "num_tokens": 524494587.0, + "step": 13513 + }, + { + "epoch": 1.7191197048721536, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.356863975524902, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8833650946617126, + "num_tokens": 524528481.0, + "step": 13514 + }, + { + "epoch": 1.7192469151507441, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.265298843383789, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8832483887672424, + "num_tokens": 524564890.0, + "step": 13515 + }, + { + "epoch": 1.7193741254293347, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.238697052001953, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.9060068130493164, + "num_tokens": 524599700.0, + "step": 13516 + }, + { + "epoch": 1.7195013357079252, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.236929893493652, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8846754431724548, + "num_tokens": 524639896.0, + "step": 13517 + }, + { + "epoch": 1.7196285459865157, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.253782272338867, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8898272514343262, + "num_tokens": 524675567.0, + "step": 13518 + }, + { + "epoch": 1.7197557562651062, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.189083099365234, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.889167308807373, + "num_tokens": 524718513.0, + "step": 13519 + }, + { + "epoch": 1.7198829665436968, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.193819046020508, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8890131711959839, + "num_tokens": 524759232.0, + "step": 13520 + }, + { + "epoch": 1.7200101768222873, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.251585960388184, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.9051113724708557, + "num_tokens": 524803158.0, + "step": 13521 + }, + { + "epoch": 1.7201373871008778, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.250054359436035, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8807949423789978, + "num_tokens": 524844518.0, + "step": 13522 + }, + { + "epoch": 1.7202645973794684, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.347376823425293, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8816863894462585, + "num_tokens": 524882507.0, + "step": 13523 + }, + { + "epoch": 1.7203918076580589, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.283581733703613, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8805114030838013, + "num_tokens": 524920027.0, + "step": 13524 + }, + { + "epoch": 1.7205190179366494, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.276805877685547, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8913038969039917, + "num_tokens": 524953057.0, + "step": 13525 + }, + { + "epoch": 1.72064622821524, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.235698699951172, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.88742995262146, + "num_tokens": 524991449.0, + "step": 13526 + }, + { + "epoch": 1.7207734384938305, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.323573112487793, + "learning_rate": 1e-06, + "loss": 0.3995, + "mean_token_accuracy": 0.8660370111465454, + "num_tokens": 525031523.0, + "step": 13527 + }, + { + "epoch": 1.7209006487724208, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.187549591064453, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8939389586448669, + "num_tokens": 525071495.0, + "step": 13528 + }, + { + "epoch": 1.7210278590510113, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.246566772460938e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.274506568908691, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8867396116256714, + "num_tokens": 525107611.0, + "step": 13529 + }, + { + "epoch": 1.7211550693296018, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.19033145904541, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8922094106674194, + "num_tokens": 525147889.0, + "step": 13530 + }, + { + "epoch": 1.7212822796081924, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.288055419921875, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8817492723464966, + "num_tokens": 525182827.0, + "step": 13531 + }, + { + "epoch": 1.7214094898867829, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.223164558410645, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.9070075154304504, + "num_tokens": 525219779.0, + "step": 13532 + }, + { + "epoch": 1.7215367001653734, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.275251388549805, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8882843255996704, + "num_tokens": 525257547.0, + "step": 13533 + }, + { + "epoch": 1.7216639104439637, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.195225715637207, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8868482708930969, + "num_tokens": 525294796.0, + "step": 13534 + }, + { + "epoch": 1.7217911207225542, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.203006744384766, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.893378734588623, + "num_tokens": 525336540.0, + "step": 13535 + }, + { + "epoch": 1.7219183310011448, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.254740715026855, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8878923654556274, + "num_tokens": 525375008.0, + "step": 13536 + }, + { + "epoch": 1.7220455412797353, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.252784729003906, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8817445039749146, + "num_tokens": 525415890.0, + "step": 13537 + }, + { + "epoch": 1.7221727515583258, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.316054344177246, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8963598012924194, + "num_tokens": 525451379.0, + "step": 13538 + }, + { + "epoch": 1.7222999618369164, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.180205345153809, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.906761884689331, + "num_tokens": 525491400.0, + "step": 13539 + }, + { + "epoch": 1.7224271721155069, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.246173858642578, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8778790831565857, + "num_tokens": 525532282.0, + "step": 13540 + }, + { + "epoch": 1.7225543823940974, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.200800895690918, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.896507740020752, + "num_tokens": 525569359.0, + "step": 13541 + }, + { + "epoch": 1.722681592672688, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.248051643371582, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8872888088226318, + "num_tokens": 525607502.0, + "step": 13542 + }, + { + "epoch": 1.7228088029512785, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.249403953552246, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8761345148086548, + "num_tokens": 525646843.0, + "step": 13543 + }, + { + "epoch": 1.722936013229869, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.256258964538574, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8909273743629456, + "num_tokens": 525684213.0, + "step": 13544 + }, + { + "epoch": 1.7230632235084595, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.318071365356445, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8741016387939453, + "num_tokens": 525726821.0, + "step": 13545 + }, + { + "epoch": 1.72319043378705, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.218899726867676, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8865682482719421, + "num_tokens": 525763828.0, + "step": 13546 + }, + { + "epoch": 1.7233176440656406, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.2598876953125, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.886607825756073, + "num_tokens": 525801436.0, + "step": 13547 + }, + { + "epoch": 1.7234448543442311, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.28780460357666, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8843337297439575, + "num_tokens": 525835112.0, + "step": 13548 + }, + { + "epoch": 1.7235720646228216, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.232995986938477, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8852002620697021, + "num_tokens": 525875638.0, + "step": 13549 + }, + { + "epoch": 1.7236992749014122, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.174055099487305, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8937891721725464, + "num_tokens": 525919340.0, + "step": 13550 + }, + { + "epoch": 1.7238264851800027, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.311555862426758, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.9019674062728882, + "num_tokens": 525955252.0, + "step": 13551 + }, + { + "epoch": 1.723953695458593, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.241983413696289, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8916710019111633, + "num_tokens": 525995309.0, + "step": 13552 + }, + { + "epoch": 1.7240809057371835, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.36544132232666, + "learning_rate": 1e-06, + "loss": 0.3647, + "mean_token_accuracy": 0.8752457499504089, + "num_tokens": 526036513.0, + "step": 13553 + }, + { + "epoch": 1.724208116015774, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.213740348815918, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8841264247894287, + "num_tokens": 526073953.0, + "step": 13554 + }, + { + "epoch": 1.7243353262943646, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.300897598266602, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8795259594917297, + "num_tokens": 526118574.0, + "step": 13555 + }, + { + "epoch": 1.7244625365729551, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.235539436340332, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8843669891357422, + "num_tokens": 526155334.0, + "step": 13556 + }, + { + "epoch": 1.7245897468515456, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.198423385620117, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8912947773933411, + "num_tokens": 526198819.0, + "step": 13557 + }, + { + "epoch": 1.724716957130136, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.13767147064209, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.892491340637207, + "num_tokens": 526239820.0, + "step": 13558 + }, + { + "epoch": 1.7248441674087265, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.235748291015625, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8915221691131592, + "num_tokens": 526280877.0, + "step": 13559 + }, + { + "epoch": 1.724971377687317, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.240747451782227, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8844802379608154, + "num_tokens": 526321821.0, + "step": 13560 + }, + { + "epoch": 1.7250985879659075, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.226075172424316, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8773953318595886, + "num_tokens": 526360755.0, + "step": 13561 + }, + { + "epoch": 1.725225798244498, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.38803768157959, + "learning_rate": 1e-06, + "loss": 0.3834, + "mean_token_accuracy": 0.8722317218780518, + "num_tokens": 526396021.0, + "step": 13562 + }, + { + "epoch": 1.7253530085230886, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.224162101745605, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8747413754463196, + "num_tokens": 526437404.0, + "step": 13563 + }, + { + "epoch": 1.7254802188016791, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.129300117492676, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8925046324729919, + "num_tokens": 526478338.0, + "step": 13564 + }, + { + "epoch": 1.7256074290802697, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.319223403930664, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8731634020805359, + "num_tokens": 526515116.0, + "step": 13565 + }, + { + "epoch": 1.7257346393588602, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.128167152404785, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8803810477256775, + "num_tokens": 526560203.0, + "step": 13566 + }, + { + "epoch": 1.7258618496374507, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.187013626098633, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.878671407699585, + "num_tokens": 526600432.0, + "step": 13567 + }, + { + "epoch": 1.7259890599160412, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.287392616271973, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8947944045066833, + "num_tokens": 526633796.0, + "step": 13568 + }, + { + "epoch": 1.7261162701946318, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.225954055786133, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8812385201454163, + "num_tokens": 526678609.0, + "step": 13569 + }, + { + "epoch": 1.7262434804732223, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.225128173828125, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8827789425849915, + "num_tokens": 526719475.0, + "step": 13570 + }, + { + "epoch": 1.7263706907518128, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.2763032913208, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8999207019805908, + "num_tokens": 526752292.0, + "step": 13571 + }, + { + "epoch": 1.7264979010304033, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.359220504760742, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8827539086341858, + "num_tokens": 526788011.0, + "step": 13572 + }, + { + "epoch": 1.7266251113089939, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.22840404510498, + "learning_rate": 1e-06, + "loss": 0.361, + "mean_token_accuracy": 0.8791542053222656, + "num_tokens": 526830142.0, + "step": 13573 + }, + { + "epoch": 1.7267523215875844, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.194560050964355, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8893266320228577, + "num_tokens": 526871611.0, + "step": 13574 + }, + { + "epoch": 1.726879531866175, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.214256286621094, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8879939317703247, + "num_tokens": 526913674.0, + "step": 13575 + }, + { + "epoch": 1.7270067421447655, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.363211631774902, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.884166955947876, + "num_tokens": 526948248.0, + "step": 13576 + }, + { + "epoch": 1.7271339524233558, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.442097663879395, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8834909200668335, + "num_tokens": 526980012.0, + "step": 13577 + }, + { + "epoch": 1.7272611627019463, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.220891952514648, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8849958181381226, + "num_tokens": 527021732.0, + "step": 13578 + }, + { + "epoch": 1.7273883729805368, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.25119400024414, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8754781484603882, + "num_tokens": 527061074.0, + "step": 13579 + }, + { + "epoch": 1.7275155832591274, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.269810676574707, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8977155685424805, + "num_tokens": 527098915.0, + "step": 13580 + }, + { + "epoch": 1.7276427935377179, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.213438987731934, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8754160404205322, + "num_tokens": 527142431.0, + "step": 13581 + }, + { + "epoch": 1.7277700038163084, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.276522636413574, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8880448341369629, + "num_tokens": 527181342.0, + "step": 13582 + }, + { + "epoch": 1.7278972140948987, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.225249290466309, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8863406777381897, + "num_tokens": 527220530.0, + "step": 13583 + }, + { + "epoch": 1.7280244243734892, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.311824798583984, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8904808759689331, + "num_tokens": 527253184.0, + "step": 13584 + }, + { + "epoch": 1.7281516346520798, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.125627517700195, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8992555141448975, + "num_tokens": 527297772.0, + "step": 13585 + }, + { + "epoch": 1.7282788449306703, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.250896453857422, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8936423063278198, + "num_tokens": 527338024.0, + "step": 13586 + }, + { + "epoch": 1.7284060552092608, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.274679183959961, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8743666410446167, + "num_tokens": 527374829.0, + "step": 13587 + }, + { + "epoch": 1.7285332654878514, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.234930992126465, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8806737661361694, + "num_tokens": 527412283.0, + "step": 13588 + }, + { + "epoch": 1.7286604757664419, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.271230697631836, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8767908811569214, + "num_tokens": 527453061.0, + "step": 13589 + }, + { + "epoch": 1.7287876860450324, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.191980361938477, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8970277905464172, + "num_tokens": 527491439.0, + "step": 13590 + }, + { + "epoch": 1.728914896323623, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.374922752380371, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8859843015670776, + "num_tokens": 527524027.0, + "step": 13591 + }, + { + "epoch": 1.7290421066022135, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.170201301574707, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8877498507499695, + "num_tokens": 527564715.0, + "step": 13592 + }, + { + "epoch": 1.729169316880804, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.219966888427734, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.8732496500015259, + "num_tokens": 527605502.0, + "step": 13593 + }, + { + "epoch": 1.7292965271593945, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.195728302001953, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8829541802406311, + "num_tokens": 527645212.0, + "step": 13594 + }, + { + "epoch": 1.729423737437985, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.162333488464355, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.899168848991394, + "num_tokens": 527685641.0, + "step": 13595 + }, + { + "epoch": 1.7295509477165756, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.295332908630371, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8803259134292603, + "num_tokens": 527722313.0, + "step": 13596 + }, + { + "epoch": 1.729678157995166, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.15331745147705, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8890766501426697, + "num_tokens": 527767410.0, + "step": 13597 + }, + { + "epoch": 1.7298053682737566, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.156964302062988, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8961215615272522, + "num_tokens": 527810878.0, + "step": 13598 + }, + { + "epoch": 1.7299325785523472, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.19664478302002, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8923239707946777, + "num_tokens": 527847110.0, + "step": 13599 + }, + { + "epoch": 1.7300597888309377, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.266441345214844, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8903264403343201, + "num_tokens": 527880953.0, + "step": 13600 + }, + { + "epoch": 1.730186999109528, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.240743637084961, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8895975351333618, + "num_tokens": 527919320.0, + "step": 13601 + }, + { + "epoch": 1.7303142093881185, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.238438606262207, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8820933103561401, + "num_tokens": 527956156.0, + "step": 13602 + }, + { + "epoch": 1.730441419666709, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.178849220275879, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8821226358413696, + "num_tokens": 528002456.0, + "step": 13603 + }, + { + "epoch": 1.7305686299452996, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.157413482666016, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8798165321350098, + "num_tokens": 528050634.0, + "step": 13604 + }, + { + "epoch": 1.7306958402238901, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.347479820251465, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8756579160690308, + "num_tokens": 528090684.0, + "step": 13605 + }, + { + "epoch": 1.7308230505024806, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.207942008972168, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.890183687210083, + "num_tokens": 528132155.0, + "step": 13606 + }, + { + "epoch": 1.730950260781071, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.142547607421875, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8868736028671265, + "num_tokens": 528179379.0, + "step": 13607 + }, + { + "epoch": 1.7310774710596615, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.387887001037598, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8824974298477173, + "num_tokens": 528211149.0, + "step": 13608 + }, + { + "epoch": 1.731204681338252, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.183857917785645, + "learning_rate": 1e-06, + "loss": 0.3734, + "mean_token_accuracy": 0.8740420341491699, + "num_tokens": 528254766.0, + "step": 13609 + }, + { + "epoch": 1.7313318916168425, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.213528633117676, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8874462842941284, + "num_tokens": 528293152.0, + "step": 13610 + }, + { + "epoch": 1.731459101895433, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.2055025100708, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8926504850387573, + "num_tokens": 528332362.0, + "step": 13611 + }, + { + "epoch": 1.7315863121740236, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.274136543273926, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8980487585067749, + "num_tokens": 528367897.0, + "step": 13612 + }, + { + "epoch": 1.7317135224526141, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.173506736755371, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8943237662315369, + "num_tokens": 528413395.0, + "step": 13613 + }, + { + "epoch": 1.7318407327312046, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.2371826171875, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8912180662155151, + "num_tokens": 528450140.0, + "step": 13614 + }, + { + "epoch": 1.7319679430097952, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.24096965789795, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8838350176811218, + "num_tokens": 528487855.0, + "step": 13615 + }, + { + "epoch": 1.7320951532883857, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.162117958068848, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8889986276626587, + "num_tokens": 528529010.0, + "step": 13616 + }, + { + "epoch": 1.7322223635669762, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.287093162536621, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8839614391326904, + "num_tokens": 528562918.0, + "step": 13617 + }, + { + "epoch": 1.7323495738455668, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.192872047424316, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8941812515258789, + "num_tokens": 528604149.0, + "step": 13618 + }, + { + "epoch": 1.7324767841241573, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.1907320022583, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.891321063041687, + "num_tokens": 528646462.0, + "step": 13619 + }, + { + "epoch": 1.7326039944027478, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.146703720092773, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8885703086853027, + "num_tokens": 528692854.0, + "step": 13620 + }, + { + "epoch": 1.7327312046813383, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.313628196716309, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8910800218582153, + "num_tokens": 528724548.0, + "step": 13621 + }, + { + "epoch": 1.7328584149599289, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.231738090515137, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8781683444976807, + "num_tokens": 528768236.0, + "step": 13622 + }, + { + "epoch": 1.7329856252385194, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.146267890930176, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8864166140556335, + "num_tokens": 528809904.0, + "step": 13623 + }, + { + "epoch": 1.73311283551711, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.230840682983398, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8854387998580933, + "num_tokens": 528849502.0, + "step": 13624 + }, + { + "epoch": 1.7332400457957005, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.258458137512207, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.887295126914978, + "num_tokens": 528884097.0, + "step": 13625 + }, + { + "epoch": 1.7333672560742908, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.252129554748535, + "learning_rate": 1e-06, + "loss": 0.3978, + "mean_token_accuracy": 0.8667121529579163, + "num_tokens": 528925262.0, + "step": 13626 + }, + { + "epoch": 1.7334944663528813, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.157899856567383, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8814693689346313, + "num_tokens": 528968151.0, + "step": 13627 + }, + { + "epoch": 1.7336216766314718, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.249000549316406, + "learning_rate": 1e-06, + "loss": 0.3631, + "mean_token_accuracy": 0.8797624111175537, + "num_tokens": 529005706.0, + "step": 13628 + }, + { + "epoch": 1.7337488869100623, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.280731201171875, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8921912908554077, + "num_tokens": 529038769.0, + "step": 13629 + }, + { + "epoch": 1.7338760971886529, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.337433815002441, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8831592202186584, + "num_tokens": 529072755.0, + "step": 13630 + }, + { + "epoch": 1.7340033074672434, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.19475269317627, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.884591281414032, + "num_tokens": 529111208.0, + "step": 13631 + }, + { + "epoch": 1.7341305177458337, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.179539680480957, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8871444463729858, + "num_tokens": 529153886.0, + "step": 13632 + }, + { + "epoch": 1.7342577280244242, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.230337142944336, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8939425349235535, + "num_tokens": 529194075.0, + "step": 13633 + }, + { + "epoch": 1.7343849383030148, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.22674560546875, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8925384283065796, + "num_tokens": 529228797.0, + "step": 13634 + }, + { + "epoch": 1.7345121485816053, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.210799217224121, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.886684775352478, + "num_tokens": 529268617.0, + "step": 13635 + }, + { + "epoch": 1.7346393588601958, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.202788352966309, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8796089887619019, + "num_tokens": 529314223.0, + "step": 13636 + }, + { + "epoch": 1.7347665691387864, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.24817943572998, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.9020647406578064, + "num_tokens": 529347357.0, + "step": 13637 + }, + { + "epoch": 1.7348937794173769, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.3336820602417, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8835134506225586, + "num_tokens": 529382680.0, + "step": 13638 + }, + { + "epoch": 1.7350209896959674, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.311452865600586, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8849464058876038, + "num_tokens": 529415380.0, + "step": 13639 + }, + { + "epoch": 1.735148199974558, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.176030158996582, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8906676769256592, + "num_tokens": 529459974.0, + "step": 13640 + }, + { + "epoch": 1.7352754102531485, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.277118682861328, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8744856119155884, + "num_tokens": 529500853.0, + "step": 13641 + }, + { + "epoch": 1.735402620531739, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.307229995727539, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8846202492713928, + "num_tokens": 529533211.0, + "step": 13642 + }, + { + "epoch": 1.7355298308103295, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.198051452636719, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.8751533031463623, + "num_tokens": 529571194.0, + "step": 13643 + }, + { + "epoch": 1.73565704108892, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.267580032348633, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8735582232475281, + "num_tokens": 529609848.0, + "step": 13644 + }, + { + "epoch": 1.7357842513675106, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.267163276672363, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8802249431610107, + "num_tokens": 529655836.0, + "step": 13645 + }, + { + "epoch": 1.735911461646101, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.22300910949707, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8841741681098938, + "num_tokens": 529690577.0, + "step": 13646 + }, + { + "epoch": 1.7360386719246916, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.235652923583984, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8862887620925903, + "num_tokens": 529733031.0, + "step": 13647 + }, + { + "epoch": 1.7361658822032822, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.144824981689453, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8822316527366638, + "num_tokens": 529777088.0, + "step": 13648 + }, + { + "epoch": 1.7362930924818727, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.269342422485352, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8880387544631958, + "num_tokens": 529812565.0, + "step": 13649 + }, + { + "epoch": 1.736420302760463, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.204878807067871, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8914198875427246, + "num_tokens": 529853706.0, + "step": 13650 + }, + { + "epoch": 1.7365475130390535, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.208267211914062, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8788043260574341, + "num_tokens": 529897939.0, + "step": 13651 + }, + { + "epoch": 1.736674723317644, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.303353309631348, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8746579885482788, + "num_tokens": 529937405.0, + "step": 13652 + }, + { + "epoch": 1.7368019335962346, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.258268356323242, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8799082040786743, + "num_tokens": 529972802.0, + "step": 13653 + }, + { + "epoch": 1.736929143874825, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.24791145324707, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8950833082199097, + "num_tokens": 530009507.0, + "step": 13654 + }, + { + "epoch": 1.7370563541534156, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.296369552612305, + "learning_rate": 1e-06, + "loss": 0.3908, + "mean_token_accuracy": 0.8683469295501709, + "num_tokens": 530051176.0, + "step": 13655 + }, + { + "epoch": 1.737183564432006, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.345880508422852, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8806633949279785, + "num_tokens": 530088808.0, + "step": 13656 + }, + { + "epoch": 1.7373107747105965, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.222427368164062, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8944454789161682, + "num_tokens": 530126632.0, + "step": 13657 + }, + { + "epoch": 1.737437984989187, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.263690948486328, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8903512954711914, + "num_tokens": 530168631.0, + "step": 13658 + }, + { + "epoch": 1.7375651952677775, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.219528198242188, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8791224956512451, + "num_tokens": 530210985.0, + "step": 13659 + }, + { + "epoch": 1.737692405546368, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.151922225952148, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8882530927658081, + "num_tokens": 530253944.0, + "step": 13660 + }, + { + "epoch": 1.7378196158249586, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.38383960723877, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8742852210998535, + "num_tokens": 530290426.0, + "step": 13661 + }, + { + "epoch": 1.7379468261035491, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.17434310913086, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.9008466005325317, + "num_tokens": 530331903.0, + "step": 13662 + }, + { + "epoch": 1.7380740363821396, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.23757553100586, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8855793476104736, + "num_tokens": 530372114.0, + "step": 13663 + }, + { + "epoch": 1.7382012466607302, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.268648147583008, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8827264308929443, + "num_tokens": 530412970.0, + "step": 13664 + }, + { + "epoch": 1.7383284569393207, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.165185928344727, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8944324851036072, + "num_tokens": 530455396.0, + "step": 13665 + }, + { + "epoch": 1.7384556672179112, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.188132286071777, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8916829228401184, + "num_tokens": 530496360.0, + "step": 13666 + }, + { + "epoch": 1.7385828774965018, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.216370582580566, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.9044405817985535, + "num_tokens": 530535029.0, + "step": 13667 + }, + { + "epoch": 1.7387100877750923, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.188488006591797, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8788490891456604, + "num_tokens": 530581320.0, + "step": 13668 + }, + { + "epoch": 1.7388372980536828, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.269657135009766, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8874196410179138, + "num_tokens": 530621387.0, + "step": 13669 + }, + { + "epoch": 1.7389645083322733, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.210309028625488, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8983748555183411, + "num_tokens": 530658376.0, + "step": 13670 + }, + { + "epoch": 1.7390917186108639, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.321861267089844, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8808197975158691, + "num_tokens": 530699090.0, + "step": 13671 + }, + { + "epoch": 1.7392189288894544, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.260208129882812, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8826486468315125, + "num_tokens": 530737606.0, + "step": 13672 + }, + { + "epoch": 1.739346139168045, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.38560676574707, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8839008212089539, + "num_tokens": 530767594.0, + "step": 13673 + }, + { + "epoch": 1.7394733494466355, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.229318618774414, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8891324400901794, + "num_tokens": 530806134.0, + "step": 13674 + }, + { + "epoch": 1.7396005597252258, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.256584167480469, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8938502669334412, + "num_tokens": 530844527.0, + "step": 13675 + }, + { + "epoch": 1.7397277700038163, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.279423713684082, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8739869594573975, + "num_tokens": 530884200.0, + "step": 13676 + }, + { + "epoch": 1.7398549802824068, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.221413612365723, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8809351325035095, + "num_tokens": 530925654.0, + "step": 13677 + }, + { + "epoch": 1.7399821905609973, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.231191635131836, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8808549642562866, + "num_tokens": 530965885.0, + "step": 13678 + }, + { + "epoch": 1.7401094008395879, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.244966506958008, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8875450491905212, + "num_tokens": 531002545.0, + "step": 13679 + }, + { + "epoch": 1.7402366111181784, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.152647018432617, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8822994232177734, + "num_tokens": 531045081.0, + "step": 13680 + }, + { + "epoch": 1.7403638213967687, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.171520233154297, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9019995331764221, + "num_tokens": 531083410.0, + "step": 13681 + }, + { + "epoch": 1.7404910316753592, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.279463768005371, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.9043726921081543, + "num_tokens": 531116210.0, + "step": 13682 + }, + { + "epoch": 1.7406182419539498, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.302985191345215, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8838543891906738, + "num_tokens": 531153479.0, + "step": 13683 + }, + { + "epoch": 1.7407454522325403, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.2278470993042, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8904339075088501, + "num_tokens": 531192510.0, + "step": 13684 + }, + { + "epoch": 1.7408726625111308, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.289536476135254, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8954477310180664, + "num_tokens": 531229319.0, + "step": 13685 + }, + { + "epoch": 1.7409998727897213, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.243351936340332, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.8739479184150696, + "num_tokens": 531275739.0, + "step": 13686 + }, + { + "epoch": 1.7411270830683119, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.361530303955078, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8800022006034851, + "num_tokens": 531310038.0, + "step": 13687 + }, + { + "epoch": 1.7412542933469024, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.256805419921875, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8861187100410461, + "num_tokens": 531352950.0, + "step": 13688 + }, + { + "epoch": 1.741381503625493, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.210554122924805, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8864579796791077, + "num_tokens": 531394563.0, + "step": 13689 + }, + { + "epoch": 1.7415087139040835, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.234410285949707, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.9052683115005493, + "num_tokens": 531428667.0, + "step": 13690 + }, + { + "epoch": 1.741635924182674, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.22698974609375, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8919029831886292, + "num_tokens": 531466627.0, + "step": 13691 + }, + { + "epoch": 1.7417631344612645, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.26986312866211, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8863691091537476, + "num_tokens": 531500908.0, + "step": 13692 + }, + { + "epoch": 1.741890344739855, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.239145278930664, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8799534440040588, + "num_tokens": 531542963.0, + "step": 13693 + }, + { + "epoch": 1.7420175550184456, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.180824279785156, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8910959959030151, + "num_tokens": 531584866.0, + "step": 13694 + }, + { + "epoch": 1.742144765297036, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.227090835571289, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8851512670516968, + "num_tokens": 531623874.0, + "step": 13695 + }, + { + "epoch": 1.7422719755756266, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.300477981567383, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.897144079208374, + "num_tokens": 531657266.0, + "step": 13696 + }, + { + "epoch": 1.7423991858542172, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.234423637390137, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.886675238609314, + "num_tokens": 531693471.0, + "step": 13697 + }, + { + "epoch": 1.7425263961328077, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.169551849365234, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.89628666639328, + "num_tokens": 531739193.0, + "step": 13698 + }, + { + "epoch": 1.742653606411398, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.303936004638672, + "learning_rate": 1e-06, + "loss": 0.3676, + "mean_token_accuracy": 0.872965395450592, + "num_tokens": 531776169.0, + "step": 13699 + }, + { + "epoch": 1.7427808166899885, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.306665420532227, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8847965002059937, + "num_tokens": 531812101.0, + "step": 13700 + }, + { + "epoch": 1.742908026968579, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.30319881439209, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8902738094329834, + "num_tokens": 531849306.0, + "step": 13701 + }, + { + "epoch": 1.7430352372471696, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.212777137756348, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.879244327545166, + "num_tokens": 531893021.0, + "step": 13702 + }, + { + "epoch": 1.74316244752576, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.205889701843262, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8947333097457886, + "num_tokens": 531934926.0, + "step": 13703 + }, + { + "epoch": 1.7432896578043506, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.208968162536621, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9006702303886414, + "num_tokens": 531970765.0, + "step": 13704 + }, + { + "epoch": 1.743416868082941, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.20926284790039, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.888248085975647, + "num_tokens": 532008952.0, + "step": 13705 + }, + { + "epoch": 1.7435440783615315, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.24763011932373, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8986522555351257, + "num_tokens": 532047077.0, + "step": 13706 + }, + { + "epoch": 1.743671288640122, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.296518325805664, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8748141527175903, + "num_tokens": 532082107.0, + "step": 13707 + }, + { + "epoch": 1.7437984989187125, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.3297700881958, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8911863565444946, + "num_tokens": 532113596.0, + "step": 13708 + }, + { + "epoch": 1.743925709197303, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.151881217956543, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9005355834960938, + "num_tokens": 532150956.0, + "step": 13709 + }, + { + "epoch": 1.7440529194758936, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.292673110961914, + "learning_rate": 1e-06, + "loss": 0.3875, + "mean_token_accuracy": 0.8715186715126038, + "num_tokens": 532191865.0, + "step": 13710 + }, + { + "epoch": 1.744180129754484, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.186182975769043, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8925451636314392, + "num_tokens": 532236229.0, + "step": 13711 + }, + { + "epoch": 1.7443073400330746, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.227123260498047, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8894925117492676, + "num_tokens": 532271088.0, + "step": 13712 + }, + { + "epoch": 1.7444345503116652, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.295856475830078, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8823309540748596, + "num_tokens": 532309200.0, + "step": 13713 + }, + { + "epoch": 1.7445617605902557, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.153340339660645, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.89996337890625, + "num_tokens": 532347388.0, + "step": 13714 + }, + { + "epoch": 1.7446889708688462, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.319791793823242, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8872771263122559, + "num_tokens": 532382407.0, + "step": 13715 + }, + { + "epoch": 1.7448161811474368, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.212285041809082, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8977028727531433, + "num_tokens": 532425230.0, + "step": 13716 + }, + { + "epoch": 1.7449433914260273, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.313104629516602, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8909609913825989, + "num_tokens": 532455651.0, + "step": 13717 + }, + { + "epoch": 1.7450706017046178, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.231419563293457, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8853698372840881, + "num_tokens": 532497556.0, + "step": 13718 + }, + { + "epoch": 1.7451978119832083, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.180069923400879, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9009150862693787, + "num_tokens": 532537561.0, + "step": 13719 + }, + { + "epoch": 1.7453250222617989, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.260431289672852, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8771153688430786, + "num_tokens": 532578565.0, + "step": 13720 + }, + { + "epoch": 1.7454522325403894, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.253795623779297, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.88996422290802, + "num_tokens": 532621510.0, + "step": 13721 + }, + { + "epoch": 1.74557944281898, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.104745864868164, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9088020920753479, + "num_tokens": 532663177.0, + "step": 13722 + }, + { + "epoch": 1.7457066530975704, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.23229694366455, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8956781625747681, + "num_tokens": 532705373.0, + "step": 13723 + }, + { + "epoch": 1.7458338633761608, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.365777969360352, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8805610537528992, + "num_tokens": 532737575.0, + "step": 13724 + }, + { + "epoch": 1.7459610736547513, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.26938247680664, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8800560235977173, + "num_tokens": 532778214.0, + "step": 13725 + }, + { + "epoch": 1.7460882839333418, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.150439262390137, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9014987945556641, + "num_tokens": 532825994.0, + "step": 13726 + }, + { + "epoch": 1.7462154942119323, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.386246681213379, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8871580958366394, + "num_tokens": 532864192.0, + "step": 13727 + }, + { + "epoch": 1.7463427044905229, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.259949684143066, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8893904685974121, + "num_tokens": 532902667.0, + "step": 13728 + }, + { + "epoch": 1.7464699147691134, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.189467430114746, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8991758823394775, + "num_tokens": 532940625.0, + "step": 13729 + }, + { + "epoch": 1.7465971250477037, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.24227237701416, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8916273713111877, + "num_tokens": 532978220.0, + "step": 13730 + }, + { + "epoch": 1.7467243353262942, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.254045486450195, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.892729640007019, + "num_tokens": 533017533.0, + "step": 13731 + }, + { + "epoch": 1.7468515456048848, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.23955249786377, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.882621705532074, + "num_tokens": 533058684.0, + "step": 13732 + }, + { + "epoch": 1.7469787558834753, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.222688674926758, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8961658477783203, + "num_tokens": 533100739.0, + "step": 13733 + }, + { + "epoch": 1.7471059661620658, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.25598430633545, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8932210206985474, + "num_tokens": 533138077.0, + "step": 13734 + }, + { + "epoch": 1.7472331764406563, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.262866973876953, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8956799507141113, + "num_tokens": 533172585.0, + "step": 13735 + }, + { + "epoch": 1.7473603867192469, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.234193801879883, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8868793249130249, + "num_tokens": 533211723.0, + "step": 13736 + }, + { + "epoch": 1.7474875969978374, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.203471183776855, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8881807923316956, + "num_tokens": 533253623.0, + "step": 13737 + }, + { + "epoch": 1.747614807276428, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.270576477050781, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8944451212882996, + "num_tokens": 533290328.0, + "step": 13738 + }, + { + "epoch": 1.7477420175550185, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.324755668640137, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8905576467514038, + "num_tokens": 533323544.0, + "step": 13739 + }, + { + "epoch": 1.747869227833609, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.238343238830566, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8758000135421753, + "num_tokens": 533366353.0, + "step": 13740 + }, + { + "epoch": 1.7479964381121995, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.167304992675781, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8854242563247681, + "num_tokens": 533410010.0, + "step": 13741 + }, + { + "epoch": 1.74812364839079, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.224648475646973, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8867273926734924, + "num_tokens": 533449294.0, + "step": 13742 + }, + { + "epoch": 1.7482508586693806, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.289579391479492, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8938145637512207, + "num_tokens": 533479048.0, + "step": 13743 + }, + { + "epoch": 1.748378068947971, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.176417350769043, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8932784795761108, + "num_tokens": 533520472.0, + "step": 13744 + }, + { + "epoch": 1.7485052792265616, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.23312759399414, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.89748215675354, + "num_tokens": 533555071.0, + "step": 13745 + }, + { + "epoch": 1.7486324895051522, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.266709327697754, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.9029585123062134, + "num_tokens": 533589671.0, + "step": 13746 + }, + { + "epoch": 1.7487596997837427, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.343949317932129, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8751109838485718, + "num_tokens": 533625004.0, + "step": 13747 + }, + { + "epoch": 1.748886910062333, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.213325500488281, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8909718990325928, + "num_tokens": 533664175.0, + "step": 13748 + }, + { + "epoch": 1.7490141203409235, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.304571151733398, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8827071189880371, + "num_tokens": 533704649.0, + "step": 13749 + }, + { + "epoch": 1.749141330619514, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.308151245117188, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8941935300827026, + "num_tokens": 533738698.0, + "step": 13750 + }, + { + "epoch": 1.7492685408981046, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.148193359375, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.881404459476471, + "num_tokens": 533779923.0, + "step": 13751 + }, + { + "epoch": 1.749395751176695, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.379528045654297, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8926413059234619, + "num_tokens": 533811260.0, + "step": 13752 + }, + { + "epoch": 1.7495229614552856, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.176018714904785, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8869320154190063, + "num_tokens": 533851804.0, + "step": 13753 + }, + { + "epoch": 1.749650171733876, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.341761589050293, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8871047496795654, + "num_tokens": 533883031.0, + "step": 13754 + }, + { + "epoch": 1.7497773820124665, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.292688369750977, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8948796391487122, + "num_tokens": 533918453.0, + "step": 13755 + }, + { + "epoch": 1.749904592291057, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.250885963439941, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8863753080368042, + "num_tokens": 533954600.0, + "step": 13756 + }, + { + "epoch": 1.7500318025696475, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.271353721618652, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8851100206375122, + "num_tokens": 533996732.0, + "step": 13757 + }, + { + "epoch": 1.750159012848238, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.325155258178711, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8748349547386169, + "num_tokens": 534030351.0, + "step": 13758 + }, + { + "epoch": 1.7502862231268286, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.267212867736816, + "learning_rate": 1e-06, + "loss": 0.3714, + "mean_token_accuracy": 0.8768270015716553, + "num_tokens": 534071236.0, + "step": 13759 + }, + { + "epoch": 1.750413433405419, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.227386474609375, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8947024941444397, + "num_tokens": 534112270.0, + "step": 13760 + }, + { + "epoch": 1.7505406436840096, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.26145076751709, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8879247307777405, + "num_tokens": 534146523.0, + "step": 13761 + }, + { + "epoch": 1.7506678539626002, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.265560150146484, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8888399600982666, + "num_tokens": 534181816.0, + "step": 13762 + }, + { + "epoch": 1.7507950642411907, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.357582092285156, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.8715680837631226, + "num_tokens": 534220928.0, + "step": 13763 + }, + { + "epoch": 1.7509222745197812, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.198902130126953, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8856774568557739, + "num_tokens": 534260135.0, + "step": 13764 + }, + { + "epoch": 1.7510494847983717, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.282832145690918, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8852989077568054, + "num_tokens": 534293960.0, + "step": 13765 + }, + { + "epoch": 1.7511766950769623, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.340537071228027, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8899612426757812, + "num_tokens": 534327609.0, + "step": 13766 + }, + { + "epoch": 1.7513039053555528, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.267144203186035, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8766768574714661, + "num_tokens": 534362489.0, + "step": 13767 + }, + { + "epoch": 1.7514311156341433, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.286989212036133, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8905341625213623, + "num_tokens": 534399111.0, + "step": 13768 + }, + { + "epoch": 1.7515583259127339, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.237923622131348, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.894481897354126, + "num_tokens": 534438642.0, + "step": 13769 + }, + { + "epoch": 1.7516855361913244, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.282458305358887, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8826063871383667, + "num_tokens": 534478302.0, + "step": 13770 + }, + { + "epoch": 1.751812746469915, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.211647033691406, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8900927305221558, + "num_tokens": 534520988.0, + "step": 13771 + }, + { + "epoch": 1.7519399567485054, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.286104202270508, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8979083299636841, + "num_tokens": 534556237.0, + "step": 13772 + }, + { + "epoch": 1.7520671670270958, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.393525123596191, + "learning_rate": 1e-06, + "loss": 0.3885, + "mean_token_accuracy": 0.871084451675415, + "num_tokens": 534591543.0, + "step": 13773 + }, + { + "epoch": 1.7521943773056863, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.227214813232422, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8973369002342224, + "num_tokens": 534631085.0, + "step": 13774 + }, + { + "epoch": 1.7523215875842768, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.28722095489502, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8950211405754089, + "num_tokens": 534667043.0, + "step": 13775 + }, + { + "epoch": 1.7524487978628673, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.297184944152832, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8816855549812317, + "num_tokens": 534707849.0, + "step": 13776 + }, + { + "epoch": 1.7525760081414579, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.309310913085938, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8925156593322754, + "num_tokens": 534741169.0, + "step": 13777 + }, + { + "epoch": 1.7527032184200484, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.270655632019043, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8880366683006287, + "num_tokens": 534782246.0, + "step": 13778 + }, + { + "epoch": 1.7528304286986387, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.245073318481445, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8997795581817627, + "num_tokens": 534819095.0, + "step": 13779 + }, + { + "epoch": 1.7529576389772292, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.309287071228027, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8808947801589966, + "num_tokens": 534858406.0, + "step": 13780 + }, + { + "epoch": 1.7530848492558198, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.249709129333496, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.890829861164093, + "num_tokens": 534899778.0, + "step": 13781 + }, + { + "epoch": 1.7532120595344103, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.208844184875488, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8944085240364075, + "num_tokens": 534941332.0, + "step": 13782 + }, + { + "epoch": 1.7533392698130008, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.291855812072754, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.888168454170227, + "num_tokens": 534976933.0, + "step": 13783 + }, + { + "epoch": 1.7534664800915913, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.282270431518555, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8817819356918335, + "num_tokens": 535014415.0, + "step": 13784 + }, + { + "epoch": 1.7535936903701819, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.286556243896484, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8824911117553711, + "num_tokens": 535050323.0, + "step": 13785 + }, + { + "epoch": 1.7537209006487724, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.327683448791504, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8951060771942139, + "num_tokens": 535086093.0, + "step": 13786 + }, + { + "epoch": 1.753848110927363, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.323397636413574, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8775882720947266, + "num_tokens": 535120716.0, + "step": 13787 + }, + { + "epoch": 1.7539753212059535, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.174615859985352, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8967263698577881, + "num_tokens": 535163075.0, + "step": 13788 + }, + { + "epoch": 1.754102531484544, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.216645240783691, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8886522054672241, + "num_tokens": 535201064.0, + "step": 13789 + }, + { + "epoch": 1.7542297417631345, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.229466438293457, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.9045979380607605, + "num_tokens": 535235818.0, + "step": 13790 + }, + { + "epoch": 1.754356952041725, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.102806091308594, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.909230649471283, + "num_tokens": 535281627.0, + "step": 13791 + }, + { + "epoch": 1.7544841623203156, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.19555377960205, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8993469476699829, + "num_tokens": 535321874.0, + "step": 13792 + }, + { + "epoch": 1.754611372598906, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.299317359924316, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8973252177238464, + "num_tokens": 535355402.0, + "step": 13793 + }, + { + "epoch": 1.7547385828774966, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.18774700164795, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8925819396972656, + "num_tokens": 535396764.0, + "step": 13794 + }, + { + "epoch": 1.7548657931560872, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.34743881225586, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8804712295532227, + "num_tokens": 535432198.0, + "step": 13795 + }, + { + "epoch": 1.7549930034346777, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.414728164672852, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8813991546630859, + "num_tokens": 535464774.0, + "step": 13796 + }, + { + "epoch": 1.755120213713268, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.109561920166016, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8891591429710388, + "num_tokens": 535508538.0, + "step": 13797 + }, + { + "epoch": 1.7552474239918585, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.318140983581543, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8936683535575867, + "num_tokens": 535547328.0, + "step": 13798 + }, + { + "epoch": 1.755374634270449, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.168999671936035, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8959232568740845, + "num_tokens": 535590634.0, + "step": 13799 + }, + { + "epoch": 1.7555018445490396, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.250736236572266, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8877060413360596, + "num_tokens": 535629098.0, + "step": 13800 + }, + { + "epoch": 1.75562905482763, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.398186683654785, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8855820298194885, + "num_tokens": 535661487.0, + "step": 13801 + }, + { + "epoch": 1.7557562651062206, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.34605598449707, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8825128674507141, + "num_tokens": 535693560.0, + "step": 13802 + }, + { + "epoch": 1.755883475384811, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.287495613098145, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8693289756774902, + "num_tokens": 535738385.0, + "step": 13803 + }, + { + "epoch": 1.7560106856634015, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.283650398254395, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8866287469863892, + "num_tokens": 535775346.0, + "step": 13804 + }, + { + "epoch": 1.756137895941992, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.252901077270508, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8965232372283936, + "num_tokens": 535815743.0, + "step": 13805 + }, + { + "epoch": 1.7562651062205825, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.371736526489258, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8833814859390259, + "num_tokens": 535849993.0, + "step": 13806 + }, + { + "epoch": 1.756392316499173, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.23497486114502, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8728060722351074, + "num_tokens": 535890627.0, + "step": 13807 + }, + { + "epoch": 1.7565195267777636, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.225706100463867, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8882369995117188, + "num_tokens": 535928648.0, + "step": 13808 + }, + { + "epoch": 1.756646737056354, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.08791446685791, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.889140784740448, + "num_tokens": 535976067.0, + "step": 13809 + }, + { + "epoch": 1.7567739473349446, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.250319480895996, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8808585405349731, + "num_tokens": 536014303.0, + "step": 13810 + }, + { + "epoch": 1.7569011576135352, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.196490287780762, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9045364260673523, + "num_tokens": 536050451.0, + "step": 13811 + }, + { + "epoch": 1.7570283678921257, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.152740478515625, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.9007408022880554, + "num_tokens": 536091100.0, + "step": 13812 + }, + { + "epoch": 1.7571555781707162, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.185526847839355, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9055761694908142, + "num_tokens": 536126591.0, + "step": 13813 + }, + { + "epoch": 1.7572827884493067, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.25601863861084, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8951670527458191, + "num_tokens": 536162095.0, + "step": 13814 + }, + { + "epoch": 1.7574099987278973, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.130295753479004, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8951816558837891, + "num_tokens": 536204319.0, + "step": 13815 + }, + { + "epoch": 1.7575372090064878, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.169520378112793, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8978578448295593, + "num_tokens": 536248358.0, + "step": 13816 + }, + { + "epoch": 1.7576644192850783, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.303677558898926, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8900899887084961, + "num_tokens": 536283722.0, + "step": 13817 + }, + { + "epoch": 1.7577916295636689, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.18807315826416, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.886433482170105, + "num_tokens": 536327344.0, + "step": 13818 + }, + { + "epoch": 1.7579188398422594, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.375507354736328, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8964113593101501, + "num_tokens": 536361576.0, + "step": 13819 + }, + { + "epoch": 1.75804605012085, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.226533889770508, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8910607695579529, + "num_tokens": 536399987.0, + "step": 13820 + }, + { + "epoch": 1.7581732603994404, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.231880187988281, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8843249082565308, + "num_tokens": 536437436.0, + "step": 13821 + }, + { + "epoch": 1.7583004706780307, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.28153133392334, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8845514059066772, + "num_tokens": 536476752.0, + "step": 13822 + }, + { + "epoch": 1.7584276809566213, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.179252624511719, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8749925494194031, + "num_tokens": 536521123.0, + "step": 13823 + }, + { + "epoch": 1.7585548912352118, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.347290992736816, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8896415829658508, + "num_tokens": 536556550.0, + "step": 13824 + }, + { + "epoch": 1.7586821015138023, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.234977722167969, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8937993049621582, + "num_tokens": 536592445.0, + "step": 13825 + }, + { + "epoch": 1.7588093117923929, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.291537284851074, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8995376825332642, + "num_tokens": 536625540.0, + "step": 13826 + }, + { + "epoch": 1.7589365220709834, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.364990234375, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8914092183113098, + "num_tokens": 536654624.0, + "step": 13827 + }, + { + "epoch": 1.7590637323495737, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.280226707458496, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8903930187225342, + "num_tokens": 536694450.0, + "step": 13828 + }, + { + "epoch": 1.7591909426281642, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.300681114196777, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8879435658454895, + "num_tokens": 536732597.0, + "step": 13829 + }, + { + "epoch": 1.7593181529067548, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.185803413391113, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8892874717712402, + "num_tokens": 536774117.0, + "step": 13830 + }, + { + "epoch": 1.7594453631853453, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.37047004699707, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8797636032104492, + "num_tokens": 536808998.0, + "step": 13831 + }, + { + "epoch": 1.7595725734639358, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.248602867126465, + "learning_rate": 1e-06, + "loss": 0.3576, + "mean_token_accuracy": 0.8805794715881348, + "num_tokens": 536852794.0, + "step": 13832 + }, + { + "epoch": 1.7596997837425263, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.251119613647461, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8796408772468567, + "num_tokens": 536894206.0, + "step": 13833 + }, + { + "epoch": 1.7598269940211169, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.269618034362793, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8865785002708435, + "num_tokens": 536930151.0, + "step": 13834 + }, + { + "epoch": 1.7599542042997074, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.229341506958008, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9075119495391846, + "num_tokens": 536969691.0, + "step": 13835 + }, + { + "epoch": 1.760081414578298, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.29474925994873, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8895331621170044, + "num_tokens": 537004794.0, + "step": 13836 + }, + { + "epoch": 1.7602086248568884, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.284472465515137, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8787667751312256, + "num_tokens": 537045417.0, + "step": 13837 + }, + { + "epoch": 1.760335835135479, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.346489906311035, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8801040649414062, + "num_tokens": 537076903.0, + "step": 13838 + }, + { + "epoch": 1.7604630454140695, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.267304420471191, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8929409384727478, + "num_tokens": 537115044.0, + "step": 13839 + }, + { + "epoch": 1.76059025569266, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.294044494628906, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8875482082366943, + "num_tokens": 537148107.0, + "step": 13840 + }, + { + "epoch": 1.7607174659712506, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.283735275268555, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8834261894226074, + "num_tokens": 537189144.0, + "step": 13841 + }, + { + "epoch": 1.760844676249841, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.150105476379395, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8960559368133545, + "num_tokens": 537232810.0, + "step": 13842 + }, + { + "epoch": 1.7609718865284316, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.346156120300293, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.8731775283813477, + "num_tokens": 537271220.0, + "step": 13843 + }, + { + "epoch": 1.7610990968070221, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.250710487365723, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8872194290161133, + "num_tokens": 537310086.0, + "step": 13844 + }, + { + "epoch": 1.7612263070856127, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.20977783203125, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8841474652290344, + "num_tokens": 537348632.0, + "step": 13845 + }, + { + "epoch": 1.761353517364203, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.322503089904785, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8788812160491943, + "num_tokens": 537386155.0, + "step": 13846 + }, + { + "epoch": 1.7614807276427935, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.242936134338379, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8897642493247986, + "num_tokens": 537425261.0, + "step": 13847 + }, + { + "epoch": 1.761607937921384, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.227680206298828, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8829522132873535, + "num_tokens": 537464662.0, + "step": 13848 + }, + { + "epoch": 1.7617351481999746, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.191756248474121, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8833277821540833, + "num_tokens": 537505654.0, + "step": 13849 + }, + { + "epoch": 1.761862358478565, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.469200134277344, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8831987380981445, + "num_tokens": 537544498.0, + "step": 13850 + }, + { + "epoch": 1.7619895687571556, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.196343421936035, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9005376696586609, + "num_tokens": 537579343.0, + "step": 13851 + }, + { + "epoch": 1.762116779035746, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.208222389221191, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8953160047531128, + "num_tokens": 537615662.0, + "step": 13852 + }, + { + "epoch": 1.7622439893143365, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.23377799987793, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8964813947677612, + "num_tokens": 537652066.0, + "step": 13853 + }, + { + "epoch": 1.762371199592927, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.241971969604492, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.903935432434082, + "num_tokens": 537684955.0, + "step": 13854 + }, + { + "epoch": 1.7624984098715175, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.324795722961426, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8888198137283325, + "num_tokens": 537719428.0, + "step": 13855 + }, + { + "epoch": 1.762625620150108, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.182744979858398, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8898773193359375, + "num_tokens": 537762702.0, + "step": 13856 + }, + { + "epoch": 1.7627528304286986, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.22656536102295, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8980377316474915, + "num_tokens": 537799662.0, + "step": 13857 + }, + { + "epoch": 1.762880040707289, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.224207878112793, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8804560899734497, + "num_tokens": 537842712.0, + "step": 13858 + }, + { + "epoch": 1.7630072509858796, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.356945037841797, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8918558359146118, + "num_tokens": 537878750.0, + "step": 13859 + }, + { + "epoch": 1.7631344612644702, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.207565307617188, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8935575485229492, + "num_tokens": 537919723.0, + "step": 13860 + }, + { + "epoch": 1.7632616715430607, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.208894729614258, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8886762857437134, + "num_tokens": 537963158.0, + "step": 13861 + }, + { + "epoch": 1.7633888818216512, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.292798042297363, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8749278783798218, + "num_tokens": 537999461.0, + "step": 13862 + }, + { + "epoch": 1.7635160921002417, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.248384475708008, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8996509909629822, + "num_tokens": 538037774.0, + "step": 13863 + }, + { + "epoch": 1.7636433023788323, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.206130981445312, + "learning_rate": 1e-06, + "loss": 0.3784, + "mean_token_accuracy": 0.8742495179176331, + "num_tokens": 538080129.0, + "step": 13864 + }, + { + "epoch": 1.7637705126574228, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.283553123474121, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8975908756256104, + "num_tokens": 538115215.0, + "step": 13865 + }, + { + "epoch": 1.7638977229360133, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.217633247375488, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8875432014465332, + "num_tokens": 538157346.0, + "step": 13866 + }, + { + "epoch": 1.7640249332146039, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.284636497497559, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8907797932624817, + "num_tokens": 538194747.0, + "step": 13867 + }, + { + "epoch": 1.7641521434931944, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.186800956726074, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8884545564651489, + "num_tokens": 538240785.0, + "step": 13868 + }, + { + "epoch": 1.764279353771785, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.333069801330566, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8846175670623779, + "num_tokens": 538280312.0, + "step": 13869 + }, + { + "epoch": 1.7644065640503754, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.273269653320312, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9008667469024658, + "num_tokens": 538315782.0, + "step": 13870 + }, + { + "epoch": 1.7645337743289657, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.157649993896484, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8972345590591431, + "num_tokens": 538360260.0, + "step": 13871 + }, + { + "epoch": 1.7646609846075563, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.226414680480957, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8970280289649963, + "num_tokens": 538404147.0, + "step": 13872 + }, + { + "epoch": 1.7647881948861468, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.320991516113281, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.891513466835022, + "num_tokens": 538439249.0, + "step": 13873 + }, + { + "epoch": 1.7649154051647373, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.239113807678223, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8983827829360962, + "num_tokens": 538476282.0, + "step": 13874 + }, + { + "epoch": 1.7650426154433279, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.20336627960205, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8914331197738647, + "num_tokens": 538515257.0, + "step": 13875 + }, + { + "epoch": 1.7651698257219184, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.259663581848145, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8841941356658936, + "num_tokens": 538552915.0, + "step": 13876 + }, + { + "epoch": 1.7652970360005087, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.225228309631348, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8820769190788269, + "num_tokens": 538592667.0, + "step": 13877 + }, + { + "epoch": 1.7654242462790992, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.454253196716309, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8911647796630859, + "num_tokens": 538626364.0, + "step": 13878 + }, + { + "epoch": 1.7655514565576897, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.145683288574219, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8898543119430542, + "num_tokens": 538672328.0, + "step": 13879 + }, + { + "epoch": 1.7656786668362803, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.272929191589355, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8826757073402405, + "num_tokens": 538706921.0, + "step": 13880 + }, + { + "epoch": 1.7658058771148708, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.347885131835938, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8893786668777466, + "num_tokens": 538742433.0, + "step": 13881 + }, + { + "epoch": 1.7659330873934613, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.255791664123535, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.896170973777771, + "num_tokens": 538776394.0, + "step": 13882 + }, + { + "epoch": 1.7660602976720519, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.213722229003906, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8824479579925537, + "num_tokens": 538811416.0, + "step": 13883 + }, + { + "epoch": 1.7661875079506424, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.224074363708496, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.878555178642273, + "num_tokens": 538854006.0, + "step": 13884 + }, + { + "epoch": 1.766314718229233, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.248970985412598, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8901962041854858, + "num_tokens": 538894879.0, + "step": 13885 + }, + { + "epoch": 1.7664419285078234, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.22747802734375, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8875834941864014, + "num_tokens": 538935909.0, + "step": 13886 + }, + { + "epoch": 1.766569138786414, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.231263160705566, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8912941217422485, + "num_tokens": 538979963.0, + "step": 13887 + }, + { + "epoch": 1.7666963490650045, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.208597183227539, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8945932388305664, + "num_tokens": 539021662.0, + "step": 13888 + }, + { + "epoch": 1.766823559343595, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.303657531738281, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8863412141799927, + "num_tokens": 539061440.0, + "step": 13889 + }, + { + "epoch": 1.7669507696221856, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.267570495605469, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8898348808288574, + "num_tokens": 539100150.0, + "step": 13890 + }, + { + "epoch": 1.767077979900776, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.279253959655762, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8910988569259644, + "num_tokens": 539134736.0, + "step": 13891 + }, + { + "epoch": 1.7672051901793666, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.261442184448242, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.8684034943580627, + "num_tokens": 539172931.0, + "step": 13892 + }, + { + "epoch": 1.7673324004579571, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.287483215332031, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8911261558532715, + "num_tokens": 539206168.0, + "step": 13893 + }, + { + "epoch": 1.7674596107365477, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.253656387329102, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.88772052526474, + "num_tokens": 539241983.0, + "step": 13894 + }, + { + "epoch": 1.767586821015138, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.198843955993652, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8791247606277466, + "num_tokens": 539286254.0, + "step": 13895 + }, + { + "epoch": 1.7677140312937285, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.240923881530762, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.888725757598877, + "num_tokens": 539330243.0, + "step": 13896 + }, + { + "epoch": 1.767841241572319, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.283954620361328, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8742734789848328, + "num_tokens": 539367578.0, + "step": 13897 + }, + { + "epoch": 1.7679684518509096, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.32543659210205, + "learning_rate": 1e-06, + "loss": 0.3908, + "mean_token_accuracy": 0.8691686391830444, + "num_tokens": 539407442.0, + "step": 13898 + }, + { + "epoch": 1.7680956621295, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.275493621826172, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8711060285568237, + "num_tokens": 539450555.0, + "step": 13899 + }, + { + "epoch": 1.7682228724080906, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.440131187438965, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8832093477249146, + "num_tokens": 539483019.0, + "step": 13900 + }, + { + "epoch": 1.768350082686681, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.273015975952148, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8908566236495972, + "num_tokens": 539518169.0, + "step": 13901 + }, + { + "epoch": 1.7684772929652715, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.193267822265625, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8794510960578918, + "num_tokens": 539558402.0, + "step": 13902 + }, + { + "epoch": 1.768604503243862, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.25426197052002, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8846633434295654, + "num_tokens": 539598890.0, + "step": 13903 + }, + { + "epoch": 1.7687317135224525, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.194547653198242, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8970801830291748, + "num_tokens": 539638193.0, + "step": 13904 + }, + { + "epoch": 1.768858923801043, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.250837326049805, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8876410722732544, + "num_tokens": 539678217.0, + "step": 13905 + }, + { + "epoch": 1.7689861340796336, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.18961238861084, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8791410326957703, + "num_tokens": 539726218.0, + "step": 13906 + }, + { + "epoch": 1.769113344358224, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.38486099243164, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8785126209259033, + "num_tokens": 539758919.0, + "step": 13907 + }, + { + "epoch": 1.7692405546368146, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.227095603942871, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8812946677207947, + "num_tokens": 539800895.0, + "step": 13908 + }, + { + "epoch": 1.7693677649154052, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.16817855834961, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8932120203971863, + "num_tokens": 539844355.0, + "step": 13909 + }, + { + "epoch": 1.7694949751939957, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.286808967590332, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8775745630264282, + "num_tokens": 539887927.0, + "step": 13910 + }, + { + "epoch": 1.7696221854725862, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.247114181518555, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8918192386627197, + "num_tokens": 539928265.0, + "step": 13911 + }, + { + "epoch": 1.7697493957511767, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.24012279510498, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8851596713066101, + "num_tokens": 539966986.0, + "step": 13912 + }, + { + "epoch": 1.7698766060297673, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.290364265441895, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8827754855155945, + "num_tokens": 540006946.0, + "step": 13913 + }, + { + "epoch": 1.7700038163083578, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.315962791442871, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8831558227539062, + "num_tokens": 540042546.0, + "step": 13914 + }, + { + "epoch": 1.7701310265869483, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.239287376403809, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8890880346298218, + "num_tokens": 540082353.0, + "step": 13915 + }, + { + "epoch": 1.7702582368655388, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.225722312927246, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8931846618652344, + "num_tokens": 540121701.0, + "step": 13916 + }, + { + "epoch": 1.7703854471441294, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.247365951538086, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.895721971988678, + "num_tokens": 540161438.0, + "step": 13917 + }, + { + "epoch": 1.77051265742272, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.199312210083008, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.901847779750824, + "num_tokens": 540198606.0, + "step": 13918 + }, + { + "epoch": 1.7706398677013102, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.331976890563965, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8845784664154053, + "num_tokens": 540239200.0, + "step": 13919 + }, + { + "epoch": 1.7707670779799007, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.149523735046387, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8877456188201904, + "num_tokens": 540287540.0, + "step": 13920 + }, + { + "epoch": 1.7708942882584913, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.325788497924805, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8836426138877869, + "num_tokens": 540326708.0, + "step": 13921 + }, + { + "epoch": 1.7710214985370818, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.20738410949707, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.9042633771896362, + "num_tokens": 540369376.0, + "step": 13922 + }, + { + "epoch": 1.7711487088156723, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.274239540100098, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8993422985076904, + "num_tokens": 540405699.0, + "step": 13923 + }, + { + "epoch": 1.7712759190942629, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.338526725769043, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8928966522216797, + "num_tokens": 540439356.0, + "step": 13924 + }, + { + "epoch": 1.7714031293728534, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.280257225036621, + "learning_rate": 1e-06, + "loss": 0.3999, + "mean_token_accuracy": 0.8645788431167603, + "num_tokens": 540480856.0, + "step": 13925 + }, + { + "epoch": 1.7715303396514437, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.229684829711914, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8941439390182495, + "num_tokens": 540520774.0, + "step": 13926 + }, + { + "epoch": 1.7716575499300342, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.318638801574707, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8807699084281921, + "num_tokens": 540561690.0, + "step": 13927 + }, + { + "epoch": 1.7717847602086247, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.241995811462402, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8838887810707092, + "num_tokens": 540599410.0, + "step": 13928 + }, + { + "epoch": 1.7719119704872153, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.250707626342773, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.897983193397522, + "num_tokens": 540637765.0, + "step": 13929 + }, + { + "epoch": 1.7720391807658058, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.37547779083252, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8744314312934875, + "num_tokens": 540675606.0, + "step": 13930 + }, + { + "epoch": 1.7721663910443963, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.33447551727295, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8794982433319092, + "num_tokens": 540715932.0, + "step": 13931 + }, + { + "epoch": 1.7722936013229869, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.269208908081055, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8859518766403198, + "num_tokens": 540752662.0, + "step": 13932 + }, + { + "epoch": 1.7724208116015774, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.243704795837402, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.9061294198036194, + "num_tokens": 540791942.0, + "step": 13933 + }, + { + "epoch": 1.772548021880168, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.318683624267578, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8807429671287537, + "num_tokens": 540832291.0, + "step": 13934 + }, + { + "epoch": 1.7726752321587584, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.27542781829834, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8839312791824341, + "num_tokens": 540872092.0, + "step": 13935 + }, + { + "epoch": 1.772802442437349, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.251747131347656, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8860687017440796, + "num_tokens": 540912110.0, + "step": 13936 + }, + { + "epoch": 1.7729296527159395, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.237993240356445, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8933091163635254, + "num_tokens": 540948432.0, + "step": 13937 + }, + { + "epoch": 1.77305686299453, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.279260635375977, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8954035639762878, + "num_tokens": 540987944.0, + "step": 13938 + }, + { + "epoch": 1.7731840732731206, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.17711353302002, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8968865871429443, + "num_tokens": 541034677.0, + "step": 13939 + }, + { + "epoch": 1.773311283551711, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.302501678466797, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8816224932670593, + "num_tokens": 541080666.0, + "step": 13940 + }, + { + "epoch": 1.7734384938303016, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.196588516235352, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.9000446200370789, + "num_tokens": 541120984.0, + "step": 13941 + }, + { + "epoch": 1.7735657041088921, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.274764060974121, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8906723856925964, + "num_tokens": 541158283.0, + "step": 13942 + }, + { + "epoch": 1.7736929143874827, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.310128211975098, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8858835697174072, + "num_tokens": 541201445.0, + "step": 13943 + }, + { + "epoch": 1.773820124666073, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.271947860717773, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.9037215709686279, + "num_tokens": 541240523.0, + "step": 13944 + }, + { + "epoch": 1.7739473349446635, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.2409029006958, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8976770639419556, + "num_tokens": 541277009.0, + "step": 13945 + }, + { + "epoch": 1.774074545223254, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.329072952270508, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8859469294548035, + "num_tokens": 541315089.0, + "step": 13946 + }, + { + "epoch": 1.7742017555018446, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.259763717651367, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8876582384109497, + "num_tokens": 541356100.0, + "step": 13947 + }, + { + "epoch": 1.774328965780435, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.314129829406738, + "learning_rate": 1e-06, + "loss": 0.3576, + "mean_token_accuracy": 0.8799619674682617, + "num_tokens": 541395891.0, + "step": 13948 + }, + { + "epoch": 1.7744561760590256, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.194133758544922, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8939209580421448, + "num_tokens": 541437253.0, + "step": 13949 + }, + { + "epoch": 1.774583386337616, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.330924987792969, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.881750226020813, + "num_tokens": 541471588.0, + "step": 13950 + }, + { + "epoch": 1.7747105966162064, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.330474853515625, + "learning_rate": 1e-06, + "loss": 0.3783, + "mean_token_accuracy": 0.8740943670272827, + "num_tokens": 541508808.0, + "step": 13951 + }, + { + "epoch": 1.774837806894797, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.049041748046875e-05, + "grad_norm": 9.208456039428711, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.884576141834259, + "num_tokens": 541554535.0, + "step": 13952 + }, + { + "epoch": 1.7749650171733875, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.237874031066895, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8931981325149536, + "num_tokens": 541594681.0, + "step": 13953 + }, + { + "epoch": 1.775092227451978, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.289429664611816, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8883163332939148, + "num_tokens": 541636859.0, + "step": 13954 + }, + { + "epoch": 1.7752194377305686, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.330574035644531, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.877302885055542, + "num_tokens": 541670724.0, + "step": 13955 + }, + { + "epoch": 1.775346648009159, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.387444496154785, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8870830535888672, + "num_tokens": 541701420.0, + "step": 13956 + }, + { + "epoch": 1.7754738582877496, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.267669677734375, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8903375864028931, + "num_tokens": 541739772.0, + "step": 13957 + }, + { + "epoch": 1.7756010685663401, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.234345436096191, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8955268859863281, + "num_tokens": 541780063.0, + "step": 13958 + }, + { + "epoch": 1.7757282788449307, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.335921287536621, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8856664896011353, + "num_tokens": 541817196.0, + "step": 13959 + }, + { + "epoch": 1.7758554891235212, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.158475875854492, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8999750018119812, + "num_tokens": 541857037.0, + "step": 13960 + }, + { + "epoch": 1.7759826994021117, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.266514778137207, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8915812373161316, + "num_tokens": 541895452.0, + "step": 13961 + }, + { + "epoch": 1.7761099096807023, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.485772132873535, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8918377757072449, + "num_tokens": 541920950.0, + "step": 13962 + }, + { + "epoch": 1.7762371199592928, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.183196067810059, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.9038791656494141, + "num_tokens": 541966156.0, + "step": 13963 + }, + { + "epoch": 1.7763643302378833, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.248035430908203, + "learning_rate": 1e-06, + "loss": 0.3681, + "mean_token_accuracy": 0.874180257320404, + "num_tokens": 542011048.0, + "step": 13964 + }, + { + "epoch": 1.7764915405164738, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.253949165344238, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.9004567861557007, + "num_tokens": 542049804.0, + "step": 13965 + }, + { + "epoch": 1.7766187507950644, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.251152992248535, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.895545482635498, + "num_tokens": 542091931.0, + "step": 13966 + }, + { + "epoch": 1.776745961073655, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.356657028198242, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8804656267166138, + "num_tokens": 542128430.0, + "step": 13967 + }, + { + "epoch": 1.7768731713522452, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.246675491333008, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8960909843444824, + "num_tokens": 542164311.0, + "step": 13968 + }, + { + "epoch": 1.7770003816308357, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.242026329040527, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.891598641872406, + "num_tokens": 542204767.0, + "step": 13969 + }, + { + "epoch": 1.7771275919094263, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.302849769592285, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8840325474739075, + "num_tokens": 542245211.0, + "step": 13970 + }, + { + "epoch": 1.7772548021880168, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.28471851348877, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8817211389541626, + "num_tokens": 542281466.0, + "step": 13971 + }, + { + "epoch": 1.7773820124666073, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.300566673278809, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.881646990776062, + "num_tokens": 542321497.0, + "step": 13972 + }, + { + "epoch": 1.7775092227451978, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.277061462402344, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8953230977058411, + "num_tokens": 542356126.0, + "step": 13973 + }, + { + "epoch": 1.7776364330237884, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.319250106811523, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8736083507537842, + "num_tokens": 542395001.0, + "step": 13974 + }, + { + "epoch": 1.7777636433023787, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.295272827148438, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.886090874671936, + "num_tokens": 542431283.0, + "step": 13975 + }, + { + "epoch": 1.7778908535809692, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.329227447509766, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8785079717636108, + "num_tokens": 542469995.0, + "step": 13976 + }, + { + "epoch": 1.7780180638595597, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.313620567321777, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8853700160980225, + "num_tokens": 542509616.0, + "step": 13977 + }, + { + "epoch": 1.7781452741381503, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.240018844604492, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8996074199676514, + "num_tokens": 542549972.0, + "step": 13978 + }, + { + "epoch": 1.7782724844167408, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.23425006866455, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8912363052368164, + "num_tokens": 542593755.0, + "step": 13979 + }, + { + "epoch": 1.7783996946953313, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.26601791381836, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8886095881462097, + "num_tokens": 542634577.0, + "step": 13980 + }, + { + "epoch": 1.7785269049739219, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.361539840698242, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.890733540058136, + "num_tokens": 542673447.0, + "step": 13981 + }, + { + "epoch": 1.7786541152525124, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.178089141845703, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8970641493797302, + "num_tokens": 542711203.0, + "step": 13982 + }, + { + "epoch": 1.778781325531103, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.281571388244629, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8895982503890991, + "num_tokens": 542753465.0, + "step": 13983 + }, + { + "epoch": 1.7789085358096934, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.186410903930664, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8821585178375244, + "num_tokens": 542798212.0, + "step": 13984 + }, + { + "epoch": 1.779035746088284, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.350586891174316, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8872634768486023, + "num_tokens": 542836904.0, + "step": 13985 + }, + { + "epoch": 1.7791629563668745, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.27332592010498, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8770829439163208, + "num_tokens": 542879240.0, + "step": 13986 + }, + { + "epoch": 1.779290166645465, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.278573989868164, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8955382108688354, + "num_tokens": 542909477.0, + "step": 13987 + }, + { + "epoch": 1.7794173769240555, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.320828437805176, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8784348368644714, + "num_tokens": 542948776.0, + "step": 13988 + }, + { + "epoch": 1.779544587202646, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.308389663696289, + "learning_rate": 1e-06, + "loss": 0.3689, + "mean_token_accuracy": 0.875147819519043, + "num_tokens": 542993989.0, + "step": 13989 + }, + { + "epoch": 1.7796717974812366, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.263710021972656, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8919633030891418, + "num_tokens": 543032847.0, + "step": 13990 + }, + { + "epoch": 1.7797990077598271, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.251577377319336, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8877395391464233, + "num_tokens": 543072386.0, + "step": 13991 + }, + { + "epoch": 1.7799262180384177, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.371180534362793, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8933873176574707, + "num_tokens": 543110406.0, + "step": 13992 + }, + { + "epoch": 1.780053428317008, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.25583553314209, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8857094049453735, + "num_tokens": 543153657.0, + "step": 13993 + }, + { + "epoch": 1.7801806385955985, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.24395751953125, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8883150815963745, + "num_tokens": 543192775.0, + "step": 13994 + }, + { + "epoch": 1.780307848874189, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.311505317687988, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8796111345291138, + "num_tokens": 543230215.0, + "step": 13995 + }, + { + "epoch": 1.7804350591527796, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.284605979919434, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8900052309036255, + "num_tokens": 543270423.0, + "step": 13996 + }, + { + "epoch": 1.78056226943137, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.326399803161621, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8999391198158264, + "num_tokens": 543302506.0, + "step": 13997 + }, + { + "epoch": 1.7806894797099606, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.212889671325684, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.9011557698249817, + "num_tokens": 543338471.0, + "step": 13998 + }, + { + "epoch": 1.780816689988551, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.373826026916504, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8937000036239624, + "num_tokens": 543373568.0, + "step": 13999 + }, + { + "epoch": 1.7809439002671414, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.258285522460938, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8934867978096008, + "num_tokens": 543408688.0, + "step": 14000 + }, + { + "epoch": 1.781071110545732, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.321833610534668, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8793433904647827, + "num_tokens": 543449494.0, + "step": 14001 + }, + { + "epoch": 1.7811983208243225, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.144815444946289, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.899196207523346, + "num_tokens": 543496432.0, + "step": 14002 + }, + { + "epoch": 1.781325531102913, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.312110900878906, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8899878263473511, + "num_tokens": 543533043.0, + "step": 14003 + }, + { + "epoch": 1.7814527413815036, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.180535316467285, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8924063444137573, + "num_tokens": 543575752.0, + "step": 14004 + }, + { + "epoch": 1.781579951660094, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.304040908813477, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.887134313583374, + "num_tokens": 543617586.0, + "step": 14005 + }, + { + "epoch": 1.7817071619386846, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.33129596710205, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8890587091445923, + "num_tokens": 543650177.0, + "step": 14006 + }, + { + "epoch": 1.7818343722172751, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.218463897705078, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8914160132408142, + "num_tokens": 543690505.0, + "step": 14007 + }, + { + "epoch": 1.7819615824958657, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.39177417755127, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8740918040275574, + "num_tokens": 543725249.0, + "step": 14008 + }, + { + "epoch": 1.7820887927744562, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.36231803894043, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8804073929786682, + "num_tokens": 543762618.0, + "step": 14009 + }, + { + "epoch": 1.7822160030530467, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.309492111206055, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8871051669120789, + "num_tokens": 543801144.0, + "step": 14010 + }, + { + "epoch": 1.7823432133316373, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.325568199157715, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8704935312271118, + "num_tokens": 543836942.0, + "step": 14011 + }, + { + "epoch": 1.7824704236102278, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.369073867797852, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8875755071640015, + "num_tokens": 543872590.0, + "step": 14012 + }, + { + "epoch": 1.7825976338888183, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.129511833190918, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8771089315414429, + "num_tokens": 543916844.0, + "step": 14013 + }, + { + "epoch": 1.7827248441674088, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.3677339553833, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8863782286643982, + "num_tokens": 543952799.0, + "step": 14014 + }, + { + "epoch": 1.7828520544459994, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.191435813903809, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8902627825737, + "num_tokens": 544000449.0, + "step": 14015 + }, + { + "epoch": 1.78297926472459, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.242135047912598, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8899997472763062, + "num_tokens": 544042900.0, + "step": 14016 + }, + { + "epoch": 1.7831064750031802, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.336228370666504, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8999664783477783, + "num_tokens": 544079833.0, + "step": 14017 + }, + { + "epoch": 1.7832336852817707, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.304571151733398, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8779648542404175, + "num_tokens": 544117337.0, + "step": 14018 + }, + { + "epoch": 1.7833608955603613, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.33066463470459, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8871340751647949, + "num_tokens": 544153537.0, + "step": 14019 + }, + { + "epoch": 1.7834881058389518, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.261874198913574, + "learning_rate": 1e-06, + "loss": 0.3671, + "mean_token_accuracy": 0.8741350173950195, + "num_tokens": 544194833.0, + "step": 14020 + }, + { + "epoch": 1.7836153161175423, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.254899024963379, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8946532011032104, + "num_tokens": 544233191.0, + "step": 14021 + }, + { + "epoch": 1.7837425263961328, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.382546424865723, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.880641520023346, + "num_tokens": 544269613.0, + "step": 14022 + }, + { + "epoch": 1.7838697366747234, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.32479476928711, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8873755931854248, + "num_tokens": 544305580.0, + "step": 14023 + }, + { + "epoch": 1.7839969469533137, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.222208023071289, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8857634663581848, + "num_tokens": 544344006.0, + "step": 14024 + }, + { + "epoch": 1.7841241572319042, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.420153617858887, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8792009949684143, + "num_tokens": 544380136.0, + "step": 14025 + }, + { + "epoch": 1.7842513675104947, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.235092163085938, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.873492956161499, + "num_tokens": 544422588.0, + "step": 14026 + }, + { + "epoch": 1.7843785777890853, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.253082275390625, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8919699192047119, + "num_tokens": 544462232.0, + "step": 14027 + }, + { + "epoch": 1.7845057880676758, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.313371658325195, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8832850456237793, + "num_tokens": 544498668.0, + "step": 14028 + }, + { + "epoch": 1.7846329983462663, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.281688690185547, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8838478922843933, + "num_tokens": 544531340.0, + "step": 14029 + }, + { + "epoch": 1.7847602086248568, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.243183135986328, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.878844678401947, + "num_tokens": 544578098.0, + "step": 14030 + }, + { + "epoch": 1.7848874189034474, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.325300216674805, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8759071826934814, + "num_tokens": 544614866.0, + "step": 14031 + }, + { + "epoch": 1.785014629182038, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.312850952148438, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8850387334823608, + "num_tokens": 544653132.0, + "step": 14032 + }, + { + "epoch": 1.7851418394606284, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.196550369262695, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8921902775764465, + "num_tokens": 544694115.0, + "step": 14033 + }, + { + "epoch": 1.785269049739219, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.317940711975098, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8862990736961365, + "num_tokens": 544728127.0, + "step": 14034 + }, + { + "epoch": 1.7853962600178095, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.216432571411133, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9025632739067078, + "num_tokens": 544763034.0, + "step": 14035 + }, + { + "epoch": 1.7855234702964, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.241631507873535, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8871965408325195, + "num_tokens": 544799824.0, + "step": 14036 + }, + { + "epoch": 1.7856506805749905, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.261037826538086, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8917011618614197, + "num_tokens": 544837397.0, + "step": 14037 + }, + { + "epoch": 1.785777890853581, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.216686248779297, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8881524801254272, + "num_tokens": 544878591.0, + "step": 14038 + }, + { + "epoch": 1.7859051011321716, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.349322319030762, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8862210512161255, + "num_tokens": 544913831.0, + "step": 14039 + }, + { + "epoch": 1.7860323114107621, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.330820083618164, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8864998817443848, + "num_tokens": 544951475.0, + "step": 14040 + }, + { + "epoch": 1.7861595216893527, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.251009941101074, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8860310316085815, + "num_tokens": 544989606.0, + "step": 14041 + }, + { + "epoch": 1.786286731967943, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.34695816040039, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8930394053459167, + "num_tokens": 545025087.0, + "step": 14042 + }, + { + "epoch": 1.7864139422465335, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.375350952148438, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8809839487075806, + "num_tokens": 545058965.0, + "step": 14043 + }, + { + "epoch": 1.786541152525124, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.32939624786377, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8908178806304932, + "num_tokens": 545091961.0, + "step": 14044 + }, + { + "epoch": 1.7866683628037145, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.224150657653809, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8961817026138306, + "num_tokens": 545132722.0, + "step": 14045 + }, + { + "epoch": 1.786795573082305, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.428342819213867, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8814481496810913, + "num_tokens": 545170617.0, + "step": 14046 + }, + { + "epoch": 1.7869227833608956, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.35881233215332, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.8734012246131897, + "num_tokens": 545210946.0, + "step": 14047 + }, + { + "epoch": 1.787049993639486, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.270910263061523, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8862954378128052, + "num_tokens": 545248929.0, + "step": 14048 + }, + { + "epoch": 1.7871772039180764, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.220629692077637, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8972446918487549, + "num_tokens": 545290732.0, + "step": 14049 + }, + { + "epoch": 1.787304414196667, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.262568473815918, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.897979736328125, + "num_tokens": 545326691.0, + "step": 14050 + }, + { + "epoch": 1.7874316244752575, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.24387264251709, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8979200124740601, + "num_tokens": 545364917.0, + "step": 14051 + }, + { + "epoch": 1.787558834753848, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.24725341796875, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8906182646751404, + "num_tokens": 545406190.0, + "step": 14052 + }, + { + "epoch": 1.7876860450324386, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.233232498168945, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.9030880928039551, + "num_tokens": 545442115.0, + "step": 14053 + }, + { + "epoch": 1.787813255311029, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.257482528686523, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.889033317565918, + "num_tokens": 545480904.0, + "step": 14054 + }, + { + "epoch": 1.7879404655896196, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.263219833374023, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8866552710533142, + "num_tokens": 545518181.0, + "step": 14055 + }, + { + "epoch": 1.7880676758682101, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.215668678283691, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8931255340576172, + "num_tokens": 545559935.0, + "step": 14056 + }, + { + "epoch": 1.7881948861468007, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.264223098754883, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8966919183731079, + "num_tokens": 545598405.0, + "step": 14057 + }, + { + "epoch": 1.7883220964253912, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.30074405670166, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8903064727783203, + "num_tokens": 545628990.0, + "step": 14058 + }, + { + "epoch": 1.7884493067039817, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.300145149230957, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.9008413553237915, + "num_tokens": 545664978.0, + "step": 14059 + }, + { + "epoch": 1.7885765169825723, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.253286361694336, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8877988457679749, + "num_tokens": 545704124.0, + "step": 14060 + }, + { + "epoch": 1.7887037272611628, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.298760414123535, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.901477575302124, + "num_tokens": 545744232.0, + "step": 14061 + }, + { + "epoch": 1.7888309375397533, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.1721830368042, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9062044024467468, + "num_tokens": 545787542.0, + "step": 14062 + }, + { + "epoch": 1.7889581478183438, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.298333168029785, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8816888332366943, + "num_tokens": 545825329.0, + "step": 14063 + }, + { + "epoch": 1.7890853580969344, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.182428359985352, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8940011262893677, + "num_tokens": 545870221.0, + "step": 14064 + }, + { + "epoch": 1.789212568375525, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.360513687133789, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8829278945922852, + "num_tokens": 545902715.0, + "step": 14065 + }, + { + "epoch": 1.7893397786541152, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.257007598876953, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8810369968414307, + "num_tokens": 545945867.0, + "step": 14066 + }, + { + "epoch": 1.7894669889327057, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.26745891571045, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8839755654335022, + "num_tokens": 545984617.0, + "step": 14067 + }, + { + "epoch": 1.7895941992112963, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.349332809448242, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8831626772880554, + "num_tokens": 546016913.0, + "step": 14068 + }, + { + "epoch": 1.7897214094898868, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.293225288391113, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8803970813751221, + "num_tokens": 546054958.0, + "step": 14069 + }, + { + "epoch": 1.7898486197684773, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.234369277954102, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.894705057144165, + "num_tokens": 546092193.0, + "step": 14070 + }, + { + "epoch": 1.7899758300470678, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.254724502563477, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8837834596633911, + "num_tokens": 546133040.0, + "step": 14071 + }, + { + "epoch": 1.7901030403256584, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.25422477722168, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8807613253593445, + "num_tokens": 546174465.0, + "step": 14072 + }, + { + "epoch": 1.7902302506042487, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.285860061645508, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8987108469009399, + "num_tokens": 546214789.0, + "step": 14073 + }, + { + "epoch": 1.7903574608828392, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.293195724487305, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8895244598388672, + "num_tokens": 546251300.0, + "step": 14074 + }, + { + "epoch": 1.7904846711614297, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.32249641418457, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8853229880332947, + "num_tokens": 546286179.0, + "step": 14075 + }, + { + "epoch": 1.7906118814400203, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.22642707824707, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8915886878967285, + "num_tokens": 546329525.0, + "step": 14076 + }, + { + "epoch": 1.7907390917186108, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.458380699157715, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8971115350723267, + "num_tokens": 546365333.0, + "step": 14077 + }, + { + "epoch": 1.7908663019972013, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.27767562866211, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8932402729988098, + "num_tokens": 546405187.0, + "step": 14078 + }, + { + "epoch": 1.7909935122757918, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.337021827697754, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8793496489524841, + "num_tokens": 546442101.0, + "step": 14079 + }, + { + "epoch": 1.7911207225543824, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.26270866394043, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9076932668685913, + "num_tokens": 546475044.0, + "step": 14080 + }, + { + "epoch": 1.791247932832973, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.2992582321167, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8905773758888245, + "num_tokens": 546513710.0, + "step": 14081 + }, + { + "epoch": 1.7913751431115634, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.308046340942383, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8917902708053589, + "num_tokens": 546554926.0, + "step": 14082 + }, + { + "epoch": 1.791502353390154, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.31947135925293, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8805100917816162, + "num_tokens": 546591882.0, + "step": 14083 + }, + { + "epoch": 1.7916295636687445, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.393229484558105, + "learning_rate": 1e-06, + "loss": 0.3869, + "mean_token_accuracy": 0.8678305745124817, + "num_tokens": 546627090.0, + "step": 14084 + }, + { + "epoch": 1.791756773947335, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.300498962402344, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8842923045158386, + "num_tokens": 546662662.0, + "step": 14085 + }, + { + "epoch": 1.7918839842259255, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.346623420715332, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8951335549354553, + "num_tokens": 546699101.0, + "step": 14086 + }, + { + "epoch": 1.792011194504516, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.293359756469727, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.880867063999176, + "num_tokens": 546732787.0, + "step": 14087 + }, + { + "epoch": 1.7921384047831066, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.252578735351562, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8844015002250671, + "num_tokens": 546776710.0, + "step": 14088 + }, + { + "epoch": 1.7922656150616971, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.207085609436035, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8963203430175781, + "num_tokens": 546815847.0, + "step": 14089 + }, + { + "epoch": 1.7923928253402877, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.370063781738281, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8900247812271118, + "num_tokens": 546849372.0, + "step": 14090 + }, + { + "epoch": 1.792520035618878, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.299009323120117, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.895930826663971, + "num_tokens": 546881784.0, + "step": 14091 + }, + { + "epoch": 1.7926472458974685, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.223713874816895, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8881468176841736, + "num_tokens": 546926082.0, + "step": 14092 + }, + { + "epoch": 1.792774456176059, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.297935485839844, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8901680707931519, + "num_tokens": 546960759.0, + "step": 14093 + }, + { + "epoch": 1.7929016664546495, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.244085311889648, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8857330083847046, + "num_tokens": 546999172.0, + "step": 14094 + }, + { + "epoch": 1.79302887673324, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.225228309631348, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.89630126953125, + "num_tokens": 547034821.0, + "step": 14095 + }, + { + "epoch": 1.7931560870118306, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.483213424682617, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8931241035461426, + "num_tokens": 547064603.0, + "step": 14096 + }, + { + "epoch": 1.793283297290421, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.281623840332031, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8844915628433228, + "num_tokens": 547104148.0, + "step": 14097 + }, + { + "epoch": 1.7934105075690114, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.349870681762695, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8924394845962524, + "num_tokens": 547136625.0, + "step": 14098 + }, + { + "epoch": 1.793537717847602, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.208124160766602, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.889053463935852, + "num_tokens": 547182072.0, + "step": 14099 + }, + { + "epoch": 1.7936649281261925, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.2641019821167, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8853834867477417, + "num_tokens": 547231288.0, + "step": 14100 + }, + { + "epoch": 1.793792138404783, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.274104118347168, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8865721225738525, + "num_tokens": 547270458.0, + "step": 14101 + }, + { + "epoch": 1.7939193486833735, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.477076530456543, + "learning_rate": 1e-06, + "loss": 0.3847, + "mean_token_accuracy": 0.8688685894012451, + "num_tokens": 547308824.0, + "step": 14102 + }, + { + "epoch": 1.794046558961964, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.291013717651367, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8840823173522949, + "num_tokens": 547350104.0, + "step": 14103 + }, + { + "epoch": 1.7941737692405546, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.327570915222168, + "learning_rate": 1e-06, + "loss": 0.3895, + "mean_token_accuracy": 0.8696629405021667, + "num_tokens": 547387669.0, + "step": 14104 + }, + { + "epoch": 1.7943009795191451, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.366312026977539, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8862311840057373, + "num_tokens": 547421888.0, + "step": 14105 + }, + { + "epoch": 1.7944281897977357, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.287567138671875, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.893216073513031, + "num_tokens": 547457892.0, + "step": 14106 + }, + { + "epoch": 1.7945554000763262, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.268668174743652, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8900964260101318, + "num_tokens": 547492993.0, + "step": 14107 + }, + { + "epoch": 1.7946826103549167, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.279891967773438, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.876598596572876, + "num_tokens": 547532676.0, + "step": 14108 + }, + { + "epoch": 1.7948098206335072, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.26595401763916, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8919069766998291, + "num_tokens": 547565322.0, + "step": 14109 + }, + { + "epoch": 1.7949370309120978, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.234685897827148, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.900583028793335, + "num_tokens": 547607439.0, + "step": 14110 + }, + { + "epoch": 1.7950642411906883, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.277307510375977, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8810125589370728, + "num_tokens": 547648975.0, + "step": 14111 + }, + { + "epoch": 1.7951914514692788, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.337203979492188, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8836568593978882, + "num_tokens": 547689564.0, + "step": 14112 + }, + { + "epoch": 1.7953186617478694, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.264680862426758, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8869732022285461, + "num_tokens": 547729823.0, + "step": 14113 + }, + { + "epoch": 1.7954458720264599, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.37177562713623, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8927798271179199, + "num_tokens": 547767798.0, + "step": 14114 + }, + { + "epoch": 1.7955730823050502, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.218764305114746, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8929935693740845, + "num_tokens": 547806829.0, + "step": 14115 + }, + { + "epoch": 1.7957002925836407, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.260396003723145, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8866269588470459, + "num_tokens": 547846556.0, + "step": 14116 + }, + { + "epoch": 1.7958275028622313, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.259967803955078, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8926893472671509, + "num_tokens": 547891437.0, + "step": 14117 + }, + { + "epoch": 1.7959547131408218, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.314496040344238, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8748364448547363, + "num_tokens": 547927159.0, + "step": 14118 + }, + { + "epoch": 1.7960819234194123, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.257514953613281, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8822586536407471, + "num_tokens": 547967713.0, + "step": 14119 + }, + { + "epoch": 1.7962091336980028, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.286020278930664, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8890946507453918, + "num_tokens": 548006103.0, + "step": 14120 + }, + { + "epoch": 1.7963363439765934, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.13147258758545, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.900222659111023, + "num_tokens": 548050254.0, + "step": 14121 + }, + { + "epoch": 1.7964635542551837, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.254319190979004, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8879235982894897, + "num_tokens": 548088642.0, + "step": 14122 + }, + { + "epoch": 1.7965907645337742, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.3121976852417, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8941041231155396, + "num_tokens": 548123243.0, + "step": 14123 + }, + { + "epoch": 1.7967179748123647, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.200861930847168, + "learning_rate": 1e-06, + "loss": 0.3996, + "mean_token_accuracy": 0.8675645589828491, + "num_tokens": 548172724.0, + "step": 14124 + }, + { + "epoch": 1.7968451850909553, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.188780784606934, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8963024616241455, + "num_tokens": 548213937.0, + "step": 14125 + }, + { + "epoch": 1.7969723953695458, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.277376174926758, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8859521746635437, + "num_tokens": 548252194.0, + "step": 14126 + }, + { + "epoch": 1.7970996056481363, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.262822151184082, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8925355672836304, + "num_tokens": 548289327.0, + "step": 14127 + }, + { + "epoch": 1.7972268159267268, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.205168724060059, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.897367000579834, + "num_tokens": 548330250.0, + "step": 14128 + }, + { + "epoch": 1.7973540262053174, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.237322807312012, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8859841227531433, + "num_tokens": 548374009.0, + "step": 14129 + }, + { + "epoch": 1.797481236483908, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.288538932800293, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8886868953704834, + "num_tokens": 548412887.0, + "step": 14130 + }, + { + "epoch": 1.7976084467624984, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.287707328796387, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8851175904273987, + "num_tokens": 548452571.0, + "step": 14131 + }, + { + "epoch": 1.797735657041089, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.333686828613281, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8837152123451233, + "num_tokens": 548485572.0, + "step": 14132 + }, + { + "epoch": 1.7978628673196795, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.300416946411133, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8969314098358154, + "num_tokens": 548519238.0, + "step": 14133 + }, + { + "epoch": 1.79799007759827, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.320793151855469, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8736327886581421, + "num_tokens": 548557494.0, + "step": 14134 + }, + { + "epoch": 1.7981172878768605, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.233323097229004, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.883827805519104, + "num_tokens": 548598845.0, + "step": 14135 + }, + { + "epoch": 1.798244498155451, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.275591850280762, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8949121236801147, + "num_tokens": 548639079.0, + "step": 14136 + }, + { + "epoch": 1.7983717084340416, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.187126159667969, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.9007469415664673, + "num_tokens": 548677234.0, + "step": 14137 + }, + { + "epoch": 1.7984989187126321, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.271782875061035, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.88726806640625, + "num_tokens": 548719425.0, + "step": 14138 + }, + { + "epoch": 1.7986261289912227, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.330310821533203, + "learning_rate": 1e-06, + "loss": 0.3874, + "mean_token_accuracy": 0.8683582544326782, + "num_tokens": 548756795.0, + "step": 14139 + }, + { + "epoch": 1.798753339269813, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.254467964172363, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8891459703445435, + "num_tokens": 548794301.0, + "step": 14140 + }, + { + "epoch": 1.7988805495484035, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.289748191833496, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.886466383934021, + "num_tokens": 548830076.0, + "step": 14141 + }, + { + "epoch": 1.799007759826994, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.307771682739258, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8830198645591736, + "num_tokens": 548868499.0, + "step": 14142 + }, + { + "epoch": 1.7991349701055845, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.401772499084473, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8741141557693481, + "num_tokens": 548902155.0, + "step": 14143 + }, + { + "epoch": 1.799262180384175, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.26526927947998, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.895378589630127, + "num_tokens": 548940653.0, + "step": 14144 + }, + { + "epoch": 1.7993893906627656, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.178988456726074, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8989797830581665, + "num_tokens": 548982916.0, + "step": 14145 + }, + { + "epoch": 1.799516600941356, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.306079864501953, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8884404301643372, + "num_tokens": 549023458.0, + "step": 14146 + }, + { + "epoch": 1.7996438112199464, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.29763126373291, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8946052193641663, + "num_tokens": 549062364.0, + "step": 14147 + }, + { + "epoch": 1.799771021498537, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.330999374389648, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8832224011421204, + "num_tokens": 549099305.0, + "step": 14148 + }, + { + "epoch": 1.7998982317771275, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.306952476501465, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.8778982758522034, + "num_tokens": 549140064.0, + "step": 14149 + }, + { + "epoch": 1.800025442055718, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.212523460388184, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.8755807280540466, + "num_tokens": 549183047.0, + "step": 14150 + }, + { + "epoch": 1.8001526523343085, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.337515830993652, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8809804320335388, + "num_tokens": 549221334.0, + "step": 14151 + }, + { + "epoch": 1.800279862612899, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.211235046386719, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8882295489311218, + "num_tokens": 549262918.0, + "step": 14152 + }, + { + "epoch": 1.8004070728914896, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.309866905212402, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8893666863441467, + "num_tokens": 549302694.0, + "step": 14153 + }, + { + "epoch": 1.8005342831700801, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.248376846313477, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8915243744850159, + "num_tokens": 549346654.0, + "step": 14154 + }, + { + "epoch": 1.8006614934486707, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.325577735900879, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8842129111289978, + "num_tokens": 549385486.0, + "step": 14155 + }, + { + "epoch": 1.8007887037272612, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.317536354064941, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8849279880523682, + "num_tokens": 549423034.0, + "step": 14156 + }, + { + "epoch": 1.8009159140058517, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.206463813781738, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8931390047073364, + "num_tokens": 549469351.0, + "step": 14157 + }, + { + "epoch": 1.8010431242844422, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.297719955444336, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8870428204536438, + "num_tokens": 549506869.0, + "step": 14158 + }, + { + "epoch": 1.8011703345630328, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.346963882446289, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8913478851318359, + "num_tokens": 549544835.0, + "step": 14159 + }, + { + "epoch": 1.8012975448416233, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.263654708862305, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8857786059379578, + "num_tokens": 549586604.0, + "step": 14160 + }, + { + "epoch": 1.8014247551202138, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.296055793762207, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8965362906455994, + "num_tokens": 549621910.0, + "step": 14161 + }, + { + "epoch": 1.8015519653988044, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.275206565856934, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8968613147735596, + "num_tokens": 549661450.0, + "step": 14162 + }, + { + "epoch": 1.8016791756773949, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.271024703979492, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8894072771072388, + "num_tokens": 549698831.0, + "step": 14163 + }, + { + "epoch": 1.8018063859559852, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.342262268066406, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.9017993211746216, + "num_tokens": 549735835.0, + "step": 14164 + }, + { + "epoch": 1.8019335962345757, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.31412124633789, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.890802264213562, + "num_tokens": 549773242.0, + "step": 14165 + }, + { + "epoch": 1.8020608065131662, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.259876251220703, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8838242292404175, + "num_tokens": 549817521.0, + "step": 14166 + }, + { + "epoch": 1.8021880167917568, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.365379333496094, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8756558299064636, + "num_tokens": 549851571.0, + "step": 14167 + }, + { + "epoch": 1.8023152270703473, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.257023811340332, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.888399064540863, + "num_tokens": 549893108.0, + "step": 14168 + }, + { + "epoch": 1.8024424373489378, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.310338973999023, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8864625692367554, + "num_tokens": 549931949.0, + "step": 14169 + }, + { + "epoch": 1.8025696476275284, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.377060890197754, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8979071378707886, + "num_tokens": 549961286.0, + "step": 14170 + }, + { + "epoch": 1.8026968579061187, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.273630142211914, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8961964845657349, + "num_tokens": 549994203.0, + "step": 14171 + }, + { + "epoch": 1.8028240681847092, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.316184043884277, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8874558210372925, + "num_tokens": 550029669.0, + "step": 14172 + }, + { + "epoch": 1.8029512784632997, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.327113151550293, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8798036575317383, + "num_tokens": 550065517.0, + "step": 14173 + }, + { + "epoch": 1.8030784887418903, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.23387622833252, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.896857500076294, + "num_tokens": 550104205.0, + "step": 14174 + }, + { + "epoch": 1.8032056990204808, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.297782897949219, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8984899520874023, + "num_tokens": 550143279.0, + "step": 14175 + }, + { + "epoch": 1.8033329092990713, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.250682830810547, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8975440263748169, + "num_tokens": 550179739.0, + "step": 14176 + }, + { + "epoch": 1.8034601195776618, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.249368667602539, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8764926195144653, + "num_tokens": 550224740.0, + "step": 14177 + }, + { + "epoch": 1.8035873298562524, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.26563549041748, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8843148350715637, + "num_tokens": 550268621.0, + "step": 14178 + }, + { + "epoch": 1.803714540134843, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.252972602844238, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8896461129188538, + "num_tokens": 550305066.0, + "step": 14179 + }, + { + "epoch": 1.8038417504134334, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.376864433288574, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8837016820907593, + "num_tokens": 550345617.0, + "step": 14180 + }, + { + "epoch": 1.803968960692024, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.20643138885498, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8930501937866211, + "num_tokens": 550388121.0, + "step": 14181 + }, + { + "epoch": 1.8040961709706145, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.251168251037598, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.894585371017456, + "num_tokens": 550437559.0, + "step": 14182 + }, + { + "epoch": 1.804223381249205, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.327903747558594, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8893325924873352, + "num_tokens": 550472775.0, + "step": 14183 + }, + { + "epoch": 1.8043505915277955, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.23901081085205, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8943253755569458, + "num_tokens": 550515060.0, + "step": 14184 + }, + { + "epoch": 1.804477801806386, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.327113151550293, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8868982791900635, + "num_tokens": 550552607.0, + "step": 14185 + }, + { + "epoch": 1.8046050120849766, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.262186050415039, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8940712213516235, + "num_tokens": 550592539.0, + "step": 14186 + }, + { + "epoch": 1.8047322223635671, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.3118257522583, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8785094022750854, + "num_tokens": 550631368.0, + "step": 14187 + }, + { + "epoch": 1.8048594326421576, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.249058723449707, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8901407718658447, + "num_tokens": 550670584.0, + "step": 14188 + }, + { + "epoch": 1.804986642920748, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.291810989379883, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8771769404411316, + "num_tokens": 550709511.0, + "step": 14189 + }, + { + "epoch": 1.8051138531993385, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.209626197814941, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8999619483947754, + "num_tokens": 550749540.0, + "step": 14190 + }, + { + "epoch": 1.805241063477929, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.320295333862305, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.9025019407272339, + "num_tokens": 550785349.0, + "step": 14191 + }, + { + "epoch": 1.8053682737565195, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.297633171081543, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8822354078292847, + "num_tokens": 550823854.0, + "step": 14192 + }, + { + "epoch": 1.80549548403511, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.325828552246094, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8840761780738831, + "num_tokens": 550860352.0, + "step": 14193 + }, + { + "epoch": 1.8056226943137006, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.229543685913086, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8899434208869934, + "num_tokens": 550900672.0, + "step": 14194 + }, + { + "epoch": 1.805749904592291, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.294336318969727, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8963963985443115, + "num_tokens": 550934063.0, + "step": 14195 + }, + { + "epoch": 1.8058771148708814, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.315555572509766, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8878406882286072, + "num_tokens": 550971872.0, + "step": 14196 + }, + { + "epoch": 1.806004325149472, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.272157669067383, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8871932029724121, + "num_tokens": 551016909.0, + "step": 14197 + }, + { + "epoch": 1.8061315354280625, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.290712356567383, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8865927457809448, + "num_tokens": 551058969.0, + "step": 14198 + }, + { + "epoch": 1.806258745706653, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.055002212524414e-05, + "grad_norm": 9.315023422241211, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8798820376396179, + "num_tokens": 551096968.0, + "step": 14199 + }, + { + "epoch": 1.8063859559852435, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.369133949279785, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8760948181152344, + "num_tokens": 551141462.0, + "step": 14200 + }, + { + "epoch": 1.806513166263834, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.306288719177246, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8959442377090454, + "num_tokens": 551176401.0, + "step": 14201 + }, + { + "epoch": 1.8066403765424246, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.374632835388184, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8785910606384277, + "num_tokens": 551210772.0, + "step": 14202 + }, + { + "epoch": 1.8067675868210151, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.264962196350098, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8954920768737793, + "num_tokens": 551252090.0, + "step": 14203 + }, + { + "epoch": 1.8068947970996057, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.240130424499512, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8789300918579102, + "num_tokens": 551292933.0, + "step": 14204 + }, + { + "epoch": 1.8070220073781962, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.288971900939941, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8884005546569824, + "num_tokens": 551334490.0, + "step": 14205 + }, + { + "epoch": 1.8071492176567867, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.236687660217285, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8937551975250244, + "num_tokens": 551377262.0, + "step": 14206 + }, + { + "epoch": 1.8072764279353772, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.305573463439941, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8889423608779907, + "num_tokens": 551416471.0, + "step": 14207 + }, + { + "epoch": 1.8074036382139678, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.338001251220703, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8858824968338013, + "num_tokens": 551454103.0, + "step": 14208 + }, + { + "epoch": 1.8075308484925583, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.179365158081055, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8941364288330078, + "num_tokens": 551494331.0, + "step": 14209 + }, + { + "epoch": 1.8076580587711488, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.293262481689453, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8815749287605286, + "num_tokens": 551534928.0, + "step": 14210 + }, + { + "epoch": 1.8077852690497394, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.223340034484863, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8929192423820496, + "num_tokens": 551574456.0, + "step": 14211 + }, + { + "epoch": 1.8079124793283299, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.2461519241333, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8989300727844238, + "num_tokens": 551615344.0, + "step": 14212 + }, + { + "epoch": 1.8080396896069202, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.301071166992188, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8823490738868713, + "num_tokens": 551653591.0, + "step": 14213 + }, + { + "epoch": 1.8081668998855107, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.33518123626709, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8889563083648682, + "num_tokens": 551691721.0, + "step": 14214 + }, + { + "epoch": 1.8082941101641012, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.244537353515625, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8783832788467407, + "num_tokens": 551736698.0, + "step": 14215 + }, + { + "epoch": 1.8084213204426918, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.373228073120117, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8883060812950134, + "num_tokens": 551771909.0, + "step": 14216 + }, + { + "epoch": 1.8085485307212823, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.31835651397705, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8749856948852539, + "num_tokens": 551814033.0, + "step": 14217 + }, + { + "epoch": 1.8086757409998728, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.197422981262207, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8881933689117432, + "num_tokens": 551854251.0, + "step": 14218 + }, + { + "epoch": 1.8088029512784631, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.252177238464355, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8983135223388672, + "num_tokens": 551893094.0, + "step": 14219 + }, + { + "epoch": 1.8089301615570537, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.356590270996094, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8794867992401123, + "num_tokens": 551931381.0, + "step": 14220 + }, + { + "epoch": 1.8090573718356442, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.253620147705078, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8846213817596436, + "num_tokens": 551969141.0, + "step": 14221 + }, + { + "epoch": 1.8091845821142347, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.281846046447754, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8921859264373779, + "num_tokens": 552007546.0, + "step": 14222 + }, + { + "epoch": 1.8093117923928252, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.33456802368164, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8780562877655029, + "num_tokens": 552048997.0, + "step": 14223 + }, + { + "epoch": 1.8094390026714158, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.246528625488281, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.877091646194458, + "num_tokens": 552094675.0, + "step": 14224 + }, + { + "epoch": 1.8095662129500063, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.363910675048828, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8883092403411865, + "num_tokens": 552131389.0, + "step": 14225 + }, + { + "epoch": 1.8096934232285968, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.224739074707031, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8897433280944824, + "num_tokens": 552171609.0, + "step": 14226 + }, + { + "epoch": 1.8098206335071874, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.356006622314453, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8818104267120361, + "num_tokens": 552212875.0, + "step": 14227 + }, + { + "epoch": 1.8099478437857779, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.297589302062988, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8809484243392944, + "num_tokens": 552249333.0, + "step": 14228 + }, + { + "epoch": 1.8100750540643684, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.289925575256348, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8929274082183838, + "num_tokens": 552288304.0, + "step": 14229 + }, + { + "epoch": 1.810202264342959, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.430828094482422, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8889997601509094, + "num_tokens": 552319219.0, + "step": 14230 + }, + { + "epoch": 1.8103294746215495, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.239590644836426, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8994730710983276, + "num_tokens": 552352628.0, + "step": 14231 + }, + { + "epoch": 1.81045668490014, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.366508483886719, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8958913087844849, + "num_tokens": 552384839.0, + "step": 14232 + }, + { + "epoch": 1.8105838951787305, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.212157249450684, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8989413976669312, + "num_tokens": 552423937.0, + "step": 14233 + }, + { + "epoch": 1.810711105457321, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.305177688598633, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8804057836532593, + "num_tokens": 552459247.0, + "step": 14234 + }, + { + "epoch": 1.8108383157359116, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.243714332580566, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8938688039779663, + "num_tokens": 552497863.0, + "step": 14235 + }, + { + "epoch": 1.8109655260145021, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.270627975463867, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8826775550842285, + "num_tokens": 552534183.0, + "step": 14236 + }, + { + "epoch": 1.8110927362930926, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.343555450439453, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8834935426712036, + "num_tokens": 552574221.0, + "step": 14237 + }, + { + "epoch": 1.811219946571683, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.296085357666016, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8668761253356934, + "num_tokens": 552614303.0, + "step": 14238 + }, + { + "epoch": 1.8113471568502735, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.22850513458252, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.886945366859436, + "num_tokens": 552655609.0, + "step": 14239 + }, + { + "epoch": 1.811474367128864, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.234442710876465, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8879883289337158, + "num_tokens": 552694694.0, + "step": 14240 + }, + { + "epoch": 1.8116015774074545, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.323371887207031, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8819078803062439, + "num_tokens": 552733039.0, + "step": 14241 + }, + { + "epoch": 1.811728787686045, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.341249465942383, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8994795083999634, + "num_tokens": 552765343.0, + "step": 14242 + }, + { + "epoch": 1.8118559979646356, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.213040351867676, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9002653360366821, + "num_tokens": 552803466.0, + "step": 14243 + }, + { + "epoch": 1.811983208243226, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.381853103637695, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8734077215194702, + "num_tokens": 552840795.0, + "step": 14244 + }, + { + "epoch": 1.8121104185218164, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.24117660522461, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8910858035087585, + "num_tokens": 552880218.0, + "step": 14245 + }, + { + "epoch": 1.812237628800407, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.225271224975586, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8897665739059448, + "num_tokens": 552924489.0, + "step": 14246 + }, + { + "epoch": 1.8123648390789975, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.274862289428711, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8865529298782349, + "num_tokens": 552960753.0, + "step": 14247 + }, + { + "epoch": 1.812492049357588, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.219361305236816, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.9007452726364136, + "num_tokens": 552997822.0, + "step": 14248 + }, + { + "epoch": 1.8126192596361785, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.219225883483887, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8844456672668457, + "num_tokens": 553040708.0, + "step": 14249 + }, + { + "epoch": 1.812746469914769, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.229463577270508, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.898950457572937, + "num_tokens": 553077583.0, + "step": 14250 + }, + { + "epoch": 1.8128736801933596, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.318117141723633, + "learning_rate": 1e-06, + "loss": 0.3475, + "mean_token_accuracy": 0.883525013923645, + "num_tokens": 553112894.0, + "step": 14251 + }, + { + "epoch": 1.8130008904719501, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.249931335449219, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8902922868728638, + "num_tokens": 553151895.0, + "step": 14252 + }, + { + "epoch": 1.8131281007505406, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.300625801086426, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.874914288520813, + "num_tokens": 553190425.0, + "step": 14253 + }, + { + "epoch": 1.8132553110291312, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.179359436035156, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8902693390846252, + "num_tokens": 553232390.0, + "step": 14254 + }, + { + "epoch": 1.8133825213077217, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.317136764526367, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8914382457733154, + "num_tokens": 553270364.0, + "step": 14255 + }, + { + "epoch": 1.8135097315863122, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.373571395874023, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8817702531814575, + "num_tokens": 553304590.0, + "step": 14256 + }, + { + "epoch": 1.8136369418649028, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.312511444091797, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8786294460296631, + "num_tokens": 553343699.0, + "step": 14257 + }, + { + "epoch": 1.8137641521434933, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.234810829162598, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.892420768737793, + "num_tokens": 553382214.0, + "step": 14258 + }, + { + "epoch": 1.8138913624220838, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.341964721679688, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8850765824317932, + "num_tokens": 553415447.0, + "step": 14259 + }, + { + "epoch": 1.8140185727006743, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.26168441772461, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8975266218185425, + "num_tokens": 553451216.0, + "step": 14260 + }, + { + "epoch": 1.8141457829792649, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.300102233886719, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8907737135887146, + "num_tokens": 553487920.0, + "step": 14261 + }, + { + "epoch": 1.8142729932578552, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.254451751708984, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.88700932264328, + "num_tokens": 553529339.0, + "step": 14262 + }, + { + "epoch": 1.8144002035364457, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.305464744567871, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8924371004104614, + "num_tokens": 553564645.0, + "step": 14263 + }, + { + "epoch": 1.8145274138150362, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.34652042388916, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8838369846343994, + "num_tokens": 553597799.0, + "step": 14264 + }, + { + "epoch": 1.8146546240936268, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.324382781982422, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8854866027832031, + "num_tokens": 553640024.0, + "step": 14265 + }, + { + "epoch": 1.8147818343722173, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.272200584411621, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8879314661026001, + "num_tokens": 553678960.0, + "step": 14266 + }, + { + "epoch": 1.8149090446508078, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.318166732788086, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8821715116500854, + "num_tokens": 553714739.0, + "step": 14267 + }, + { + "epoch": 1.8150362549293981, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.329676628112793, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.883065402507782, + "num_tokens": 553752321.0, + "step": 14268 + }, + { + "epoch": 1.8151634652079887, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.282090187072754, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8892027139663696, + "num_tokens": 553791967.0, + "step": 14269 + }, + { + "epoch": 1.8152906754865792, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.276449203491211, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8888188600540161, + "num_tokens": 553832667.0, + "step": 14270 + }, + { + "epoch": 1.8154178857651697, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.240585327148438, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8749163746833801, + "num_tokens": 553867667.0, + "step": 14271 + }, + { + "epoch": 1.8155450960437602, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.332454681396484, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8769984245300293, + "num_tokens": 553907047.0, + "step": 14272 + }, + { + "epoch": 1.8156723063223508, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.270551681518555, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8841981291770935, + "num_tokens": 553944160.0, + "step": 14273 + }, + { + "epoch": 1.8157995166009413, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.200567245483398, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.887248158454895, + "num_tokens": 553982726.0, + "step": 14274 + }, + { + "epoch": 1.8159267268795318, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.295372009277344, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8854967951774597, + "num_tokens": 554020642.0, + "step": 14275 + }, + { + "epoch": 1.8160539371581224, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.205733299255371, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8946906328201294, + "num_tokens": 554065710.0, + "step": 14276 + }, + { + "epoch": 1.8161811474367129, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.176487922668457, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8918119072914124, + "num_tokens": 554111823.0, + "step": 14277 + }, + { + "epoch": 1.8163083577153034, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.218330383300781, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8888514637947083, + "num_tokens": 554157942.0, + "step": 14278 + }, + { + "epoch": 1.816435567993894, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.245168685913086, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8944228887557983, + "num_tokens": 554196533.0, + "step": 14279 + }, + { + "epoch": 1.8165627782724845, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.25772762298584, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8825490474700928, + "num_tokens": 554238724.0, + "step": 14280 + }, + { + "epoch": 1.816689988551075, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.211939811706543, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.9055185914039612, + "num_tokens": 554280113.0, + "step": 14281 + }, + { + "epoch": 1.8168171988296655, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.387678146362305, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8845207095146179, + "num_tokens": 554317737.0, + "step": 14282 + }, + { + "epoch": 1.816944409108256, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.297303199768066, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8915287256240845, + "num_tokens": 554354497.0, + "step": 14283 + }, + { + "epoch": 1.8170716193868466, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.28901195526123, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8876700401306152, + "num_tokens": 554389291.0, + "step": 14284 + }, + { + "epoch": 1.817198829665437, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.304755210876465, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8824740052223206, + "num_tokens": 554428056.0, + "step": 14285 + }, + { + "epoch": 1.8173260399440276, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.298373222351074, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8757575750350952, + "num_tokens": 554467405.0, + "step": 14286 + }, + { + "epoch": 1.817453250222618, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.26357650756836, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8910503387451172, + "num_tokens": 554506808.0, + "step": 14287 + }, + { + "epoch": 1.8175804605012085, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.301155090332031, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8915948271751404, + "num_tokens": 554541971.0, + "step": 14288 + }, + { + "epoch": 1.817707670779799, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.359147071838379, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8949187994003296, + "num_tokens": 554574550.0, + "step": 14289 + }, + { + "epoch": 1.8178348810583895, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.240467071533203, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8832008838653564, + "num_tokens": 554617214.0, + "step": 14290 + }, + { + "epoch": 1.81796209133698, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.248903274536133, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8960449695587158, + "num_tokens": 554655174.0, + "step": 14291 + }, + { + "epoch": 1.8180893016155706, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.503908157348633, + "learning_rate": 1e-06, + "loss": 0.4041, + "mean_token_accuracy": 0.8715250492095947, + "num_tokens": 554686059.0, + "step": 14292 + }, + { + "epoch": 1.818216511894161, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.26850414276123, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8921520709991455, + "num_tokens": 554722723.0, + "step": 14293 + }, + { + "epoch": 1.8183437221727514, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.21480655670166, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8912086486816406, + "num_tokens": 554768177.0, + "step": 14294 + }, + { + "epoch": 1.818470932451342, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.296591758728027, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8968456387519836, + "num_tokens": 554803901.0, + "step": 14295 + }, + { + "epoch": 1.8185981427299325, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.260538101196289, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8923940658569336, + "num_tokens": 554842106.0, + "step": 14296 + }, + { + "epoch": 1.818725353008523, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.25259017944336, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8975344300270081, + "num_tokens": 554880695.0, + "step": 14297 + }, + { + "epoch": 1.8188525632871135, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.504982948303223, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8792990446090698, + "num_tokens": 554910977.0, + "step": 14298 + }, + { + "epoch": 1.818979773565704, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.333592414855957, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8881755471229553, + "num_tokens": 554943638.0, + "step": 14299 + }, + { + "epoch": 1.8191069838442946, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.1754789352417, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8939096927642822, + "num_tokens": 554987758.0, + "step": 14300 + }, + { + "epoch": 1.8192341941228851, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.307244300842285, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8813222646713257, + "num_tokens": 555030112.0, + "step": 14301 + }, + { + "epoch": 1.8193614044014756, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.3684720993042, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8836130499839783, + "num_tokens": 555066638.0, + "step": 14302 + }, + { + "epoch": 1.8194886146800662, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.315984725952148, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8996645212173462, + "num_tokens": 555103622.0, + "step": 14303 + }, + { + "epoch": 1.8196158249586567, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.39371395111084, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8944388628005981, + "num_tokens": 555138304.0, + "step": 14304 + }, + { + "epoch": 1.8197430352372472, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.348038673400879, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8751571178436279, + "num_tokens": 555175743.0, + "step": 14305 + }, + { + "epoch": 1.8198702455158378, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.355975151062012, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.886995792388916, + "num_tokens": 555216028.0, + "step": 14306 + }, + { + "epoch": 1.8199974557944283, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.287530899047852, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8768312931060791, + "num_tokens": 555253185.0, + "step": 14307 + }, + { + "epoch": 1.8201246660730188, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.303095817565918, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.887837290763855, + "num_tokens": 555290863.0, + "step": 14308 + }, + { + "epoch": 1.8202518763516093, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.23791217803955, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.9046165943145752, + "num_tokens": 555334541.0, + "step": 14309 + }, + { + "epoch": 1.8203790866301999, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.261350631713867, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8925518989562988, + "num_tokens": 555372610.0, + "step": 14310 + }, + { + "epoch": 1.8205062969087902, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.36280345916748, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8797603249549866, + "num_tokens": 555413912.0, + "step": 14311 + }, + { + "epoch": 1.8206335071873807, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.342826843261719, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8855056166648865, + "num_tokens": 555451143.0, + "step": 14312 + }, + { + "epoch": 1.8207607174659712, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.36828899383545, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.9013002514839172, + "num_tokens": 555488073.0, + "step": 14313 + }, + { + "epoch": 1.8208879277445618, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.400464057922363, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8768369555473328, + "num_tokens": 555521176.0, + "step": 14314 + }, + { + "epoch": 1.8210151380231523, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.345355987548828, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.877871036529541, + "num_tokens": 555564059.0, + "step": 14315 + }, + { + "epoch": 1.8211423483017428, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.297277450561523, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8792831897735596, + "num_tokens": 555604932.0, + "step": 14316 + }, + { + "epoch": 1.8212695585803331, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.287871360778809, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8933156728744507, + "num_tokens": 555643130.0, + "step": 14317 + }, + { + "epoch": 1.8213967688589237, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.352477073669434, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8860552310943604, + "num_tokens": 555681311.0, + "step": 14318 + }, + { + "epoch": 1.8215239791375142, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.267056465148926, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8942052125930786, + "num_tokens": 555719839.0, + "step": 14319 + }, + { + "epoch": 1.8216511894161047, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.293928146362305, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8882262110710144, + "num_tokens": 555757384.0, + "step": 14320 + }, + { + "epoch": 1.8217783996946952, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.255217552185059, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8978216648101807, + "num_tokens": 555799714.0, + "step": 14321 + }, + { + "epoch": 1.8219056099732858, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.257930755615234, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8877366781234741, + "num_tokens": 555838131.0, + "step": 14322 + }, + { + "epoch": 1.8220328202518763, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.29698657989502, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8798131942749023, + "num_tokens": 555883233.0, + "step": 14323 + }, + { + "epoch": 1.8221600305304668, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.260927200317383, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8833067417144775, + "num_tokens": 555925753.0, + "step": 14324 + }, + { + "epoch": 1.8222872408090574, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.341350555419922, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8979697227478027, + "num_tokens": 555959554.0, + "step": 14325 + }, + { + "epoch": 1.8224144510876479, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.212149620056152, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8963705897331238, + "num_tokens": 556001473.0, + "step": 14326 + }, + { + "epoch": 1.8225416613662384, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.338237762451172, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8760982751846313, + "num_tokens": 556038441.0, + "step": 14327 + }, + { + "epoch": 1.822668871644829, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.244423866271973, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8809590339660645, + "num_tokens": 556080480.0, + "step": 14328 + }, + { + "epoch": 1.8227960819234195, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.298666954040527, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8897237777709961, + "num_tokens": 556118835.0, + "step": 14329 + }, + { + "epoch": 1.82292329220201, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.21391773223877, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8965821862220764, + "num_tokens": 556156332.0, + "step": 14330 + }, + { + "epoch": 1.8230505024806005, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.286555290222168, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.892485499382019, + "num_tokens": 556193530.0, + "step": 14331 + }, + { + "epoch": 1.823177712759191, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.283567428588867, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8738473653793335, + "num_tokens": 556237028.0, + "step": 14332 + }, + { + "epoch": 1.8233049230377816, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.23945140838623, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8945165872573853, + "num_tokens": 556277666.0, + "step": 14333 + }, + { + "epoch": 1.823432133316372, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.320829391479492, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8894470930099487, + "num_tokens": 556315988.0, + "step": 14334 + }, + { + "epoch": 1.8235593435949626, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.399155616760254, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.883211076259613, + "num_tokens": 556346616.0, + "step": 14335 + }, + { + "epoch": 1.823686553873553, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.33408260345459, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.888840913772583, + "num_tokens": 556380835.0, + "step": 14336 + }, + { + "epoch": 1.8238137641521435, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.262079238891602, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8910408020019531, + "num_tokens": 556418653.0, + "step": 14337 + }, + { + "epoch": 1.823940974430734, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.381890296936035, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8834826946258545, + "num_tokens": 556451900.0, + "step": 14338 + }, + { + "epoch": 1.8240681847093245, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.354656219482422, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8916336297988892, + "num_tokens": 556484526.0, + "step": 14339 + }, + { + "epoch": 1.824195394987915, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.247773170471191, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8850817680358887, + "num_tokens": 556525764.0, + "step": 14340 + }, + { + "epoch": 1.8243226052665056, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.163555145263672, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.9008468389511108, + "num_tokens": 556567746.0, + "step": 14341 + }, + { + "epoch": 1.8244498155450959, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.170907020568848, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.884189248085022, + "num_tokens": 556611888.0, + "step": 14342 + }, + { + "epoch": 1.8245770258236864, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.351421356201172, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8789061903953552, + "num_tokens": 556649242.0, + "step": 14343 + }, + { + "epoch": 1.824704236102277, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.271026611328125, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8884797692298889, + "num_tokens": 556687195.0, + "step": 14344 + }, + { + "epoch": 1.8248314463808675, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.40217399597168, + "learning_rate": 1e-06, + "loss": 0.3892, + "mean_token_accuracy": 0.8715130686759949, + "num_tokens": 556721441.0, + "step": 14345 + }, + { + "epoch": 1.824958656659458, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.368334770202637, + "learning_rate": 1e-06, + "loss": 0.4045, + "mean_token_accuracy": 0.8641803860664368, + "num_tokens": 556758881.0, + "step": 14346 + }, + { + "epoch": 1.8250858669380485, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.253474235534668, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.902870774269104, + "num_tokens": 556798246.0, + "step": 14347 + }, + { + "epoch": 1.825213077216639, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.283544540405273, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8983966708183289, + "num_tokens": 556836339.0, + "step": 14348 + }, + { + "epoch": 1.8253402874952296, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.355915069580078, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8966240882873535, + "num_tokens": 556871146.0, + "step": 14349 + }, + { + "epoch": 1.8254674977738201, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.221505165100098, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8876777291297913, + "num_tokens": 556914195.0, + "step": 14350 + }, + { + "epoch": 1.8255947080524106, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.296127319335938, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8854414224624634, + "num_tokens": 556953894.0, + "step": 14351 + }, + { + "epoch": 1.8257219183310012, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.214881896972656, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8966971635818481, + "num_tokens": 556999609.0, + "step": 14352 + }, + { + "epoch": 1.8258491286095917, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.331060409545898, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8913120627403259, + "num_tokens": 557040412.0, + "step": 14353 + }, + { + "epoch": 1.8259763388881822, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.225000381469727, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9090399742126465, + "num_tokens": 557075220.0, + "step": 14354 + }, + { + "epoch": 1.8261035491667728, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.299063682556152, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.892244815826416, + "num_tokens": 557114029.0, + "step": 14355 + }, + { + "epoch": 1.8262307594453633, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.275635719299316, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8973735570907593, + "num_tokens": 557154708.0, + "step": 14356 + }, + { + "epoch": 1.8263579697239538, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.381511688232422, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8828020095825195, + "num_tokens": 557192728.0, + "step": 14357 + }, + { + "epoch": 1.8264851800025443, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.33400821685791, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8937540054321289, + "num_tokens": 557228722.0, + "step": 14358 + }, + { + "epoch": 1.8266123902811349, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.313368797302246, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8834500908851624, + "num_tokens": 557268439.0, + "step": 14359 + }, + { + "epoch": 1.8267396005597252, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.283980369567871, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8951653242111206, + "num_tokens": 557306808.0, + "step": 14360 + }, + { + "epoch": 1.8268668108383157, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.263291358947754, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8846084475517273, + "num_tokens": 557347578.0, + "step": 14361 + }, + { + "epoch": 1.8269940211169062, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.416890144348145, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8883916735649109, + "num_tokens": 557382373.0, + "step": 14362 + }, + { + "epoch": 1.8271212313954968, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.267380714416504, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8955660462379456, + "num_tokens": 557423140.0, + "step": 14363 + }, + { + "epoch": 1.8272484416740873, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.31534481048584, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8828178644180298, + "num_tokens": 557459959.0, + "step": 14364 + }, + { + "epoch": 1.8273756519526778, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.36849594116211, + "learning_rate": 1e-06, + "loss": 0.3734, + "mean_token_accuracy": 0.8773427605628967, + "num_tokens": 557503501.0, + "step": 14365 + }, + { + "epoch": 1.8275028622312681, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.270933151245117, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8997761607170105, + "num_tokens": 557544597.0, + "step": 14366 + }, + { + "epoch": 1.8276300725098586, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.235730171203613, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8857583999633789, + "num_tokens": 557588542.0, + "step": 14367 + }, + { + "epoch": 1.8277572827884492, + "ewc_loss": 0.016845703125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0609626770019531e-05, + "grad_norm": 9.395888328552246, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8763761520385742, + "num_tokens": 557629504.0, + "step": 14368 + }, + { + "epoch": 1.8278844930670397, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.314114570617676, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8828562498092651, + "num_tokens": 557668080.0, + "step": 14369 + }, + { + "epoch": 1.8280117033456302, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.403368949890137, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8946421146392822, + "num_tokens": 557701987.0, + "step": 14370 + }, + { + "epoch": 1.8281389136242208, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.284348487854004, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8984214067459106, + "num_tokens": 557738221.0, + "step": 14371 + }, + { + "epoch": 1.8282661239028113, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.261298179626465, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8919675946235657, + "num_tokens": 557776347.0, + "step": 14372 + }, + { + "epoch": 1.8283933341814018, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.398466110229492, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8841289281845093, + "num_tokens": 557816118.0, + "step": 14373 + }, + { + "epoch": 1.8285205444599923, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.296541213989258, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.889073371887207, + "num_tokens": 557856168.0, + "step": 14374 + }, + { + "epoch": 1.8286477547385829, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.428323745727539, + "learning_rate": 1e-06, + "loss": 0.3775, + "mean_token_accuracy": 0.8719276785850525, + "num_tokens": 557894414.0, + "step": 14375 + }, + { + "epoch": 1.8287749650171734, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.249246597290039, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8835358619689941, + "num_tokens": 557934669.0, + "step": 14376 + }, + { + "epoch": 1.828902175295764, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.289968490600586, + "learning_rate": 1e-06, + "loss": 0.3797, + "mean_token_accuracy": 0.8718457221984863, + "num_tokens": 557979071.0, + "step": 14377 + }, + { + "epoch": 1.8290293855743545, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.31932258605957, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8972675800323486, + "num_tokens": 558013098.0, + "step": 14378 + }, + { + "epoch": 1.829156595852945, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.259720802307129, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8916013240814209, + "num_tokens": 558053696.0, + "step": 14379 + }, + { + "epoch": 1.8292838061315355, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.355024337768555, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8800714015960693, + "num_tokens": 558090852.0, + "step": 14380 + }, + { + "epoch": 1.829411016410126, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.332096099853516, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8789603114128113, + "num_tokens": 558122365.0, + "step": 14381 + }, + { + "epoch": 1.8295382266887166, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.361852645874023, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8934570550918579, + "num_tokens": 558159699.0, + "step": 14382 + }, + { + "epoch": 1.829665436967307, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.237881660461426, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8865391612052917, + "num_tokens": 558203090.0, + "step": 14383 + }, + { + "epoch": 1.8297926472458976, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.336280822753906, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.9030337333679199, + "num_tokens": 558237469.0, + "step": 14384 + }, + { + "epoch": 1.829919857524488, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.271247863769531, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8996665477752686, + "num_tokens": 558272470.0, + "step": 14385 + }, + { + "epoch": 1.8300470678030785, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.356172561645508, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8819014430046082, + "num_tokens": 558308909.0, + "step": 14386 + }, + { + "epoch": 1.830174278081669, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.316330909729004, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8915428519248962, + "num_tokens": 558348484.0, + "step": 14387 + }, + { + "epoch": 1.8303014883602595, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.2855224609375, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8947808742523193, + "num_tokens": 558385404.0, + "step": 14388 + }, + { + "epoch": 1.83042869863885, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.2461576461792, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8996659517288208, + "num_tokens": 558423176.0, + "step": 14389 + }, + { + "epoch": 1.8305559089174406, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.30151653289795, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8764379024505615, + "num_tokens": 558469952.0, + "step": 14390 + }, + { + "epoch": 1.8306831191960309, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.26987075805664, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8966588973999023, + "num_tokens": 558509077.0, + "step": 14391 + }, + { + "epoch": 1.8308103294746214, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.28042221069336, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.882378101348877, + "num_tokens": 558549137.0, + "step": 14392 + }, + { + "epoch": 1.830937539753212, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.361519813537598, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8789472579956055, + "num_tokens": 558586331.0, + "step": 14393 + }, + { + "epoch": 1.8310647500318025, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.360255241394043, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8875746726989746, + "num_tokens": 558624724.0, + "step": 14394 + }, + { + "epoch": 1.831191960310393, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.317022323608398, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8891979455947876, + "num_tokens": 558660621.0, + "step": 14395 + }, + { + "epoch": 1.8313191705889835, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.271828651428223, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8819535374641418, + "num_tokens": 558702606.0, + "step": 14396 + }, + { + "epoch": 1.831446380867574, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.271588325500488, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8940523266792297, + "num_tokens": 558744262.0, + "step": 14397 + }, + { + "epoch": 1.8315735911461646, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.382879257202148, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8766695261001587, + "num_tokens": 558784581.0, + "step": 14398 + }, + { + "epoch": 1.831700801424755, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.320579528808594, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.9050430655479431, + "num_tokens": 558817351.0, + "step": 14399 + }, + { + "epoch": 1.8318280117033456, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.299519538879395, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.8737359642982483, + "num_tokens": 558864179.0, + "step": 14400 + }, + { + "epoch": 1.8319552219819362, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.381218910217285, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8833032846450806, + "num_tokens": 558901726.0, + "step": 14401 + }, + { + "epoch": 1.8320824322605267, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.274927139282227, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.881738007068634, + "num_tokens": 558944587.0, + "step": 14402 + }, + { + "epoch": 1.8322096425391172, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.312980651855469, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8958176374435425, + "num_tokens": 558982470.0, + "step": 14403 + }, + { + "epoch": 1.8323368528177078, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.252082824707031, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8983117341995239, + "num_tokens": 559025556.0, + "step": 14404 + }, + { + "epoch": 1.8324640630962983, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.3045654296875, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8770819306373596, + "num_tokens": 559071874.0, + "step": 14405 + }, + { + "epoch": 1.8325912733748888, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.319657325744629, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8952642679214478, + "num_tokens": 559109294.0, + "step": 14406 + }, + { + "epoch": 1.8327184836534793, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.215444564819336, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8929848670959473, + "num_tokens": 559153067.0, + "step": 14407 + }, + { + "epoch": 1.8328456939320699, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.353043556213379, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8878674507141113, + "num_tokens": 559192394.0, + "step": 14408 + }, + { + "epoch": 1.8329729042106602, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.327366828918457, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8950754404067993, + "num_tokens": 559225471.0, + "step": 14409 + }, + { + "epoch": 1.8331001144892507, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.348630905151367, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8827117681503296, + "num_tokens": 559264876.0, + "step": 14410 + }, + { + "epoch": 1.8332273247678412, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.288091659545898, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8943310976028442, + "num_tokens": 559302553.0, + "step": 14411 + }, + { + "epoch": 1.8333545350464318, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.280464172363281, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8811990022659302, + "num_tokens": 559342276.0, + "step": 14412 + }, + { + "epoch": 1.8334817453250223, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.278072357177734, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8762005567550659, + "num_tokens": 559387465.0, + "step": 14413 + }, + { + "epoch": 1.8336089556036128, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.358745574951172, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8839861750602722, + "num_tokens": 559425107.0, + "step": 14414 + }, + { + "epoch": 1.8337361658822031, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.306995391845703, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8928981423377991, + "num_tokens": 559465386.0, + "step": 14415 + }, + { + "epoch": 1.8338633761607936, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.229990005493164, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.893109917640686, + "num_tokens": 559504116.0, + "step": 14416 + }, + { + "epoch": 1.8339905864393842, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.349738121032715, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8854401111602783, + "num_tokens": 559540005.0, + "step": 14417 + }, + { + "epoch": 1.8341177967179747, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.2827787399292, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9042110443115234, + "num_tokens": 559573793.0, + "step": 14418 + }, + { + "epoch": 1.8342450069965652, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.26071548461914, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9023371934890747, + "num_tokens": 559608632.0, + "step": 14419 + }, + { + "epoch": 1.8343722172751558, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.483254432678223, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8832482099533081, + "num_tokens": 559649105.0, + "step": 14420 + }, + { + "epoch": 1.8344994275537463, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.225187301635742, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8949858546257019, + "num_tokens": 559691737.0, + "step": 14421 + }, + { + "epoch": 1.8346266378323368, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.232661247253418, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.901208758354187, + "num_tokens": 559726579.0, + "step": 14422 + }, + { + "epoch": 1.8347538481109273, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.288026809692383, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8844062089920044, + "num_tokens": 559772792.0, + "step": 14423 + }, + { + "epoch": 1.8348810583895179, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.240119934082031, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8955419063568115, + "num_tokens": 559815367.0, + "step": 14424 + }, + { + "epoch": 1.8350082686681084, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.372404098510742, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8735199570655823, + "num_tokens": 559856940.0, + "step": 14425 + }, + { + "epoch": 1.835135478946699, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.33586311340332, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8957719206809998, + "num_tokens": 559894530.0, + "step": 14426 + }, + { + "epoch": 1.8352626892252895, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.411822319030762, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8878459930419922, + "num_tokens": 559927701.0, + "step": 14427 + }, + { + "epoch": 1.83538989950388, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.333436012268066, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8953603506088257, + "num_tokens": 559963438.0, + "step": 14428 + }, + { + "epoch": 1.8355171097824705, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.21857738494873, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8937399983406067, + "num_tokens": 560004054.0, + "step": 14429 + }, + { + "epoch": 1.835644320061061, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.351031303405762, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8849360942840576, + "num_tokens": 560042377.0, + "step": 14430 + }, + { + "epoch": 1.8357715303396516, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.26050853729248, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8912299275398254, + "num_tokens": 560088005.0, + "step": 14431 + }, + { + "epoch": 1.835898740618242, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.239277839660645, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.892371654510498, + "num_tokens": 560127579.0, + "step": 14432 + }, + { + "epoch": 1.8360259508968326, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.201513290405273, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8906075954437256, + "num_tokens": 560169690.0, + "step": 14433 + }, + { + "epoch": 1.836153161175423, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.305821418762207, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8841907978057861, + "num_tokens": 560208913.0, + "step": 14434 + }, + { + "epoch": 1.8362803714540135, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.21364974975586, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.892408549785614, + "num_tokens": 560250110.0, + "step": 14435 + }, + { + "epoch": 1.836407581732604, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.300111770629883, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8851071000099182, + "num_tokens": 560290665.0, + "step": 14436 + }, + { + "epoch": 1.8365347920111945, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.270048141479492, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.881598174571991, + "num_tokens": 560332281.0, + "step": 14437 + }, + { + "epoch": 1.836662002289785, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.343056678771973, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8918755054473877, + "num_tokens": 560369095.0, + "step": 14438 + }, + { + "epoch": 1.8367892125683756, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.270081520080566, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8853577375411987, + "num_tokens": 560406721.0, + "step": 14439 + }, + { + "epoch": 1.8369164228469659, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.216645240783691, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8889668583869934, + "num_tokens": 560450642.0, + "step": 14440 + }, + { + "epoch": 1.8370436331255564, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.363555908203125, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8812822103500366, + "num_tokens": 560484802.0, + "step": 14441 + }, + { + "epoch": 1.837170843404147, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.302385330200195, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.88828444480896, + "num_tokens": 560524533.0, + "step": 14442 + }, + { + "epoch": 1.8372980536827375, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.250092506408691, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.897730827331543, + "num_tokens": 560565945.0, + "step": 14443 + }, + { + "epoch": 1.837425263961328, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.261544227600098, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8850177526473999, + "num_tokens": 560606167.0, + "step": 14444 + }, + { + "epoch": 1.8375524742399185, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.252900123596191, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8952460289001465, + "num_tokens": 560644153.0, + "step": 14445 + }, + { + "epoch": 1.837679684518509, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.322081565856934, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8927001357078552, + "num_tokens": 560681612.0, + "step": 14446 + }, + { + "epoch": 1.8378068947970996, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.322750091552734, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8866249322891235, + "num_tokens": 560720849.0, + "step": 14447 + }, + { + "epoch": 1.83793410507569, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.269699096679688, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8934757113456726, + "num_tokens": 560759991.0, + "step": 14448 + }, + { + "epoch": 1.8380613153542806, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.335460662841797, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8957223892211914, + "num_tokens": 560794054.0, + "step": 14449 + }, + { + "epoch": 1.8381885256328712, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.219128608703613, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.9023342132568359, + "num_tokens": 560833035.0, + "step": 14450 + }, + { + "epoch": 1.8383157359114617, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.264663696289062, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8882858753204346, + "num_tokens": 560875228.0, + "step": 14451 + }, + { + "epoch": 1.8384429461900522, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.215900421142578, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8983644843101501, + "num_tokens": 560919728.0, + "step": 14452 + }, + { + "epoch": 1.8385701564686427, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.288141250610352, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8828138709068298, + "num_tokens": 560963329.0, + "step": 14453 + }, + { + "epoch": 1.8386973667472333, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.263127326965332, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9011296033859253, + "num_tokens": 561003904.0, + "step": 14454 + }, + { + "epoch": 1.8388245770258238, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.31452751159668, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8750016093254089, + "num_tokens": 561041646.0, + "step": 14455 + }, + { + "epoch": 1.8389517873044143, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.278085708618164, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8794316649436951, + "num_tokens": 561083939.0, + "step": 14456 + }, + { + "epoch": 1.8390789975830049, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.268365859985352, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8932997584342957, + "num_tokens": 561128320.0, + "step": 14457 + }, + { + "epoch": 1.8392062078615952, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.324456214904785, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8865281343460083, + "num_tokens": 561166902.0, + "step": 14458 + }, + { + "epoch": 1.8393334181401857, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.358407974243164, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8916325569152832, + "num_tokens": 561200317.0, + "step": 14459 + }, + { + "epoch": 1.8394606284187762, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.380389213562012, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8775228261947632, + "num_tokens": 561233895.0, + "step": 14460 + }, + { + "epoch": 1.8395878386973668, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.359052658081055, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8854135274887085, + "num_tokens": 561269056.0, + "step": 14461 + }, + { + "epoch": 1.8397150489759573, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.280449867248535, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8917635679244995, + "num_tokens": 561306396.0, + "step": 14462 + }, + { + "epoch": 1.8398422592545478, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.284716606140137, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8969114422798157, + "num_tokens": 561348744.0, + "step": 14463 + }, + { + "epoch": 1.8399694695331381, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.262072563171387, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8893937468528748, + "num_tokens": 561387521.0, + "step": 14464 + }, + { + "epoch": 1.8400966798117286, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.409197807312012, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8853830099105835, + "num_tokens": 561423989.0, + "step": 14465 + }, + { + "epoch": 1.8402238900903192, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.30700397491455, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8940631151199341, + "num_tokens": 561462540.0, + "step": 14466 + }, + { + "epoch": 1.8403511003689097, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.309881210327148, + "learning_rate": 1e-06, + "loss": 0.3881, + "mean_token_accuracy": 0.8706568479537964, + "num_tokens": 561509437.0, + "step": 14467 + }, + { + "epoch": 1.8404783106475002, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.345073699951172, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8846577405929565, + "num_tokens": 561552203.0, + "step": 14468 + }, + { + "epoch": 1.8406055209260908, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.306694984436035, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8779637217521667, + "num_tokens": 561591812.0, + "step": 14469 + }, + { + "epoch": 1.8407327312046813, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.316527366638184, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8879894018173218, + "num_tokens": 561637061.0, + "step": 14470 + }, + { + "epoch": 1.8408599414832718, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.314239501953125, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8999308347702026, + "num_tokens": 561672383.0, + "step": 14471 + }, + { + "epoch": 1.8409871517618623, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.2451753616333, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8972744941711426, + "num_tokens": 561713222.0, + "step": 14472 + }, + { + "epoch": 1.8411143620404529, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 10.098039627075195, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8854020833969116, + "num_tokens": 561749794.0, + "step": 14473 + }, + { + "epoch": 1.8412415723190434, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.374114036560059, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8764820694923401, + "num_tokens": 561785022.0, + "step": 14474 + }, + { + "epoch": 1.841368782597634, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.261687278747559, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8874750137329102, + "num_tokens": 561820899.0, + "step": 14475 + }, + { + "epoch": 1.8414959928762245, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.342864036560059, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.895816445350647, + "num_tokens": 561858903.0, + "step": 14476 + }, + { + "epoch": 1.841623203154815, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.264744758605957, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8892000913619995, + "num_tokens": 561898214.0, + "step": 14477 + }, + { + "epoch": 1.8417504134334055, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.263526916503906, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8971126079559326, + "num_tokens": 561931953.0, + "step": 14478 + }, + { + "epoch": 1.841877623711996, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.28337287902832, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8891750574111938, + "num_tokens": 561971829.0, + "step": 14479 + }, + { + "epoch": 1.8420048339905866, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.276436805725098, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8866660594940186, + "num_tokens": 562012786.0, + "step": 14480 + }, + { + "epoch": 1.842132044269177, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.319314956665039, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.888668417930603, + "num_tokens": 562049306.0, + "step": 14481 + }, + { + "epoch": 1.8422592545477676, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.420210838317871, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8820832371711731, + "num_tokens": 562081188.0, + "step": 14482 + }, + { + "epoch": 1.842386464826358, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.225001335144043, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8996889591217041, + "num_tokens": 562121228.0, + "step": 14483 + }, + { + "epoch": 1.8425136751049485, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.374967575073242, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8781524896621704, + "num_tokens": 562157519.0, + "step": 14484 + }, + { + "epoch": 1.842640885383539, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.363626480102539, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8879403471946716, + "num_tokens": 562192810.0, + "step": 14485 + }, + { + "epoch": 1.8427680956621295, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.299590110778809, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8907769322395325, + "num_tokens": 562232510.0, + "step": 14486 + }, + { + "epoch": 1.84289530594072, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.343029022216797, + "learning_rate": 1e-06, + "loss": 0.4429, + "mean_token_accuracy": 0.8558444976806641, + "num_tokens": 562276520.0, + "step": 14487 + }, + { + "epoch": 1.8430225162193106, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.236440658569336, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.887370228767395, + "num_tokens": 562320068.0, + "step": 14488 + }, + { + "epoch": 1.8431497264979009, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.303535461425781, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8859969973564148, + "num_tokens": 562358201.0, + "step": 14489 + }, + { + "epoch": 1.8432769367764914, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.371009826660156, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8948650360107422, + "num_tokens": 562390141.0, + "step": 14490 + }, + { + "epoch": 1.843404147055082, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.31557559967041, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8860680460929871, + "num_tokens": 562427444.0, + "step": 14491 + }, + { + "epoch": 1.8435313573336725, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.354715347290039, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8858151435852051, + "num_tokens": 562465236.0, + "step": 14492 + }, + { + "epoch": 1.843658567612263, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.190580368041992, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.889350414276123, + "num_tokens": 562508790.0, + "step": 14493 + }, + { + "epoch": 1.8437857778908535, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.278046607971191, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.9035822749137878, + "num_tokens": 562546161.0, + "step": 14494 + }, + { + "epoch": 1.843912988169444, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.307043075561523, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.8766854405403137, + "num_tokens": 562589925.0, + "step": 14495 + }, + { + "epoch": 1.8440401984480346, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.324999809265137, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8872321844100952, + "num_tokens": 562627970.0, + "step": 14496 + }, + { + "epoch": 1.844167408726625, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.370353698730469, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8858544230461121, + "num_tokens": 562667060.0, + "step": 14497 + }, + { + "epoch": 1.8442946190052156, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.309017181396484, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8749890923500061, + "num_tokens": 562709977.0, + "step": 14498 + }, + { + "epoch": 1.8444218292838062, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.306982040405273, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.9034571647644043, + "num_tokens": 562743720.0, + "step": 14499 + }, + { + "epoch": 1.8445490395623967, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.411919593811035, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.886756420135498, + "num_tokens": 562776563.0, + "step": 14500 + }, + { + "epoch": 1.8446762498409872, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.288911819458008, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8852907419204712, + "num_tokens": 562823257.0, + "step": 14501 + }, + { + "epoch": 1.8448034601195777, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.40750503540039, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8854707479476929, + "num_tokens": 562856639.0, + "step": 14502 + }, + { + "epoch": 1.8449306703981683, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.287569046020508, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8916453123092651, + "num_tokens": 562894416.0, + "step": 14503 + }, + { + "epoch": 1.8450578806767588, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.345786094665527, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8932058215141296, + "num_tokens": 562928473.0, + "step": 14504 + }, + { + "epoch": 1.8451850909553493, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.315895080566406, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8834143280982971, + "num_tokens": 562972742.0, + "step": 14505 + }, + { + "epoch": 1.8453123012339399, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.342392921447754, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8922972083091736, + "num_tokens": 563008897.0, + "step": 14506 + }, + { + "epoch": 1.8454395115125302, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.292082786560059, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8879200220108032, + "num_tokens": 563050159.0, + "step": 14507 + }, + { + "epoch": 1.8455667217911207, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.351251602172852, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8797448873519897, + "num_tokens": 563091602.0, + "step": 14508 + }, + { + "epoch": 1.8456939320697112, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.228283882141113, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8974530696868896, + "num_tokens": 563133248.0, + "step": 14509 + }, + { + "epoch": 1.8458211423483017, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.340136528015137, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8877441883087158, + "num_tokens": 563172935.0, + "step": 14510 + }, + { + "epoch": 1.8459483526268923, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.323686599731445, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8952419757843018, + "num_tokens": 563210132.0, + "step": 14511 + }, + { + "epoch": 1.8460755629054828, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.227189064025879, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8952531814575195, + "num_tokens": 563252979.0, + "step": 14512 + }, + { + "epoch": 1.846202773184073, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.359213829040527, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8798903226852417, + "num_tokens": 563292869.0, + "step": 14513 + }, + { + "epoch": 1.8463299834626636, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.302556037902832, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8912443518638611, + "num_tokens": 563328389.0, + "step": 14514 + }, + { + "epoch": 1.8464571937412542, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.410388946533203, + "learning_rate": 1e-06, + "loss": 0.4233, + "mean_token_accuracy": 0.8601434230804443, + "num_tokens": 563368603.0, + "step": 14515 + }, + { + "epoch": 1.8465844040198447, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.273581504821777, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8839492797851562, + "num_tokens": 563409293.0, + "step": 14516 + }, + { + "epoch": 1.8467116142984352, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.2864351272583, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8981364965438843, + "num_tokens": 563443981.0, + "step": 14517 + }, + { + "epoch": 1.8468388245770258, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.348130226135254, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8857243061065674, + "num_tokens": 563478666.0, + "step": 14518 + }, + { + "epoch": 1.8469660348556163, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.231051445007324, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8939511775970459, + "num_tokens": 563519002.0, + "step": 14519 + }, + { + "epoch": 1.8470932451342068, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.378311157226562, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8956552743911743, + "num_tokens": 563550979.0, + "step": 14520 + }, + { + "epoch": 1.8472204554127973, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.309208869934082, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.880845308303833, + "num_tokens": 563587492.0, + "step": 14521 + }, + { + "epoch": 1.8473476656913879, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.278802871704102, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8899561166763306, + "num_tokens": 563624908.0, + "step": 14522 + }, + { + "epoch": 1.8474748759699784, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.292061805725098, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8916795253753662, + "num_tokens": 563666862.0, + "step": 14523 + }, + { + "epoch": 1.847602086248569, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.26911735534668, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8912283778190613, + "num_tokens": 563706324.0, + "step": 14524 + }, + { + "epoch": 1.8477292965271594, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.44243335723877, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8889555931091309, + "num_tokens": 563743146.0, + "step": 14525 + }, + { + "epoch": 1.84785650680575, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.27152156829834, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.896755576133728, + "num_tokens": 563778645.0, + "step": 14526 + }, + { + "epoch": 1.8479837170843405, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.391427993774414, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8800535202026367, + "num_tokens": 563815866.0, + "step": 14527 + }, + { + "epoch": 1.848110927362931, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.354207992553711, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8783628344535828, + "num_tokens": 563856783.0, + "step": 14528 + }, + { + "epoch": 1.8482381376415216, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.292150497436523, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8896002769470215, + "num_tokens": 563895436.0, + "step": 14529 + }, + { + "epoch": 1.848365347920112, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.263214111328125, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8888171911239624, + "num_tokens": 563935620.0, + "step": 14530 + }, + { + "epoch": 1.8484925581987026, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.328471183776855, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8852163553237915, + "num_tokens": 563974860.0, + "step": 14531 + }, + { + "epoch": 1.848619768477293, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.331151008605957, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8843053579330444, + "num_tokens": 564016596.0, + "step": 14532 + }, + { + "epoch": 1.8487469787558835, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.348796844482422, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.889549732208252, + "num_tokens": 564054037.0, + "step": 14533 + }, + { + "epoch": 1.848874189034474, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.246866226196289, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8931066393852234, + "num_tokens": 564097322.0, + "step": 14534 + }, + { + "epoch": 1.8490013993130645, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.394326210021973, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8828017711639404, + "num_tokens": 564134155.0, + "step": 14535 + }, + { + "epoch": 1.849128609591655, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.364391326904297, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8762297034263611, + "num_tokens": 564175034.0, + "step": 14536 + }, + { + "epoch": 1.8492558198702456, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.329580307006836, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8929653167724609, + "num_tokens": 564208987.0, + "step": 14537 + }, + { + "epoch": 1.8493830301488359, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.322850227355957, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8908668160438538, + "num_tokens": 564249597.0, + "step": 14538 + }, + { + "epoch": 1.8495102404274264, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.209527969360352, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8804209232330322, + "num_tokens": 564291644.0, + "step": 14539 + }, + { + "epoch": 1.849637450706017, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.296774864196777, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8833519816398621, + "num_tokens": 564333393.0, + "step": 14540 + }, + { + "epoch": 1.8497646609846075, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.426937103271484, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8843966722488403, + "num_tokens": 564366053.0, + "step": 14541 + }, + { + "epoch": 1.849891871263198, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.293440818786621, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8922114372253418, + "num_tokens": 564403844.0, + "step": 14542 + }, + { + "epoch": 1.8500190815417885, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.26018238067627, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8855740427970886, + "num_tokens": 564444369.0, + "step": 14543 + }, + { + "epoch": 1.850146291820379, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.38603687286377, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8855252861976624, + "num_tokens": 564479763.0, + "step": 14544 + }, + { + "epoch": 1.8502735020989696, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.351766586303711, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8987573385238647, + "num_tokens": 564513448.0, + "step": 14545 + }, + { + "epoch": 1.85040071237756, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.217453956604004, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9040612578392029, + "num_tokens": 564551589.0, + "step": 14546 + }, + { + "epoch": 1.8505279226561506, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.268109321594238, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8920704126358032, + "num_tokens": 564590601.0, + "step": 14547 + }, + { + "epoch": 1.8506551329347412, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.34698486328125, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8894525766372681, + "num_tokens": 564625091.0, + "step": 14548 + }, + { + "epoch": 1.8507823432133317, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.218559265136719, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8886376023292542, + "num_tokens": 564670082.0, + "step": 14549 + }, + { + "epoch": 1.8509095534919222, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.253523826599121, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9046071767807007, + "num_tokens": 564710815.0, + "step": 14550 + }, + { + "epoch": 1.8510367637705127, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.365538597106934, + "learning_rate": 1e-06, + "loss": 0.3773, + "mean_token_accuracy": 0.8721710443496704, + "num_tokens": 564744487.0, + "step": 14551 + }, + { + "epoch": 1.8511639740491033, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.313274383544922, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8916343450546265, + "num_tokens": 564786165.0, + "step": 14552 + }, + { + "epoch": 1.8512911843276938, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.314408302307129, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8966651558876038, + "num_tokens": 564816322.0, + "step": 14553 + }, + { + "epoch": 1.8514183946062843, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.361923217773438, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8766824007034302, + "num_tokens": 564851803.0, + "step": 14554 + }, + { + "epoch": 1.8515456048848749, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.38696575164795, + "learning_rate": 1e-06, + "loss": 0.3971, + "mean_token_accuracy": 0.8662687540054321, + "num_tokens": 564889764.0, + "step": 14555 + }, + { + "epoch": 1.8516728151634652, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.213062286376953, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8880135416984558, + "num_tokens": 564933254.0, + "step": 14556 + }, + { + "epoch": 1.8518000254420557, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.243060111999512, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8898869752883911, + "num_tokens": 564975171.0, + "step": 14557 + }, + { + "epoch": 1.8519272357206462, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.413084983825684, + "learning_rate": 1e-06, + "loss": 0.3835, + "mean_token_accuracy": 0.8715424537658691, + "num_tokens": 565010262.0, + "step": 14558 + }, + { + "epoch": 1.8520544459992367, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.276057243347168, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8864947557449341, + "num_tokens": 565051419.0, + "step": 14559 + }, + { + "epoch": 1.8521816562778273, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.26008415222168, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8856772184371948, + "num_tokens": 565089898.0, + "step": 14560 + }, + { + "epoch": 1.8523088665564178, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.292472839355469, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.898221492767334, + "num_tokens": 565128259.0, + "step": 14561 + }, + { + "epoch": 1.852436076835008, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.327383995056152, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8969259262084961, + "num_tokens": 565165457.0, + "step": 14562 + }, + { + "epoch": 1.8525632871135986, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.253933906555176, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9048875570297241, + "num_tokens": 565209519.0, + "step": 14563 + }, + { + "epoch": 1.8526904973921892, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.410642623901367, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8803983926773071, + "num_tokens": 565248590.0, + "step": 14564 + }, + { + "epoch": 1.8528177076707797, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.313785552978516, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8949689269065857, + "num_tokens": 565288014.0, + "step": 14565 + }, + { + "epoch": 1.8529449179493702, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.293998718261719, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8985918760299683, + "num_tokens": 565325345.0, + "step": 14566 + }, + { + "epoch": 1.8530721282279607, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.395015716552734, + "learning_rate": 1e-06, + "loss": 0.392, + "mean_token_accuracy": 0.8642070293426514, + "num_tokens": 565364569.0, + "step": 14567 + }, + { + "epoch": 1.8531993385065513, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.3090181350708, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8959344625473022, + "num_tokens": 565402851.0, + "step": 14568 + }, + { + "epoch": 1.8533265487851418, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.323801040649414, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.888281524181366, + "num_tokens": 565442916.0, + "step": 14569 + }, + { + "epoch": 1.8534537590637323, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.388235092163086, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.892656683921814, + "num_tokens": 565478124.0, + "step": 14570 + }, + { + "epoch": 1.8535809693423229, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.240080833435059, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8801194429397583, + "num_tokens": 565525541.0, + "step": 14571 + }, + { + "epoch": 1.8537081796209134, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.303665161132812, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8845837116241455, + "num_tokens": 565569754.0, + "step": 14572 + }, + { + "epoch": 1.853835389899504, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.418486595153809, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8909599781036377, + "num_tokens": 565603841.0, + "step": 14573 + }, + { + "epoch": 1.8539626001780944, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.31966495513916, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8947738409042358, + "num_tokens": 565635644.0, + "step": 14574 + }, + { + "epoch": 1.854089810456685, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.345904350280762, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.890964150428772, + "num_tokens": 565673173.0, + "step": 14575 + }, + { + "epoch": 1.8542170207352755, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.348808288574219, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.892349362373352, + "num_tokens": 565707747.0, + "step": 14576 + }, + { + "epoch": 1.854344231013866, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.315400123596191, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8823157548904419, + "num_tokens": 565744157.0, + "step": 14577 + }, + { + "epoch": 1.8544714412924566, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.303475379943848, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8830270767211914, + "num_tokens": 565783860.0, + "step": 14578 + }, + { + "epoch": 1.854598651571047, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.30134391784668, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8818601369857788, + "num_tokens": 565829961.0, + "step": 14579 + }, + { + "epoch": 1.8547258618496376, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.389137268066406, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8852720260620117, + "num_tokens": 565865058.0, + "step": 14580 + }, + { + "epoch": 1.854853072128228, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.322093963623047, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.88255774974823, + "num_tokens": 565902810.0, + "step": 14581 + }, + { + "epoch": 1.8549802824068184, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.337615966796875, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8948298096656799, + "num_tokens": 565940753.0, + "step": 14582 + }, + { + "epoch": 1.855107492685409, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.266744613647461, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9065093398094177, + "num_tokens": 565976891.0, + "step": 14583 + }, + { + "epoch": 1.8552347029639995, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.377456665039062, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8943068981170654, + "num_tokens": 566011710.0, + "step": 14584 + }, + { + "epoch": 1.85536191324259, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.35367488861084, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8899656534194946, + "num_tokens": 566049748.0, + "step": 14585 + }, + { + "epoch": 1.8554891235211806, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.362098693847656, + "learning_rate": 1e-06, + "loss": 0.3681, + "mean_token_accuracy": 0.8769158720970154, + "num_tokens": 566088919.0, + "step": 14586 + }, + { + "epoch": 1.8556163337997709, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.24657154083252, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9005845189094543, + "num_tokens": 566129985.0, + "step": 14587 + }, + { + "epoch": 1.8557435440783614, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.422059059143066, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.9003095626831055, + "num_tokens": 566165121.0, + "step": 14588 + }, + { + "epoch": 1.855870754356952, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.32391357421875, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.897343635559082, + "num_tokens": 566196609.0, + "step": 14589 + }, + { + "epoch": 1.8559979646355425, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.370292663574219, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8908551931381226, + "num_tokens": 566231806.0, + "step": 14590 + }, + { + "epoch": 1.856125174914133, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.466105461120605, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.891733705997467, + "num_tokens": 566267057.0, + "step": 14591 + }, + { + "epoch": 1.8562523851927235, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.361580848693848, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8803688287734985, + "num_tokens": 566301967.0, + "step": 14592 + }, + { + "epoch": 1.856379595471314, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.228550910949707, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8975171446800232, + "num_tokens": 566341475.0, + "step": 14593 + }, + { + "epoch": 1.8565068057499046, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.336832046508789, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8932682275772095, + "num_tokens": 566383162.0, + "step": 14594 + }, + { + "epoch": 1.856634016028495, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.299493789672852, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8779358863830566, + "num_tokens": 566423749.0, + "step": 14595 + }, + { + "epoch": 1.8567612263070856, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.264082908630371, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.897689938545227, + "num_tokens": 566465445.0, + "step": 14596 + }, + { + "epoch": 1.8568884365856761, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.329123497009277, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8890653848648071, + "num_tokens": 566503383.0, + "step": 14597 + }, + { + "epoch": 1.8570156468642667, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.276106834411621, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8830122351646423, + "num_tokens": 566546926.0, + "step": 14598 + }, + { + "epoch": 1.8571428571428572, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.270999908447266, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8905467987060547, + "num_tokens": 566585362.0, + "step": 14599 + }, + { + "epoch": 1.8572700674214477, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.383124351501465, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8838995695114136, + "num_tokens": 566621870.0, + "step": 14600 + }, + { + "epoch": 1.8573972777000383, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.27385425567627, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8935677409172058, + "num_tokens": 566662574.0, + "step": 14601 + }, + { + "epoch": 1.8575244879786288, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.319043159484863, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8817435503005981, + "num_tokens": 566702900.0, + "step": 14602 + }, + { + "epoch": 1.8576516982572193, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.376626014709473, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8907192349433899, + "num_tokens": 566736213.0, + "step": 14603 + }, + { + "epoch": 1.8577789085358098, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.287036895751953, + "learning_rate": 1e-06, + "loss": 0.3851, + "mean_token_accuracy": 0.8729691505432129, + "num_tokens": 566779119.0, + "step": 14604 + }, + { + "epoch": 1.8579061188144002, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.37076473236084, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.8794249296188354, + "num_tokens": 566813960.0, + "step": 14605 + }, + { + "epoch": 1.8580333290929907, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.26510238647461, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8914494514465332, + "num_tokens": 566858503.0, + "step": 14606 + }, + { + "epoch": 1.8581605393715812, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.320107460021973, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8787645101547241, + "num_tokens": 566899677.0, + "step": 14607 + }, + { + "epoch": 1.8582877496501717, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.381092071533203, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8793865442276001, + "num_tokens": 566935220.0, + "step": 14608 + }, + { + "epoch": 1.8584149599287623, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.304369926452637, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8894348740577698, + "num_tokens": 566975289.0, + "step": 14609 + }, + { + "epoch": 1.8585421702073528, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.45272445678711, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8850300312042236, + "num_tokens": 567007041.0, + "step": 14610 + }, + { + "epoch": 1.858669380485943, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.33480453491211, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8957622051239014, + "num_tokens": 567045062.0, + "step": 14611 + }, + { + "epoch": 1.8587965907645336, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.316307067871094, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8789293766021729, + "num_tokens": 567086018.0, + "step": 14612 + }, + { + "epoch": 1.8589238010431242, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.34321117401123, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8905966281890869, + "num_tokens": 567124387.0, + "step": 14613 + }, + { + "epoch": 1.8590510113217147, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.377347946166992, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8898738622665405, + "num_tokens": 567157723.0, + "step": 14614 + }, + { + "epoch": 1.8591782216003052, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.372447967529297, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8930399417877197, + "num_tokens": 567194407.0, + "step": 14615 + }, + { + "epoch": 1.8593054318788957, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.297303199768066, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8801677227020264, + "num_tokens": 567234386.0, + "step": 14616 + }, + { + "epoch": 1.8594326421574863, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.320175170898438, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8956383466720581, + "num_tokens": 567274640.0, + "step": 14617 + }, + { + "epoch": 1.8595598524360768, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.320550918579102, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8820887804031372, + "num_tokens": 567316408.0, + "step": 14618 + }, + { + "epoch": 1.8596870627146673, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.291315078735352, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8934776782989502, + "num_tokens": 567353292.0, + "step": 14619 + }, + { + "epoch": 1.8598142729932579, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.296887397766113, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8847930431365967, + "num_tokens": 567399942.0, + "step": 14620 + }, + { + "epoch": 1.8599414832718484, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.26616096496582, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8893042206764221, + "num_tokens": 567438551.0, + "step": 14621 + }, + { + "epoch": 1.860068693550439, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.31522274017334, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8833855390548706, + "num_tokens": 567477336.0, + "step": 14622 + }, + { + "epoch": 1.8601959038290294, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.300989151000977, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8827325105667114, + "num_tokens": 567521016.0, + "step": 14623 + }, + { + "epoch": 1.86032311410762, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.333587646484375, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8894699811935425, + "num_tokens": 567553656.0, + "step": 14624 + }, + { + "epoch": 1.8604503243862105, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.33700942993164, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8853615522384644, + "num_tokens": 567596067.0, + "step": 14625 + }, + { + "epoch": 1.860577534664801, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.436203956604004, + "learning_rate": 1e-06, + "loss": 0.409, + "mean_token_accuracy": 0.8648806810379028, + "num_tokens": 567632224.0, + "step": 14626 + }, + { + "epoch": 1.8607047449433916, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.32119369506836, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8754132986068726, + "num_tokens": 567673065.0, + "step": 14627 + }, + { + "epoch": 1.860831955221982, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.399542808532715, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8788841962814331, + "num_tokens": 567706246.0, + "step": 14628 + }, + { + "epoch": 1.8609591655005726, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.267008781433105, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8992366194725037, + "num_tokens": 567747819.0, + "step": 14629 + }, + { + "epoch": 1.861086375779163, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.30477237701416, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8858060836791992, + "num_tokens": 567788791.0, + "step": 14630 + }, + { + "epoch": 1.8612135860577534, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.239340782165527, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8876116275787354, + "num_tokens": 567832498.0, + "step": 14631 + }, + { + "epoch": 1.861340796336344, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.287371635437012, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8898501396179199, + "num_tokens": 567871435.0, + "step": 14632 + }, + { + "epoch": 1.8614680066149345, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.356671333312988, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8855575919151306, + "num_tokens": 567914418.0, + "step": 14633 + }, + { + "epoch": 1.861595216893525, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.32697868347168, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8908792734146118, + "num_tokens": 567952848.0, + "step": 14634 + }, + { + "epoch": 1.8617224271721156, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.347392082214355, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8814079761505127, + "num_tokens": 567988870.0, + "step": 14635 + }, + { + "epoch": 1.8618496374507059, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.381461143493652, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8944036960601807, + "num_tokens": 568021909.0, + "step": 14636 + }, + { + "epoch": 1.8619768477292964, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.333430290222168, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.9045211672782898, + "num_tokens": 568059070.0, + "step": 14637 + }, + { + "epoch": 1.862104058007887, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.316658020019531, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8968805074691772, + "num_tokens": 568095864.0, + "step": 14638 + }, + { + "epoch": 1.8622312682864774, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.399649620056152, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8794599771499634, + "num_tokens": 568133583.0, + "step": 14639 + }, + { + "epoch": 1.862358478565068, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.501768112182617, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8819187879562378, + "num_tokens": 568162894.0, + "step": 14640 + }, + { + "epoch": 1.8624856888436585, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.303936958312988, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8932532072067261, + "num_tokens": 568202456.0, + "step": 14641 + }, + { + "epoch": 1.862612899122249, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.37569808959961, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8812958002090454, + "num_tokens": 568240440.0, + "step": 14642 + }, + { + "epoch": 1.8627401094008396, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.300660133361816, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8949426412582397, + "num_tokens": 568275592.0, + "step": 14643 + }, + { + "epoch": 1.86286731967943, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.385541915893555, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8849464654922485, + "num_tokens": 568311632.0, + "step": 14644 + }, + { + "epoch": 1.8629945299580206, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.284637451171875, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.900396466255188, + "num_tokens": 568350432.0, + "step": 14645 + }, + { + "epoch": 1.8631217402366111, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.336387634277344, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8788225054740906, + "num_tokens": 568389980.0, + "step": 14646 + }, + { + "epoch": 1.8632489505152017, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.37286376953125, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.9041861891746521, + "num_tokens": 568428044.0, + "step": 14647 + }, + { + "epoch": 1.8633761607937922, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.341270446777344, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8859425187110901, + "num_tokens": 568472484.0, + "step": 14648 + }, + { + "epoch": 1.8635033710723827, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.318072319030762, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8799020648002625, + "num_tokens": 568511212.0, + "step": 14649 + }, + { + "epoch": 1.8636305813509733, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.374197959899902, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8959487676620483, + "num_tokens": 568547595.0, + "step": 14650 + }, + { + "epoch": 1.8637577916295638, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.234973907470703, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8981865644454956, + "num_tokens": 568588381.0, + "step": 14651 + }, + { + "epoch": 1.8638850019081543, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.336334228515625, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8881074786186218, + "num_tokens": 568627922.0, + "step": 14652 + }, + { + "epoch": 1.8640122121867448, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.32735538482666, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8946337103843689, + "num_tokens": 568670031.0, + "step": 14653 + }, + { + "epoch": 1.8641394224653351, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.37605094909668, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8741857409477234, + "num_tokens": 568708554.0, + "step": 14654 + }, + { + "epoch": 1.8642666327439257, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.381213188171387, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8879427313804626, + "num_tokens": 568747891.0, + "step": 14655 + }, + { + "epoch": 1.8643938430225162, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.257026672363281, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.898997962474823, + "num_tokens": 568786231.0, + "step": 14656 + }, + { + "epoch": 1.8645210533011067, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.354719161987305, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.888325572013855, + "num_tokens": 568828627.0, + "step": 14657 + }, + { + "epoch": 1.8646482635796973, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.324751853942871, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.890748143196106, + "num_tokens": 568863662.0, + "step": 14658 + }, + { + "epoch": 1.8647754738582878, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.310057640075684, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.892449676990509, + "num_tokens": 568902735.0, + "step": 14659 + }, + { + "epoch": 1.864902684136878, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.262900352478027, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8965880274772644, + "num_tokens": 568940182.0, + "step": 14660 + }, + { + "epoch": 1.8650298944154686, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.375789642333984, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8848662376403809, + "num_tokens": 568977856.0, + "step": 14661 + }, + { + "epoch": 1.8651571046940592, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.326166152954102, + "learning_rate": 1e-06, + "loss": 0.3884, + "mean_token_accuracy": 0.8713522553443909, + "num_tokens": 569019754.0, + "step": 14662 + }, + { + "epoch": 1.8652843149726497, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.279491424560547, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8867603540420532, + "num_tokens": 569059052.0, + "step": 14663 + }, + { + "epoch": 1.8654115252512402, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.311434745788574, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.892619252204895, + "num_tokens": 569094025.0, + "step": 14664 + }, + { + "epoch": 1.8655387355298307, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.20820140838623, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.900280237197876, + "num_tokens": 569133185.0, + "step": 14665 + }, + { + "epoch": 1.8656659458084213, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.4595308303833, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8812578916549683, + "num_tokens": 569167647.0, + "step": 14666 + }, + { + "epoch": 1.8657931560870118, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.210061073303223, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8890607357025146, + "num_tokens": 569209964.0, + "step": 14667 + }, + { + "epoch": 1.8659203663656023, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.255763053894043, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9093071222305298, + "num_tokens": 569248182.0, + "step": 14668 + }, + { + "epoch": 1.8660475766441929, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.369278907775879, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8968057632446289, + "num_tokens": 569282044.0, + "step": 14669 + }, + { + "epoch": 1.8661747869227834, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.332213401794434, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.9002140760421753, + "num_tokens": 569317301.0, + "step": 14670 + }, + { + "epoch": 1.866301997201374, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.296883583068848, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8992508053779602, + "num_tokens": 569355682.0, + "step": 14671 + }, + { + "epoch": 1.8664292074799644, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.27345085144043, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8932653069496155, + "num_tokens": 569399207.0, + "step": 14672 + }, + { + "epoch": 1.866556417758555, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.244412422180176, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8909921646118164, + "num_tokens": 569441009.0, + "step": 14673 + }, + { + "epoch": 1.8666836280371455, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.388409614562988, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8841211795806885, + "num_tokens": 569482335.0, + "step": 14674 + }, + { + "epoch": 1.866810838315736, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.342028617858887, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8872472047805786, + "num_tokens": 569519993.0, + "step": 14675 + }, + { + "epoch": 1.8669380485943265, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.328710556030273, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8920565843582153, + "num_tokens": 569560325.0, + "step": 14676 + }, + { + "epoch": 1.867065258872917, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.263461112976074, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8898720145225525, + "num_tokens": 569603559.0, + "step": 14677 + }, + { + "epoch": 1.8671924691515076, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.332096099853516, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.881858229637146, + "num_tokens": 569645424.0, + "step": 14678 + }, + { + "epoch": 1.867319679430098, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.33989429473877, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8923236131668091, + "num_tokens": 569676959.0, + "step": 14679 + }, + { + "epoch": 1.8674468897086884, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.359627723693848, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8868976831436157, + "num_tokens": 569721719.0, + "step": 14680 + }, + { + "epoch": 1.867574099987279, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.27995777130127, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8870288133621216, + "num_tokens": 569763908.0, + "step": 14681 + }, + { + "epoch": 1.8677013102658695, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.367913246154785, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8745290040969849, + "num_tokens": 569796656.0, + "step": 14682 + }, + { + "epoch": 1.86782852054446, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.336416244506836, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8831973075866699, + "num_tokens": 569838716.0, + "step": 14683 + }, + { + "epoch": 1.8679557308230506, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.272037506103516, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.9028182029724121, + "num_tokens": 569875536.0, + "step": 14684 + }, + { + "epoch": 1.8680829411016409, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.363568305969238, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8909252882003784, + "num_tokens": 569913227.0, + "step": 14685 + }, + { + "epoch": 1.8682101513802314, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.311746597290039, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8824113011360168, + "num_tokens": 569949586.0, + "step": 14686 + }, + { + "epoch": 1.868337361658822, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.353010177612305, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8858307003974915, + "num_tokens": 569987285.0, + "step": 14687 + }, + { + "epoch": 1.8684645719374124, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.288921356201172, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8915220499038696, + "num_tokens": 570025596.0, + "step": 14688 + }, + { + "epoch": 1.868591782216003, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.323786735534668, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8809992074966431, + "num_tokens": 570068264.0, + "step": 14689 + }, + { + "epoch": 1.8687189924945935, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.306414604187012, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8923805952072144, + "num_tokens": 570108023.0, + "step": 14690 + }, + { + "epoch": 1.868846202773184, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.315899848937988, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8906324505805969, + "num_tokens": 570145962.0, + "step": 14691 + }, + { + "epoch": 1.8689734130517746, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.1978759765625, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8860886096954346, + "num_tokens": 570188410.0, + "step": 14692 + }, + { + "epoch": 1.869100623330365, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.354375839233398, + "learning_rate": 1e-06, + "loss": 0.415, + "mean_token_accuracy": 0.8618173003196716, + "num_tokens": 570238853.0, + "step": 14693 + }, + { + "epoch": 1.8692278336089556, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.27180290222168, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9034837484359741, + "num_tokens": 570276651.0, + "step": 14694 + }, + { + "epoch": 1.8693550438875461, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.378804206848145, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8905655145645142, + "num_tokens": 570310244.0, + "step": 14695 + }, + { + "epoch": 1.8694822541661367, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.329339027404785, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.897114634513855, + "num_tokens": 570343305.0, + "step": 14696 + }, + { + "epoch": 1.8696094644447272, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.305706024169922, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8819645047187805, + "num_tokens": 570392078.0, + "step": 14697 + }, + { + "epoch": 1.8697366747233177, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.250165939331055, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9104212522506714, + "num_tokens": 570431692.0, + "step": 14698 + }, + { + "epoch": 1.8698638850019083, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.319305419921875, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8975966572761536, + "num_tokens": 570464852.0, + "step": 14699 + }, + { + "epoch": 1.8699910952804988, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.391279220581055, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.88271564245224, + "num_tokens": 570504075.0, + "step": 14700 + }, + { + "epoch": 1.8701183055590893, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.223740577697754, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8829329013824463, + "num_tokens": 570547108.0, + "step": 14701 + }, + { + "epoch": 1.8702455158376798, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.35064697265625, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8880354166030884, + "num_tokens": 570585914.0, + "step": 14702 + }, + { + "epoch": 1.8703727261162701, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.31186294555664, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8805581331253052, + "num_tokens": 570626429.0, + "step": 14703 + }, + { + "epoch": 1.8704999363948607, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.286909103393555, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8875566720962524, + "num_tokens": 570668612.0, + "step": 14704 + }, + { + "epoch": 1.8706271466734512, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.286938667297363, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8953617215156555, + "num_tokens": 570707128.0, + "step": 14705 + }, + { + "epoch": 1.8707543569520417, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.375024795532227, + "learning_rate": 1e-06, + "loss": 0.3753, + "mean_token_accuracy": 0.8739194869995117, + "num_tokens": 570745404.0, + "step": 14706 + }, + { + "epoch": 1.8708815672306323, + "ewc_loss": 0.0169677734375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0669231414794922e-05, + "grad_norm": 9.278585433959961, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8874787092208862, + "num_tokens": 570787121.0, + "step": 14707 + }, + { + "epoch": 1.8710087775092228, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.242417335510254, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8989223837852478, + "num_tokens": 570832219.0, + "step": 14708 + }, + { + "epoch": 1.871135987787813, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.375385284423828, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8869237899780273, + "num_tokens": 570870443.0, + "step": 14709 + }, + { + "epoch": 1.8712631980664036, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.310511589050293, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8862336874008179, + "num_tokens": 570914271.0, + "step": 14710 + }, + { + "epoch": 1.8713904083449941, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.27042007446289, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8920977711677551, + "num_tokens": 570960946.0, + "step": 14711 + }, + { + "epoch": 1.8715176186235847, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.312045097351074, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8815594911575317, + "num_tokens": 571003820.0, + "step": 14712 + }, + { + "epoch": 1.8716448289021752, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.496872901916504, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.8759796023368835, + "num_tokens": 571041823.0, + "step": 14713 + }, + { + "epoch": 1.8717720391807657, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.294662475585938, + "learning_rate": 1e-06, + "loss": 0.3774, + "mean_token_accuracy": 0.8726674318313599, + "num_tokens": 571085007.0, + "step": 14714 + }, + { + "epoch": 1.8718992494593563, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.288949966430664, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8971683382987976, + "num_tokens": 571122798.0, + "step": 14715 + }, + { + "epoch": 1.8720264597379468, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.34316349029541, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8929162621498108, + "num_tokens": 571160590.0, + "step": 14716 + }, + { + "epoch": 1.8721536700165373, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.277436256408691, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8943698406219482, + "num_tokens": 571202284.0, + "step": 14717 + }, + { + "epoch": 1.8722808802951278, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.372825622558594, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8892548084259033, + "num_tokens": 571238555.0, + "step": 14718 + }, + { + "epoch": 1.8724080905737184, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.329721450805664, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8910351991653442, + "num_tokens": 571273742.0, + "step": 14719 + }, + { + "epoch": 1.872535300852309, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.371451377868652, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8863884806632996, + "num_tokens": 571311078.0, + "step": 14720 + }, + { + "epoch": 1.8726625111308994, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.359421730041504, + "learning_rate": 1e-06, + "loss": 0.3909, + "mean_token_accuracy": 0.8666187524795532, + "num_tokens": 571353135.0, + "step": 14721 + }, + { + "epoch": 1.87278972140949, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.322640419006348, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8837795257568359, + "num_tokens": 571392822.0, + "step": 14722 + }, + { + "epoch": 1.8729169316880805, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.341581344604492, + "learning_rate": 1e-06, + "loss": 0.3826, + "mean_token_accuracy": 0.8704309463500977, + "num_tokens": 571434452.0, + "step": 14723 + }, + { + "epoch": 1.873044141966671, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.291220664978027, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8900086879730225, + "num_tokens": 571469951.0, + "step": 14724 + }, + { + "epoch": 1.8731713522452615, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.194830894470215, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8983197808265686, + "num_tokens": 571516241.0, + "step": 14725 + }, + { + "epoch": 1.873298562523852, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.349554061889648, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8924261331558228, + "num_tokens": 571552441.0, + "step": 14726 + }, + { + "epoch": 1.8734257728024426, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.338788032531738, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8922366499900818, + "num_tokens": 571590261.0, + "step": 14727 + }, + { + "epoch": 1.873552983081033, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.345057487487793, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8811467885971069, + "num_tokens": 571630140.0, + "step": 14728 + }, + { + "epoch": 1.8736801933596234, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.30765151977539, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8914610743522644, + "num_tokens": 571672579.0, + "step": 14729 + }, + { + "epoch": 1.873807403638214, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.28963565826416, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8911987543106079, + "num_tokens": 571712377.0, + "step": 14730 + }, + { + "epoch": 1.8739346139168045, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.279274940490723, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8961232304573059, + "num_tokens": 571749182.0, + "step": 14731 + }, + { + "epoch": 1.874061824195395, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.265624046325684, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8947486877441406, + "num_tokens": 571794272.0, + "step": 14732 + }, + { + "epoch": 1.8741890344739855, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.338120460510254, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.878321647644043, + "num_tokens": 571830547.0, + "step": 14733 + }, + { + "epoch": 1.8743162447525759, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.31832218170166, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8898310661315918, + "num_tokens": 571872120.0, + "step": 14734 + }, + { + "epoch": 1.8744434550311664, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.352380752563477, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8983343839645386, + "num_tokens": 571908375.0, + "step": 14735 + }, + { + "epoch": 1.874570665309757, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.281930923461914, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8875157237052917, + "num_tokens": 571949807.0, + "step": 14736 + }, + { + "epoch": 1.8746978755883474, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.35981559753418, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8836532831192017, + "num_tokens": 571987362.0, + "step": 14737 + }, + { + "epoch": 1.874825085866938, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.322307586669922, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8725606203079224, + "num_tokens": 572028897.0, + "step": 14738 + }, + { + "epoch": 1.8749522961455285, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.34522533416748, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8832061886787415, + "num_tokens": 572066566.0, + "step": 14739 + }, + { + "epoch": 1.875079506424119, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.337303161621094, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8907914757728577, + "num_tokens": 572103830.0, + "step": 14740 + }, + { + "epoch": 1.8752067167027096, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.248529434204102, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8869423270225525, + "num_tokens": 572144459.0, + "step": 14741 + }, + { + "epoch": 1.8753339269813, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.34066390991211, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8956291675567627, + "num_tokens": 572179186.0, + "step": 14742 + }, + { + "epoch": 1.8754611372598906, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.277610778808594, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8852753639221191, + "num_tokens": 572220772.0, + "step": 14743 + }, + { + "epoch": 1.8755883475384811, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.300699234008789, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8828170299530029, + "num_tokens": 572262144.0, + "step": 14744 + }, + { + "epoch": 1.8757155578170717, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.323580741882324, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.8702254891395569, + "num_tokens": 572306053.0, + "step": 14745 + }, + { + "epoch": 1.8758427680956622, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.345239639282227, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8859291076660156, + "num_tokens": 572347811.0, + "step": 14746 + }, + { + "epoch": 1.8759699783742527, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.386283874511719, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8903257250785828, + "num_tokens": 572384103.0, + "step": 14747 + }, + { + "epoch": 1.8760971886528433, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.391207695007324, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8801790475845337, + "num_tokens": 572422518.0, + "step": 14748 + }, + { + "epoch": 1.8762243989314338, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.21097183227539, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8932390213012695, + "num_tokens": 572462993.0, + "step": 14749 + }, + { + "epoch": 1.8763516092100243, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.262383460998535, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8830101490020752, + "num_tokens": 572504737.0, + "step": 14750 + }, + { + "epoch": 1.8764788194886148, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.362351417541504, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8908810615539551, + "num_tokens": 572537710.0, + "step": 14751 + }, + { + "epoch": 1.8766060297672051, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.291021347045898, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.875572681427002, + "num_tokens": 572578010.0, + "step": 14752 + }, + { + "epoch": 1.8767332400457957, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.28825855255127, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8847134113311768, + "num_tokens": 572622031.0, + "step": 14753 + }, + { + "epoch": 1.8768604503243862, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.4737548828125, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8714574575424194, + "num_tokens": 572656106.0, + "step": 14754 + }, + { + "epoch": 1.8769876606029767, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.262701988220215, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8788096904754639, + "num_tokens": 572695912.0, + "step": 14755 + }, + { + "epoch": 1.8771148708815673, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.390581130981445, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8833188414573669, + "num_tokens": 572732027.0, + "step": 14756 + }, + { + "epoch": 1.8772420811601578, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.29425048828125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.26632022857666, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8906457424163818, + "num_tokens": 572774331.0, + "step": 14757 + }, + { + "epoch": 1.877369291438748, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.382698059082031, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8824900984764099, + "num_tokens": 572810751.0, + "step": 14758 + }, + { + "epoch": 1.8774965017173386, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.378003120422363, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8853073120117188, + "num_tokens": 572847212.0, + "step": 14759 + }, + { + "epoch": 1.8776237119959291, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.342387199401855, + "learning_rate": 1e-06, + "loss": 0.3929, + "mean_token_accuracy": 0.868403434753418, + "num_tokens": 572889425.0, + "step": 14760 + }, + { + "epoch": 1.8777509222745197, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.313760757446289, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8871662616729736, + "num_tokens": 572928564.0, + "step": 14761 + }, + { + "epoch": 1.8778781325531102, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.256756782531738, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8938713073730469, + "num_tokens": 572967143.0, + "step": 14762 + }, + { + "epoch": 1.8780053428317007, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.248587608337402, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.9055783152580261, + "num_tokens": 573008107.0, + "step": 14763 + }, + { + "epoch": 1.8781325531102913, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.2798490524292, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8993182182312012, + "num_tokens": 573042782.0, + "step": 14764 + }, + { + "epoch": 1.8782597633888818, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.49140453338623, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8866512179374695, + "num_tokens": 573074353.0, + "step": 14765 + }, + { + "epoch": 1.8783869736674723, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.243427276611328, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8838026523590088, + "num_tokens": 573113719.0, + "step": 14766 + }, + { + "epoch": 1.8785141839460628, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.23678207397461, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.896225094795227, + "num_tokens": 573155994.0, + "step": 14767 + }, + { + "epoch": 1.8786413942246534, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.25953197479248, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.9071574211120605, + "num_tokens": 573192699.0, + "step": 14768 + }, + { + "epoch": 1.878768604503244, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.244810104370117, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8878424167633057, + "num_tokens": 573230550.0, + "step": 14769 + }, + { + "epoch": 1.8788958147818344, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.258670806884766, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8969085216522217, + "num_tokens": 573276221.0, + "step": 14770 + }, + { + "epoch": 1.879023025060425, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.370133399963379, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8906338214874268, + "num_tokens": 573312176.0, + "step": 14771 + }, + { + "epoch": 1.8791502353390155, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.292824745178223, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8842746019363403, + "num_tokens": 573350036.0, + "step": 14772 + }, + { + "epoch": 1.879277445617606, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.26755142211914, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8906070590019226, + "num_tokens": 573389199.0, + "step": 14773 + }, + { + "epoch": 1.8794046558961965, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.216631889343262, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8943781852722168, + "num_tokens": 573437371.0, + "step": 14774 + }, + { + "epoch": 1.879531866174787, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.280071258544922, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8849486112594604, + "num_tokens": 573481004.0, + "step": 14775 + }, + { + "epoch": 1.8796590764533776, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.315184593200684, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8948156833648682, + "num_tokens": 573516709.0, + "step": 14776 + }, + { + "epoch": 1.879786286731968, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.3841552734375, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8888992071151733, + "num_tokens": 573548335.0, + "step": 14777 + }, + { + "epoch": 1.8799134970105584, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.341232299804688, + "learning_rate": 1e-06, + "loss": 0.3729, + "mean_token_accuracy": 0.8738380074501038, + "num_tokens": 573589904.0, + "step": 14778 + }, + { + "epoch": 1.880040707289149, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.276883125305176, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9016678333282471, + "num_tokens": 573628324.0, + "step": 14779 + }, + { + "epoch": 1.8801679175677395, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.329902648925781, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8785278797149658, + "num_tokens": 573668696.0, + "step": 14780 + }, + { + "epoch": 1.88029512784633, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.310722351074219, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8900448083877563, + "num_tokens": 573706792.0, + "step": 14781 + }, + { + "epoch": 1.8804223381249205, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.293800354003906, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8932244777679443, + "num_tokens": 573747027.0, + "step": 14782 + }, + { + "epoch": 1.8805495484035109, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.270179748535156, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8969683647155762, + "num_tokens": 573780981.0, + "step": 14783 + }, + { + "epoch": 1.8806767586821014, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.351837158203125, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.9004943370819092, + "num_tokens": 573823467.0, + "step": 14784 + }, + { + "epoch": 1.880803968960692, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.27276611328125, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8898304104804993, + "num_tokens": 573862116.0, + "step": 14785 + }, + { + "epoch": 1.8809311792392824, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.382440567016602, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.8742638230323792, + "num_tokens": 573900460.0, + "step": 14786 + }, + { + "epoch": 1.881058389517873, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.349989891052246, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8831056356430054, + "num_tokens": 573935743.0, + "step": 14787 + }, + { + "epoch": 1.8811855997964635, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.2396821975708, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8946957588195801, + "num_tokens": 573974525.0, + "step": 14788 + }, + { + "epoch": 1.881312810075054, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.34630012512207, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8866742849349976, + "num_tokens": 574016585.0, + "step": 14789 + }, + { + "epoch": 1.8814400203536445, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.397662162780762, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8719053268432617, + "num_tokens": 574054749.0, + "step": 14790 + }, + { + "epoch": 1.881567230632235, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.268413543701172, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8845886588096619, + "num_tokens": 574096481.0, + "step": 14791 + }, + { + "epoch": 1.8816944409108256, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.340332984924316, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.88181072473526, + "num_tokens": 574134827.0, + "step": 14792 + }, + { + "epoch": 1.8818216511894161, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.24120807647705, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8921493291854858, + "num_tokens": 574180600.0, + "step": 14793 + }, + { + "epoch": 1.8819488614680067, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.252670288085938, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8812671899795532, + "num_tokens": 574221985.0, + "step": 14794 + }, + { + "epoch": 1.8820760717465972, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.42597770690918, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8922450542449951, + "num_tokens": 574251923.0, + "step": 14795 + }, + { + "epoch": 1.8822032820251877, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.31901741027832, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8912606239318848, + "num_tokens": 574291208.0, + "step": 14796 + }, + { + "epoch": 1.8823304923037782, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.389521598815918, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8730233311653137, + "num_tokens": 574331909.0, + "step": 14797 + }, + { + "epoch": 1.8824577025823688, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.180291175842285, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8955186009407043, + "num_tokens": 574382976.0, + "step": 14798 + }, + { + "epoch": 1.8825849128609593, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.290984153747559, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8863068222999573, + "num_tokens": 574425887.0, + "step": 14799 + }, + { + "epoch": 1.8827121231395498, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.443490028381348, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8794293403625488, + "num_tokens": 574462243.0, + "step": 14800 + }, + { + "epoch": 1.8828393334181401, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.261401176452637, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8945479989051819, + "num_tokens": 574503418.0, + "step": 14801 + }, + { + "epoch": 1.8829665436967307, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.30610466003418, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8805859088897705, + "num_tokens": 574543519.0, + "step": 14802 + }, + { + "epoch": 1.8830937539753212, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.383458137512207, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8943199515342712, + "num_tokens": 574578990.0, + "step": 14803 + }, + { + "epoch": 1.8832209642539117, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.365151405334473, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8772222399711609, + "num_tokens": 574616527.0, + "step": 14804 + }, + { + "epoch": 1.8833481745325023, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.382887840270996, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8890190124511719, + "num_tokens": 574652830.0, + "step": 14805 + }, + { + "epoch": 1.8834753848110928, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.281993865966797, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8835747241973877, + "num_tokens": 574698415.0, + "step": 14806 + }, + { + "epoch": 1.883602595089683, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.299704551696777, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8915044069290161, + "num_tokens": 574732964.0, + "step": 14807 + }, + { + "epoch": 1.8837298053682736, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.337505340576172, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8866791725158691, + "num_tokens": 574776426.0, + "step": 14808 + }, + { + "epoch": 1.8838570156468641, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.28927993774414, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8900333642959595, + "num_tokens": 574815670.0, + "step": 14809 + }, + { + "epoch": 1.8839842259254547, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.383955001831055, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8799014091491699, + "num_tokens": 574855505.0, + "step": 14810 + }, + { + "epoch": 1.8841114362040452, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.27120590209961, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8913955092430115, + "num_tokens": 574893440.0, + "step": 14811 + }, + { + "epoch": 1.8842386464826357, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.38565444946289, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8825520277023315, + "num_tokens": 574934076.0, + "step": 14812 + }, + { + "epoch": 1.8843658567612263, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.227255821228027, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.897260308265686, + "num_tokens": 574975615.0, + "step": 14813 + }, + { + "epoch": 1.8844930670398168, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.37160873413086, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8939679265022278, + "num_tokens": 575015784.0, + "step": 14814 + }, + { + "epoch": 1.8846202773184073, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.36767292022705, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8883572220802307, + "num_tokens": 575053776.0, + "step": 14815 + }, + { + "epoch": 1.8847474875969978, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.31790828704834, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8900237083435059, + "num_tokens": 575090895.0, + "step": 14816 + }, + { + "epoch": 1.8848746978755884, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.409981727600098, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8834932446479797, + "num_tokens": 575126621.0, + "step": 14817 + }, + { + "epoch": 1.885001908154179, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.3983154296875, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.8734982013702393, + "num_tokens": 575164649.0, + "step": 14818 + }, + { + "epoch": 1.8851291184327694, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.277679443359375, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8868658542633057, + "num_tokens": 575204140.0, + "step": 14819 + }, + { + "epoch": 1.88525632871136, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.349620819091797, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8846653699874878, + "num_tokens": 575241140.0, + "step": 14820 + }, + { + "epoch": 1.8853835389899505, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.331656455993652, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8843352198600769, + "num_tokens": 575275519.0, + "step": 14821 + }, + { + "epoch": 1.885510749268541, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.262092590332031, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.888918399810791, + "num_tokens": 575317888.0, + "step": 14822 + }, + { + "epoch": 1.8856379595471315, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.324421882629395, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9013491868972778, + "num_tokens": 575350810.0, + "step": 14823 + }, + { + "epoch": 1.885765169825722, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.355225563049316, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8768649101257324, + "num_tokens": 575389533.0, + "step": 14824 + }, + { + "epoch": 1.8858923801043126, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.25968074798584, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8979454040527344, + "num_tokens": 575431197.0, + "step": 14825 + }, + { + "epoch": 1.886019590382903, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.36867618560791, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8841995000839233, + "num_tokens": 575467257.0, + "step": 14826 + }, + { + "epoch": 1.8861468006614934, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.362598419189453, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.8759227395057678, + "num_tokens": 575505396.0, + "step": 14827 + }, + { + "epoch": 1.886274010940084, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.281004905700684, + "learning_rate": 1e-06, + "loss": 0.3576, + "mean_token_accuracy": 0.881014347076416, + "num_tokens": 575548854.0, + "step": 14828 + }, + { + "epoch": 1.8864012212186745, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.27534008026123, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8971853256225586, + "num_tokens": 575591385.0, + "step": 14829 + }, + { + "epoch": 1.886528431497265, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.413783073425293, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8900983333587646, + "num_tokens": 575623045.0, + "step": 14830 + }, + { + "epoch": 1.8866556417758555, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.419276237487793, + "learning_rate": 1e-06, + "loss": 0.4108, + "mean_token_accuracy": 0.8604562282562256, + "num_tokens": 575664584.0, + "step": 14831 + }, + { + "epoch": 1.8867828520544458, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.284817695617676, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.901644229888916, + "num_tokens": 575703252.0, + "step": 14832 + }, + { + "epoch": 1.8869100623330364, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.288481712341309, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8925799131393433, + "num_tokens": 575746379.0, + "step": 14833 + }, + { + "epoch": 1.887037272611627, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.371288299560547, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8921679258346558, + "num_tokens": 575780498.0, + "step": 14834 + }, + { + "epoch": 1.8871644828902174, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.252433776855469, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8858505487442017, + "num_tokens": 575823026.0, + "step": 14835 + }, + { + "epoch": 1.887291693168808, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.403234481811523, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8879021406173706, + "num_tokens": 575858459.0, + "step": 14836 + }, + { + "epoch": 1.8874189034473985, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.278462409973145, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.889107346534729, + "num_tokens": 575895813.0, + "step": 14837 + }, + { + "epoch": 1.887546113725989, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.353449821472168, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8841812610626221, + "num_tokens": 575935800.0, + "step": 14838 + }, + { + "epoch": 1.8876733240045795, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.289376258850098, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8989194631576538, + "num_tokens": 575968606.0, + "step": 14839 + }, + { + "epoch": 1.88780053428317, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.2922945022583, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8850136399269104, + "num_tokens": 576008490.0, + "step": 14840 + }, + { + "epoch": 1.8879277445617606, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.377257347106934, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8894091844558716, + "num_tokens": 576041490.0, + "step": 14841 + }, + { + "epoch": 1.8880549548403511, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.276165008544922, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8844745755195618, + "num_tokens": 576082485.0, + "step": 14842 + }, + { + "epoch": 1.8881821651189417, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.361061096191406, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8848971128463745, + "num_tokens": 576113144.0, + "step": 14843 + }, + { + "epoch": 1.8883093753975322, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.260595321655273, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.9006354808807373, + "num_tokens": 576154051.0, + "step": 14844 + }, + { + "epoch": 1.8884365856761227, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.349491119384766, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8773916959762573, + "num_tokens": 576191151.0, + "step": 14845 + }, + { + "epoch": 1.8885637959547132, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.294508934020996, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8885036706924438, + "num_tokens": 576235489.0, + "step": 14846 + }, + { + "epoch": 1.8886910062333038, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.364327430725098, + "learning_rate": 1e-06, + "loss": 0.3796, + "mean_token_accuracy": 0.8697645664215088, + "num_tokens": 576274329.0, + "step": 14847 + }, + { + "epoch": 1.8888182165118943, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.375326156616211, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8773798942565918, + "num_tokens": 576312770.0, + "step": 14848 + }, + { + "epoch": 1.8889454267904848, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.264567375183105, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8951116800308228, + "num_tokens": 576353897.0, + "step": 14849 + }, + { + "epoch": 1.8890726370690751, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.32744026184082, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8860968947410583, + "num_tokens": 576394991.0, + "step": 14850 + }, + { + "epoch": 1.8891998473476657, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.269389152526855, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8848839402198792, + "num_tokens": 576433761.0, + "step": 14851 + }, + { + "epoch": 1.8893270576262562, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.404180526733398, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8923017382621765, + "num_tokens": 576465512.0, + "step": 14852 + }, + { + "epoch": 1.8894542679048467, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.374045372009277, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8782435655593872, + "num_tokens": 576504982.0, + "step": 14853 + }, + { + "epoch": 1.8895814781834372, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.33997631072998, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8831954598426819, + "num_tokens": 576544825.0, + "step": 14854 + }, + { + "epoch": 1.8897086884620278, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.348098754882812, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8889579176902771, + "num_tokens": 576582497.0, + "step": 14855 + }, + { + "epoch": 1.889835898740618, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.318552017211914, + "learning_rate": 1e-06, + "loss": 0.3783, + "mean_token_accuracy": 0.8709459900856018, + "num_tokens": 576622605.0, + "step": 14856 + }, + { + "epoch": 1.8899631090192086, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.195106506347656, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9093855619430542, + "num_tokens": 576664269.0, + "step": 14857 + }, + { + "epoch": 1.8900903192977991, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.382696151733398, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8846405744552612, + "num_tokens": 576700852.0, + "step": 14858 + }, + { + "epoch": 1.8902175295763897, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.25895881652832, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.883604884147644, + "num_tokens": 576744508.0, + "step": 14859 + }, + { + "epoch": 1.8903447398549802, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.294914245605469, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8952053785324097, + "num_tokens": 576785571.0, + "step": 14860 + }, + { + "epoch": 1.8904719501335707, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.298832893371582, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.883879542350769, + "num_tokens": 576828400.0, + "step": 14861 + }, + { + "epoch": 1.8905991604121613, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.45624828338623, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8799739480018616, + "num_tokens": 576864437.0, + "step": 14862 + }, + { + "epoch": 1.8907263706907518, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.399236679077148, + "learning_rate": 1e-06, + "loss": 0.3988, + "mean_token_accuracy": 0.8634939193725586, + "num_tokens": 576903772.0, + "step": 14863 + }, + { + "epoch": 1.8908535809693423, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.270516395568848, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.89035564661026, + "num_tokens": 576943870.0, + "step": 14864 + }, + { + "epoch": 1.8909807912479328, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.404486656188965, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8780175447463989, + "num_tokens": 576986500.0, + "step": 14865 + }, + { + "epoch": 1.8911080015265234, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.248695373535156, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8997598886489868, + "num_tokens": 577026918.0, + "step": 14866 + }, + { + "epoch": 1.891235211805114, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.416106224060059, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8912912607192993, + "num_tokens": 577059871.0, + "step": 14867 + }, + { + "epoch": 1.8913624220837044, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.314868927001953, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8813539743423462, + "num_tokens": 577100063.0, + "step": 14868 + }, + { + "epoch": 1.891489632362295, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.273991584777832, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9014773964881897, + "num_tokens": 577139178.0, + "step": 14869 + }, + { + "epoch": 1.8916168426408855, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.305481910705566, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8757468461990356, + "num_tokens": 577185272.0, + "step": 14870 + }, + { + "epoch": 1.891744052919476, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.344352722167969, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8898060321807861, + "num_tokens": 577220597.0, + "step": 14871 + }, + { + "epoch": 1.8918712631980665, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.258159637451172, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8965063691139221, + "num_tokens": 577262736.0, + "step": 14872 + }, + { + "epoch": 1.891998473476657, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.306530952453613, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.886794924736023, + "num_tokens": 577296973.0, + "step": 14873 + }, + { + "epoch": 1.8921256837552476, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.387243270874023, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8914262056350708, + "num_tokens": 577331743.0, + "step": 14874 + }, + { + "epoch": 1.892252894033838, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.362613677978516, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8793193101882935, + "num_tokens": 577371493.0, + "step": 14875 + }, + { + "epoch": 1.8923801043124284, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.317687034606934, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8812694549560547, + "num_tokens": 577415682.0, + "step": 14876 + }, + { + "epoch": 1.892507314591019, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.268406867980957, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8901925683021545, + "num_tokens": 577457307.0, + "step": 14877 + }, + { + "epoch": 1.8926345248696095, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.281700134277344, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8996888399124146, + "num_tokens": 577494831.0, + "step": 14878 + }, + { + "epoch": 1.8927617351482, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.296576499938965, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8859288692474365, + "num_tokens": 577534989.0, + "step": 14879 + }, + { + "epoch": 1.8928889454267905, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.344375610351562, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.884010910987854, + "num_tokens": 577574360.0, + "step": 14880 + }, + { + "epoch": 1.8930161557053808, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.330601692199707, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8755512237548828, + "num_tokens": 577613158.0, + "step": 14881 + }, + { + "epoch": 1.8931433659839714, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.249765396118164, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8928521275520325, + "num_tokens": 577655704.0, + "step": 14882 + }, + { + "epoch": 1.893270576262562, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.265480041503906, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8908967971801758, + "num_tokens": 577695100.0, + "step": 14883 + }, + { + "epoch": 1.8933977865411524, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.417591094970703, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8797520995140076, + "num_tokens": 577727615.0, + "step": 14884 + }, + { + "epoch": 1.893524996819743, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.232818603515625, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8792392015457153, + "num_tokens": 577772177.0, + "step": 14885 + }, + { + "epoch": 1.8936522070983335, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.298550605773926, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8860219717025757, + "num_tokens": 577813220.0, + "step": 14886 + }, + { + "epoch": 1.893779417376924, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.287181854248047, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8838749527931213, + "num_tokens": 577857818.0, + "step": 14887 + }, + { + "epoch": 1.8939066276555145, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.340728759765625, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8911995887756348, + "num_tokens": 577892456.0, + "step": 14888 + }, + { + "epoch": 1.894033837934105, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.3950834274292, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8953779935836792, + "num_tokens": 577923054.0, + "step": 14889 + }, + { + "epoch": 1.8941610482126956, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.254401206970215, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8840673565864563, + "num_tokens": 577963094.0, + "step": 14890 + }, + { + "epoch": 1.8942882584912861, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.341111183166504, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.9011363983154297, + "num_tokens": 577999650.0, + "step": 14891 + }, + { + "epoch": 1.8944154687698767, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.276955604553223, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8729870319366455, + "num_tokens": 578045866.0, + "step": 14892 + }, + { + "epoch": 1.8945426790484672, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.285713195800781, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8884707689285278, + "num_tokens": 578086365.0, + "step": 14893 + }, + { + "epoch": 1.8946698893270577, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.346609115600586, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8832381963729858, + "num_tokens": 578129884.0, + "step": 14894 + }, + { + "epoch": 1.8947970996056482, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.364029884338379, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8903887867927551, + "num_tokens": 578169825.0, + "step": 14895 + }, + { + "epoch": 1.8949243098842388, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.417844772338867, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8872075080871582, + "num_tokens": 578204790.0, + "step": 14896 + }, + { + "epoch": 1.8950515201628293, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.352335929870605, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8808865547180176, + "num_tokens": 578242953.0, + "step": 14897 + }, + { + "epoch": 1.8951787304414198, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.380776405334473, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8819161653518677, + "num_tokens": 578277215.0, + "step": 14898 + }, + { + "epoch": 1.8953059407200101, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.356194496154785, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8948282599449158, + "num_tokens": 578316882.0, + "step": 14899 + }, + { + "epoch": 1.8954331509986007, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.303508758544922, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.9032214283943176, + "num_tokens": 578350786.0, + "step": 14900 + }, + { + "epoch": 1.8955603612771912, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.331836700439453, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8877959251403809, + "num_tokens": 578395096.0, + "step": 14901 + }, + { + "epoch": 1.8956875715557817, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.408177375793457, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8919733762741089, + "num_tokens": 578427260.0, + "step": 14902 + }, + { + "epoch": 1.8958147818343722, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.292954444885254, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8932898044586182, + "num_tokens": 578464916.0, + "step": 14903 + }, + { + "epoch": 1.8959419921129628, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.405438423156738, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8816575407981873, + "num_tokens": 578502712.0, + "step": 14904 + }, + { + "epoch": 1.896069202391553, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.35215950012207, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8864683508872986, + "num_tokens": 578542536.0, + "step": 14905 + }, + { + "epoch": 1.8961964126701436, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.40787410736084, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8877421617507935, + "num_tokens": 578577457.0, + "step": 14906 + }, + { + "epoch": 1.8963236229487341, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.335386276245117, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.9022672772407532, + "num_tokens": 578609907.0, + "step": 14907 + }, + { + "epoch": 1.8964508332273247, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.335103034973145, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8840817809104919, + "num_tokens": 578650047.0, + "step": 14908 + }, + { + "epoch": 1.8965780435059152, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.4152193069458, + "learning_rate": 1e-06, + "loss": 0.3954, + "mean_token_accuracy": 0.8685951828956604, + "num_tokens": 578691073.0, + "step": 14909 + }, + { + "epoch": 1.8967052537845057, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.393730163574219, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.9019989967346191, + "num_tokens": 578724003.0, + "step": 14910 + }, + { + "epoch": 1.8968324640630962, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.399950981140137, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8765114545822144, + "num_tokens": 578757993.0, + "step": 14911 + }, + { + "epoch": 1.8969596743416868, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.332698822021484, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8892453908920288, + "num_tokens": 578796751.0, + "step": 14912 + }, + { + "epoch": 1.8970868846202773, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.26618480682373, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8880123496055603, + "num_tokens": 578837971.0, + "step": 14913 + }, + { + "epoch": 1.8972140948988678, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.28338623046875, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8894961476325989, + "num_tokens": 578880236.0, + "step": 14914 + }, + { + "epoch": 1.8973413051774584, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.302592277526855, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8882890939712524, + "num_tokens": 578918754.0, + "step": 14915 + }, + { + "epoch": 1.8974685154560489, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.285883903503418, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8841728568077087, + "num_tokens": 578963657.0, + "step": 14916 + }, + { + "epoch": 1.8975957257346394, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.286510467529297, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9074326753616333, + "num_tokens": 579002250.0, + "step": 14917 + }, + { + "epoch": 1.89772293601323, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.329490661621094, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8797122240066528, + "num_tokens": 579044506.0, + "step": 14918 + }, + { + "epoch": 1.8978501462918205, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.334721565246582, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.9057726263999939, + "num_tokens": 579083287.0, + "step": 14919 + }, + { + "epoch": 1.897977356570411, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.388195037841797, + "learning_rate": 1e-06, + "loss": 0.3475, + "mean_token_accuracy": 0.8823577761650085, + "num_tokens": 579120570.0, + "step": 14920 + }, + { + "epoch": 1.8981045668490015, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.37764835357666, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8817626237869263, + "num_tokens": 579159968.0, + "step": 14921 + }, + { + "epoch": 1.898231777127592, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.306581497192383, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8862310647964478, + "num_tokens": 579197184.0, + "step": 14922 + }, + { + "epoch": 1.8983589874061826, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.358369827270508, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8897255659103394, + "num_tokens": 579231738.0, + "step": 14923 + }, + { + "epoch": 1.898486197684773, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.399517059326172, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8853830099105835, + "num_tokens": 579270852.0, + "step": 14924 + }, + { + "epoch": 1.8986134079633634, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.31601333618164, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8758795261383057, + "num_tokens": 579308823.0, + "step": 14925 + }, + { + "epoch": 1.898740618241954, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.308111190795898, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8822067975997925, + "num_tokens": 579352623.0, + "step": 14926 + }, + { + "epoch": 1.8988678285205445, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.404317855834961, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.8702924251556396, + "num_tokens": 579395217.0, + "step": 14927 + }, + { + "epoch": 1.898995038799135, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.242875099182129, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8772485256195068, + "num_tokens": 579436908.0, + "step": 14928 + }, + { + "epoch": 1.8991222490777255, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.373685836791992, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8928468227386475, + "num_tokens": 579473016.0, + "step": 14929 + }, + { + "epoch": 1.8992494593563158, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.260998725891113, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8880595564842224, + "num_tokens": 579514598.0, + "step": 14930 + }, + { + "epoch": 1.8993766696349064, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.251919746398926, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.889894425868988, + "num_tokens": 579554464.0, + "step": 14931 + }, + { + "epoch": 1.899503879913497, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.334576606750488, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8969829678535461, + "num_tokens": 579601334.0, + "step": 14932 + }, + { + "epoch": 1.8996310901920874, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.35841178894043, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.9003229141235352, + "num_tokens": 579637486.0, + "step": 14933 + }, + { + "epoch": 1.899758300470678, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.359512329101562, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8808783292770386, + "num_tokens": 579672110.0, + "step": 14934 + }, + { + "epoch": 1.8998855107492685, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.323740005493164, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8962942957878113, + "num_tokens": 579707086.0, + "step": 14935 + }, + { + "epoch": 1.900012721027859, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.43467903137207, + "learning_rate": 1e-06, + "loss": 0.3882, + "mean_token_accuracy": 0.8705223798751831, + "num_tokens": 579746786.0, + "step": 14936 + }, + { + "epoch": 1.9001399313064495, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.231893539428711, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8992123603820801, + "num_tokens": 579788378.0, + "step": 14937 + }, + { + "epoch": 1.90026714158504, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.37077808380127, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8932720422744751, + "num_tokens": 579820986.0, + "step": 14938 + }, + { + "epoch": 1.9003943518636306, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.370993614196777, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8937755823135376, + "num_tokens": 579854205.0, + "step": 14939 + }, + { + "epoch": 1.9005215621422211, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.398985862731934, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8915660381317139, + "num_tokens": 579891802.0, + "step": 14940 + }, + { + "epoch": 1.9006487724208116, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.351334571838379, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.881818413734436, + "num_tokens": 579929229.0, + "step": 14941 + }, + { + "epoch": 1.9007759826994022, + "ewc_loss": 0.01708984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.283095359802246, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8867191076278687, + "num_tokens": 579969770.0, + "step": 14942 + }, + { + "epoch": 1.9009031929779927, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.387213706970215, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8849177956581116, + "num_tokens": 580008770.0, + "step": 14943 + }, + { + "epoch": 1.9010304032565832, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.42790699005127, + "learning_rate": 1e-06, + "loss": 0.402, + "mean_token_accuracy": 0.8644848465919495, + "num_tokens": 580050389.0, + "step": 14944 + }, + { + "epoch": 1.9011576135351738, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.340737342834473, + "learning_rate": 1e-06, + "loss": 0.3647, + "mean_token_accuracy": 0.8753253221511841, + "num_tokens": 580087124.0, + "step": 14945 + }, + { + "epoch": 1.9012848238137643, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.332796096801758, + "learning_rate": 1e-06, + "loss": 0.3901, + "mean_token_accuracy": 0.8675960302352905, + "num_tokens": 580127711.0, + "step": 14946 + }, + { + "epoch": 1.9014120340923548, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.330473899841309, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9013913869857788, + "num_tokens": 580165644.0, + "step": 14947 + }, + { + "epoch": 1.9015392443709451, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.375351905822754, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8927761316299438, + "num_tokens": 580204658.0, + "step": 14948 + }, + { + "epoch": 1.9016664546495357, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.376917839050293, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8845663070678711, + "num_tokens": 580240953.0, + "step": 14949 + }, + { + "epoch": 1.9017936649281262, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.384244918823242, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8942573070526123, + "num_tokens": 580275756.0, + "step": 14950 + }, + { + "epoch": 1.9019208752067167, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.278797149658203, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8894254565238953, + "num_tokens": 580318333.0, + "step": 14951 + }, + { + "epoch": 1.9020480854853072, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.43830680847168, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8844301700592041, + "num_tokens": 580360014.0, + "step": 14952 + }, + { + "epoch": 1.9021752957638978, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.285845756530762, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.888725996017456, + "num_tokens": 580399194.0, + "step": 14953 + }, + { + "epoch": 1.902302506042488, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.320741653442383, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9010329246520996, + "num_tokens": 580433125.0, + "step": 14954 + }, + { + "epoch": 1.9024297163210786, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.284565925598145, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8960613012313843, + "num_tokens": 580472697.0, + "step": 14955 + }, + { + "epoch": 1.9025569265996691, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.255707740783691, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9047119617462158, + "num_tokens": 580517784.0, + "step": 14956 + }, + { + "epoch": 1.9026841368782597, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.3485746383667, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8825536966323853, + "num_tokens": 580560447.0, + "step": 14957 + }, + { + "epoch": 1.9028113471568502, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.250584602355957, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8955367207527161, + "num_tokens": 580600686.0, + "step": 14958 + }, + { + "epoch": 1.9029385574354407, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.357730865478516, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.897506594657898, + "num_tokens": 580639661.0, + "step": 14959 + }, + { + "epoch": 1.9030657677140312, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.354310989379883, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8920348882675171, + "num_tokens": 580677338.0, + "step": 14960 + }, + { + "epoch": 1.9031929779926218, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.405719757080078, + "learning_rate": 1e-06, + "loss": 0.381, + "mean_token_accuracy": 0.8718132376670837, + "num_tokens": 580711382.0, + "step": 14961 + }, + { + "epoch": 1.9033201882712123, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.333442687988281, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8912357091903687, + "num_tokens": 580755178.0, + "step": 14962 + }, + { + "epoch": 1.9034473985498028, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.35741901397705, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8797450065612793, + "num_tokens": 580792832.0, + "step": 14963 + }, + { + "epoch": 1.9035746088283934, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.490944862365723, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.886103630065918, + "num_tokens": 580829653.0, + "step": 14964 + }, + { + "epoch": 1.9037018191069839, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.340999603271484, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8822214007377625, + "num_tokens": 580866046.0, + "step": 14965 + }, + { + "epoch": 1.9038290293855744, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.334785461425781, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.9017512798309326, + "num_tokens": 580902698.0, + "step": 14966 + }, + { + "epoch": 1.903956239664165, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.276528358459473, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8837826251983643, + "num_tokens": 580950058.0, + "step": 14967 + }, + { + "epoch": 1.9040834499427555, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.481119155883789, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8851966261863708, + "num_tokens": 580982446.0, + "step": 14968 + }, + { + "epoch": 1.904210660221346, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.341934204101562e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.335845947265625, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8759622573852539, + "num_tokens": 581027766.0, + "step": 14969 + }, + { + "epoch": 1.9043378704999365, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.389617919921875e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.494832992553711, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8965620994567871, + "num_tokens": 581073769.0, + "step": 14970 + }, + { + "epoch": 1.904465080778527, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.389617919921875e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.355938911437988, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.891754686832428, + "num_tokens": 581110853.0, + "step": 14971 + }, + { + "epoch": 1.9045922910571176, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.389617919921875e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.404468536376953, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8835358619689941, + "num_tokens": 581147786.0, + "step": 14972 + }, + { + "epoch": 1.9047195013357079, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.389617919921875e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.43264389038086, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8829450011253357, + "num_tokens": 581184974.0, + "step": 14973 + }, + { + "epoch": 1.9048467116142984, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.389617919921875e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.278573989868164, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.9004745483398438, + "num_tokens": 581225109.0, + "step": 14974 + }, + { + "epoch": 1.904973921892889, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.389617919921875e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.394676208496094, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8918972611427307, + "num_tokens": 581260477.0, + "step": 14975 + }, + { + "epoch": 1.9051011321714795, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.389617919921875e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.29262638092041, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8872429132461548, + "num_tokens": 581305400.0, + "step": 14976 + }, + { + "epoch": 1.90522834245007, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.389617919921875e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.417024612426758, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8861991167068481, + "num_tokens": 581344160.0, + "step": 14977 + }, + { + "epoch": 1.9053555527286605, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.389617919921875e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.354888916015625, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8954290747642517, + "num_tokens": 581383779.0, + "step": 14978 + }, + { + "epoch": 1.9054827630072508, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.389617919921875e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.353626251220703, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8932236433029175, + "num_tokens": 581420612.0, + "step": 14979 + }, + { + "epoch": 1.9056099732858414, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.389617919921875e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.350981712341309, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.9032814502716064, + "num_tokens": 581460634.0, + "step": 14980 + }, + { + "epoch": 1.905737183564432, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.389617919921875e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.293567657470703, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.896742582321167, + "num_tokens": 581499706.0, + "step": 14981 + }, + { + "epoch": 1.9058643938430224, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.363675117492676, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8852100968360901, + "num_tokens": 581541400.0, + "step": 14982 + }, + { + "epoch": 1.905991604121613, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.466225624084473, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8717153668403625, + "num_tokens": 581579736.0, + "step": 14983 + }, + { + "epoch": 1.9061188144002035, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.541935920715332, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8772307634353638, + "num_tokens": 581615269.0, + "step": 14984 + }, + { + "epoch": 1.906246024678794, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.398555755615234, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8808225989341736, + "num_tokens": 581656243.0, + "step": 14985 + }, + { + "epoch": 1.9063732349573845, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.28132438659668, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8875526785850525, + "num_tokens": 581697103.0, + "step": 14986 + }, + { + "epoch": 1.906500445235975, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.335726737976074, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8890875577926636, + "num_tokens": 581740194.0, + "step": 14987 + }, + { + "epoch": 1.9066276555145656, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.367561340332031, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8853607177734375, + "num_tokens": 581777625.0, + "step": 14988 + }, + { + "epoch": 1.9067548657931561, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.479969024658203, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8772504329681396, + "num_tokens": 581813406.0, + "step": 14989 + }, + { + "epoch": 1.9068820760717466, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.368905067443848, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8928231000900269, + "num_tokens": 581849059.0, + "step": 14990 + }, + { + "epoch": 1.9070092863503372, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.309157371520996, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8953943252563477, + "num_tokens": 581880812.0, + "step": 14991 + }, + { + "epoch": 1.9071364966289277, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.310260772705078, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8783731460571289, + "num_tokens": 581923701.0, + "step": 14992 + }, + { + "epoch": 1.9072637069075182, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.396282196044922, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8853734731674194, + "num_tokens": 581962102.0, + "step": 14993 + }, + { + "epoch": 1.9073909171861088, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.312356948852539, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8882102966308594, + "num_tokens": 582002121.0, + "step": 14994 + }, + { + "epoch": 1.9075181274646993, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.495545387268066, + "learning_rate": 1e-06, + "loss": 0.3985, + "mean_token_accuracy": 0.8671301603317261, + "num_tokens": 582035105.0, + "step": 14995 + }, + { + "epoch": 1.9076453377432898, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.4127779006958, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8824464678764343, + "num_tokens": 582072096.0, + "step": 14996 + }, + { + "epoch": 1.9077725480218801, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.378050804138184, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8948361873626709, + "num_tokens": 582107778.0, + "step": 14997 + }, + { + "epoch": 1.9078997583004706, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.40201187133789, + "learning_rate": 1e-06, + "loss": 0.4094, + "mean_token_accuracy": 0.8619633316993713, + "num_tokens": 582146112.0, + "step": 14998 + }, + { + "epoch": 1.9080269685790612, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.280965805053711, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8949037790298462, + "num_tokens": 582187987.0, + "step": 14999 + }, + { + "epoch": 1.9081541788576517, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.365973472595215, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8978796601295471, + "num_tokens": 582228511.0, + "step": 15000 + }, + { + "epoch": 1.9082813891362422, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.32725715637207, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9103505611419678, + "num_tokens": 582265785.0, + "step": 15001 + }, + { + "epoch": 1.9084085994148328, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.370490074157715, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8732931613922119, + "num_tokens": 582306470.0, + "step": 15002 + }, + { + "epoch": 1.908535809693423, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.381945610046387, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8871437907218933, + "num_tokens": 582345077.0, + "step": 15003 + }, + { + "epoch": 1.9086630199720136, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.225229263305664, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8900610208511353, + "num_tokens": 582387032.0, + "step": 15004 + }, + { + "epoch": 1.9087902302506041, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.47097110748291, + "learning_rate": 1e-06, + "loss": 0.3597, + "mean_token_accuracy": 0.8792647123336792, + "num_tokens": 582422164.0, + "step": 15005 + }, + { + "epoch": 1.9089174405291947, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.379493713378906, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8865495920181274, + "num_tokens": 582457955.0, + "step": 15006 + }, + { + "epoch": 1.9090446508077852, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.339491844177246, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8952928781509399, + "num_tokens": 582495200.0, + "step": 15007 + }, + { + "epoch": 1.9091718610863757, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.388357162475586, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8942807912826538, + "num_tokens": 582529217.0, + "step": 15008 + }, + { + "epoch": 1.9092990713649662, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.347024917602539, + "learning_rate": 1e-06, + "loss": 0.3676, + "mean_token_accuracy": 0.8750720620155334, + "num_tokens": 582570697.0, + "step": 15009 + }, + { + "epoch": 1.9094262816435568, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.425711631774902, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.8784339427947998, + "num_tokens": 582613995.0, + "step": 15010 + }, + { + "epoch": 1.9095534919221473, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.38730525970459, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8901036977767944, + "num_tokens": 582656032.0, + "step": 15011 + }, + { + "epoch": 1.9096807022007378, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.360671043395996, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8925363421440125, + "num_tokens": 582696314.0, + "step": 15012 + }, + { + "epoch": 1.9098079124793284, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.382018089294434, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.871737003326416, + "num_tokens": 582739986.0, + "step": 15013 + }, + { + "epoch": 1.9099351227579189, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.45167350769043, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8887521028518677, + "num_tokens": 582774351.0, + "step": 15014 + }, + { + "epoch": 1.9100623330365094, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.384469032287598, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.887024998664856, + "num_tokens": 582811106.0, + "step": 15015 + }, + { + "epoch": 1.9101895433151, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.307182312011719, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8819442987442017, + "num_tokens": 582852376.0, + "step": 15016 + }, + { + "epoch": 1.9103167535936905, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.413684844970703, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8877415657043457, + "num_tokens": 582890748.0, + "step": 15017 + }, + { + "epoch": 1.910443963872281, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.267822265625, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8827712535858154, + "num_tokens": 582938518.0, + "step": 15018 + }, + { + "epoch": 1.9105711741508715, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.329389572143555, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8966857194900513, + "num_tokens": 582978244.0, + "step": 15019 + }, + { + "epoch": 1.910698384429462, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.435202598571777, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8850513696670532, + "num_tokens": 583015896.0, + "step": 15020 + }, + { + "epoch": 1.9108255947080524, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.415755271911621, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8904852867126465, + "num_tokens": 583051770.0, + "step": 15021 + }, + { + "epoch": 1.9109528049866429, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.403508186340332, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8893625140190125, + "num_tokens": 583084744.0, + "step": 15022 + }, + { + "epoch": 1.9110800152652334, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.412360191345215, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8914703130722046, + "num_tokens": 583115877.0, + "step": 15023 + }, + { + "epoch": 1.911207225543824, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.388288497924805, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8957381248474121, + "num_tokens": 583147817.0, + "step": 15024 + }, + { + "epoch": 1.9113344358224145, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.31422233581543, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8884209394454956, + "num_tokens": 583185136.0, + "step": 15025 + }, + { + "epoch": 1.911461646101005, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.31395149230957, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8840782642364502, + "num_tokens": 583224941.0, + "step": 15026 + }, + { + "epoch": 1.9115888563795955, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.35099983215332, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8935552835464478, + "num_tokens": 583263668.0, + "step": 15027 + }, + { + "epoch": 1.9117160666581858, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.4422025680542, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8921449184417725, + "num_tokens": 583295167.0, + "step": 15028 + }, + { + "epoch": 1.9118432769367764, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.363171577453613, + "learning_rate": 1e-06, + "loss": 0.3689, + "mean_token_accuracy": 0.8769924640655518, + "num_tokens": 583338536.0, + "step": 15029 + }, + { + "epoch": 1.9119704872153669, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.306984901428223, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8880413770675659, + "num_tokens": 583378299.0, + "step": 15030 + }, + { + "epoch": 1.9120976974939574, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.340831756591797, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.876069188117981, + "num_tokens": 583422482.0, + "step": 15031 + }, + { + "epoch": 1.912224907772548, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.263422966003418, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8965893983840942, + "num_tokens": 583466460.0, + "step": 15032 + }, + { + "epoch": 1.9123521180511385, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.337381362915039, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8820092082023621, + "num_tokens": 583509943.0, + "step": 15033 + }, + { + "epoch": 1.912479328329729, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.394295692443848, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8816123008728027, + "num_tokens": 583549201.0, + "step": 15034 + }, + { + "epoch": 1.9126065386083195, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.349447250366211, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8790019750595093, + "num_tokens": 583592049.0, + "step": 15035 + }, + { + "epoch": 1.91273374888691, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.325861930847168, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8900092244148254, + "num_tokens": 583631744.0, + "step": 15036 + }, + { + "epoch": 1.9128609591655006, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.308032989501953, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.9006456136703491, + "num_tokens": 583673468.0, + "step": 15037 + }, + { + "epoch": 1.9129881694440911, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.324285507202148, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8949394226074219, + "num_tokens": 583710649.0, + "step": 15038 + }, + { + "epoch": 1.9131153797226816, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.370218276977539, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8832557797431946, + "num_tokens": 583751987.0, + "step": 15039 + }, + { + "epoch": 1.9132425900012722, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.391363143920898, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8927657604217529, + "num_tokens": 583789335.0, + "step": 15040 + }, + { + "epoch": 1.9133698002798627, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.334488868713379, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8919095396995544, + "num_tokens": 583830368.0, + "step": 15041 + }, + { + "epoch": 1.9134970105584532, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.343320846557617, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8825777769088745, + "num_tokens": 583872508.0, + "step": 15042 + }, + { + "epoch": 1.9136242208370438, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.375421524047852, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.88193678855896, + "num_tokens": 583912859.0, + "step": 15043 + }, + { + "epoch": 1.9137514311156343, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.337936401367188, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8890307545661926, + "num_tokens": 583952977.0, + "step": 15044 + }, + { + "epoch": 1.9138786413942248, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.369490623474121, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8974679708480835, + "num_tokens": 583987040.0, + "step": 15045 + }, + { + "epoch": 1.9140058516728151, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.353750228881836, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8953827023506165, + "num_tokens": 584024593.0, + "step": 15046 + }, + { + "epoch": 1.9141330619514056, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.437301635742188e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.291647911071777, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.9029362797737122, + "num_tokens": 584064853.0, + "step": 15047 + }, + { + "epoch": 1.9142602722299962, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.408971786499023, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8837244510650635, + "num_tokens": 584102898.0, + "step": 15048 + }, + { + "epoch": 1.9143874825085867, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.384471893310547, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8907008171081543, + "num_tokens": 584141859.0, + "step": 15049 + }, + { + "epoch": 1.9145146927871772, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.322931289672852, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8879171013832092, + "num_tokens": 584185828.0, + "step": 15050 + }, + { + "epoch": 1.9146419030657678, + "ewc_loss": 0.0172119140625, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0728836059570312e-05, + "grad_norm": 9.376045227050781, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8952515125274658, + "num_tokens": 584221941.0, + "step": 15051 + }, + { + "epoch": 1.914769113344358, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.402505874633789, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8952527046203613, + "num_tokens": 584257205.0, + "step": 15052 + }, + { + "epoch": 1.9148963236229486, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.359865188598633, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.9010106325149536, + "num_tokens": 584296133.0, + "step": 15053 + }, + { + "epoch": 1.9150235339015391, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.384703636169434, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8996022939682007, + "num_tokens": 584328424.0, + "step": 15054 + }, + { + "epoch": 1.9151507441801296, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.328314781188965, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.9004116058349609, + "num_tokens": 584367621.0, + "step": 15055 + }, + { + "epoch": 1.9152779544587202, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.357120513916016, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8991886377334595, + "num_tokens": 584407044.0, + "step": 15056 + }, + { + "epoch": 1.9154051647373107, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.411202430725098, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.879895031452179, + "num_tokens": 584441992.0, + "step": 15057 + }, + { + "epoch": 1.9155323750159012, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.409404754638672, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8921760320663452, + "num_tokens": 584476026.0, + "step": 15058 + }, + { + "epoch": 1.9156595852944918, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.39661979675293, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8942785263061523, + "num_tokens": 584509269.0, + "step": 15059 + }, + { + "epoch": 1.9157867955730823, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.260562896728516, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8912183046340942, + "num_tokens": 584549325.0, + "step": 15060 + }, + { + "epoch": 1.9159140058516728, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.39212417602539, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8874120712280273, + "num_tokens": 584589176.0, + "step": 15061 + }, + { + "epoch": 1.9160412161302633, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.242435455322266, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.9010533094406128, + "num_tokens": 584627869.0, + "step": 15062 + }, + { + "epoch": 1.9161684264088539, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.480262756347656, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8750898241996765, + "num_tokens": 584666872.0, + "step": 15063 + }, + { + "epoch": 1.9162956366874444, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.309799194335938, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8862171173095703, + "num_tokens": 584706356.0, + "step": 15064 + }, + { + "epoch": 1.916422846966035, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.455567359924316, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8713694214820862, + "num_tokens": 584746031.0, + "step": 15065 + }, + { + "epoch": 1.9165500572446255, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.484817504882812, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8845794200897217, + "num_tokens": 584778292.0, + "step": 15066 + }, + { + "epoch": 1.916677267523216, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.499556541442871, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8803085088729858, + "num_tokens": 584809740.0, + "step": 15067 + }, + { + "epoch": 1.9168044778018065, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.331523895263672, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.878624439239502, + "num_tokens": 584850675.0, + "step": 15068 + }, + { + "epoch": 1.916931688080397, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.375691413879395, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8964478969573975, + "num_tokens": 584888037.0, + "step": 15069 + }, + { + "epoch": 1.9170588983589874, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.366653442382812, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.889012336730957, + "num_tokens": 584924749.0, + "step": 15070 + }, + { + "epoch": 1.9171861086375779, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.350553512573242, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8886286020278931, + "num_tokens": 584964913.0, + "step": 15071 + }, + { + "epoch": 1.9173133189161684, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.350598335266113, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8933019638061523, + "num_tokens": 585004686.0, + "step": 15072 + }, + { + "epoch": 1.917440529194759, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.331149101257324, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8957431316375732, + "num_tokens": 585044760.0, + "step": 15073 + }, + { + "epoch": 1.9175677394733495, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.366922378540039, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8821660280227661, + "num_tokens": 585079950.0, + "step": 15074 + }, + { + "epoch": 1.91769494975194, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.358658790588379, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8992148637771606, + "num_tokens": 585121646.0, + "step": 15075 + }, + { + "epoch": 1.9178221600305305, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.345335960388184, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8974200487136841, + "num_tokens": 585160447.0, + "step": 15076 + }, + { + "epoch": 1.9179493703091208, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.35152816772461, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8843189477920532, + "num_tokens": 585199995.0, + "step": 15077 + }, + { + "epoch": 1.9180765805877114, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.304529190063477, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8883370757102966, + "num_tokens": 585240931.0, + "step": 15078 + }, + { + "epoch": 1.9182037908663019, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.426149368286133, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.874274730682373, + "num_tokens": 585280358.0, + "step": 15079 + }, + { + "epoch": 1.9183310011448924, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.381521224975586, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8799593448638916, + "num_tokens": 585317015.0, + "step": 15080 + }, + { + "epoch": 1.918458211423483, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.33025074005127, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8847127556800842, + "num_tokens": 585355794.0, + "step": 15081 + }, + { + "epoch": 1.9185854217020735, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.282330513000488, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8891903162002563, + "num_tokens": 585398371.0, + "step": 15082 + }, + { + "epoch": 1.918712631980664, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.409505844116211, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8831409215927124, + "num_tokens": 585439780.0, + "step": 15083 + }, + { + "epoch": 1.9188398422592545, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.377790451049805, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8951560854911804, + "num_tokens": 585473529.0, + "step": 15084 + }, + { + "epoch": 1.918967052537845, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.289702415466309, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.884219229221344, + "num_tokens": 585517553.0, + "step": 15085 + }, + { + "epoch": 1.9190942628164356, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.344095230102539, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8868274688720703, + "num_tokens": 585554944.0, + "step": 15086 + }, + { + "epoch": 1.919221473095026, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.287297248840332, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8922498226165771, + "num_tokens": 585596658.0, + "step": 15087 + }, + { + "epoch": 1.9193486833736166, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.371170997619629, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8817505836486816, + "num_tokens": 585634523.0, + "step": 15088 + }, + { + "epoch": 1.9194758936522072, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.410835266113281, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8828853368759155, + "num_tokens": 585669106.0, + "step": 15089 + }, + { + "epoch": 1.9196031039307977, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.285914421081543, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8913695216178894, + "num_tokens": 585704043.0, + "step": 15090 + }, + { + "epoch": 1.9197303142093882, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.439719200134277, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8746649026870728, + "num_tokens": 585737814.0, + "step": 15091 + }, + { + "epoch": 1.9198575244879788, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.350181579589844, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8938076496124268, + "num_tokens": 585773048.0, + "step": 15092 + }, + { + "epoch": 1.9199847347665693, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.288784980773926, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8782708644866943, + "num_tokens": 585820639.0, + "step": 15093 + }, + { + "epoch": 1.9201119450451598, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.384182929992676, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8911080956459045, + "num_tokens": 585860305.0, + "step": 15094 + }, + { + "epoch": 1.9202391553237501, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.368314743041992, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8872774839401245, + "num_tokens": 585900873.0, + "step": 15095 + }, + { + "epoch": 1.9203663656023406, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.35892391204834, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.887694239616394, + "num_tokens": 585933496.0, + "step": 15096 + }, + { + "epoch": 1.9204935758809312, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.351466178894043, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8929859399795532, + "num_tokens": 585970321.0, + "step": 15097 + }, + { + "epoch": 1.9206207861595217, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.361894607543945, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8805732131004333, + "num_tokens": 586008551.0, + "step": 15098 + }, + { + "epoch": 1.9207479964381122, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.355714797973633, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.895035982131958, + "num_tokens": 586048433.0, + "step": 15099 + }, + { + "epoch": 1.9208752067167028, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.361366271972656, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8903279304504395, + "num_tokens": 586087744.0, + "step": 15100 + }, + { + "epoch": 1.921002416995293, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.462607383728027, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.878200888633728, + "num_tokens": 586128928.0, + "step": 15101 + }, + { + "epoch": 1.9211296272738836, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.405143737792969, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8879244923591614, + "num_tokens": 586168546.0, + "step": 15102 + }, + { + "epoch": 1.9212568375524741, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.224528312683105, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.912848949432373, + "num_tokens": 586207670.0, + "step": 15103 + }, + { + "epoch": 1.9213840478310646, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.362787246704102, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8890098333358765, + "num_tokens": 586245770.0, + "step": 15104 + }, + { + "epoch": 1.9215112581096552, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.375798225402832, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8832052946090698, + "num_tokens": 586288569.0, + "step": 15105 + }, + { + "epoch": 1.9216384683882457, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.303905487060547, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8831089735031128, + "num_tokens": 586335458.0, + "step": 15106 + }, + { + "epoch": 1.9217656786668362, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.433422088623047, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8802075982093811, + "num_tokens": 586370674.0, + "step": 15107 + }, + { + "epoch": 1.9218928889454268, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.445211410522461, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.883825421333313, + "num_tokens": 586405109.0, + "step": 15108 + }, + { + "epoch": 1.9220200992240173, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.341106414794922, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.893042802810669, + "num_tokens": 586446237.0, + "step": 15109 + }, + { + "epoch": 1.9221473095026078, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.364392280578613, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8803253173828125, + "num_tokens": 586481337.0, + "step": 15110 + }, + { + "epoch": 1.9222745197811983, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.336592674255371, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8855320811271667, + "num_tokens": 586528068.0, + "step": 15111 + }, + { + "epoch": 1.9224017300597889, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.310383796691895, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.9012607336044312, + "num_tokens": 586563622.0, + "step": 15112 + }, + { + "epoch": 1.9225289403383794, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.4849853515625e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.449114799499512, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8820734024047852, + "num_tokens": 586594696.0, + "step": 15113 + }, + { + "epoch": 1.92265615061697, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.4080171585083, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8874449133872986, + "num_tokens": 586628638.0, + "step": 15114 + }, + { + "epoch": 1.9227833608955605, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.321802139282227, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8888159990310669, + "num_tokens": 586666187.0, + "step": 15115 + }, + { + "epoch": 1.922910571174151, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.3462553024292, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8913083076477051, + "num_tokens": 586704907.0, + "step": 15116 + }, + { + "epoch": 1.9230377814527415, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.303505897521973, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8815447688102722, + "num_tokens": 586745550.0, + "step": 15117 + }, + { + "epoch": 1.923164991731332, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.430512428283691, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8764736652374268, + "num_tokens": 586784446.0, + "step": 15118 + }, + { + "epoch": 1.9232922020099223, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.352335929870605, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8954989910125732, + "num_tokens": 586820012.0, + "step": 15119 + }, + { + "epoch": 1.9234194122885129, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.33935546875, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8832219839096069, + "num_tokens": 586859025.0, + "step": 15120 + }, + { + "epoch": 1.9235466225671034, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.348651885986328, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8970165252685547, + "num_tokens": 586891025.0, + "step": 15121 + }, + { + "epoch": 1.923673832845694, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.450529098510742, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8776227235794067, + "num_tokens": 586928760.0, + "step": 15122 + }, + { + "epoch": 1.9238010431242845, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.36177921295166, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8893675208091736, + "num_tokens": 586966419.0, + "step": 15123 + }, + { + "epoch": 1.923928253402875, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.272122383117676, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8915454149246216, + "num_tokens": 587009362.0, + "step": 15124 + }, + { + "epoch": 1.9240554636814655, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.4772310256958, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8811382055282593, + "num_tokens": 587043169.0, + "step": 15125 + }, + { + "epoch": 1.9241826739600558, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.338894844055176, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8747672438621521, + "num_tokens": 587084554.0, + "step": 15126 + }, + { + "epoch": 1.9243098842386464, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.21255874633789, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.9020322561264038, + "num_tokens": 587132145.0, + "step": 15127 + }, + { + "epoch": 1.9244370945172369, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.498333930969238, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8735907077789307, + "num_tokens": 587168752.0, + "step": 15128 + }, + { + "epoch": 1.9245643047958274, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.31503677368164, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8934003114700317, + "num_tokens": 587208030.0, + "step": 15129 + }, + { + "epoch": 1.924691515074418, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.431888580322266, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8750013113021851, + "num_tokens": 587239807.0, + "step": 15130 + }, + { + "epoch": 1.9248187253530085, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.367677688598633, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9005942344665527, + "num_tokens": 587276662.0, + "step": 15131 + }, + { + "epoch": 1.924945935631599, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.267434120178223, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8884403705596924, + "num_tokens": 587321342.0, + "step": 15132 + }, + { + "epoch": 1.9250731459101895, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.327958106994629, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8927953839302063, + "num_tokens": 587359815.0, + "step": 15133 + }, + { + "epoch": 1.92520035618878, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.34752082824707, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8827704191207886, + "num_tokens": 587399133.0, + "step": 15134 + }, + { + "epoch": 1.9253275664673706, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.375358581542969, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8947290778160095, + "num_tokens": 587434278.0, + "step": 15135 + }, + { + "epoch": 1.925454776745961, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.38357162475586, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8836100101470947, + "num_tokens": 587475935.0, + "step": 15136 + }, + { + "epoch": 1.9255819870245516, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.277396202087402, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8918839693069458, + "num_tokens": 587516041.0, + "step": 15137 + }, + { + "epoch": 1.9257091973031422, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.461880683898926, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.890047550201416, + "num_tokens": 587548821.0, + "step": 15138 + }, + { + "epoch": 1.9258364075817327, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.388287544250488, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8920210599899292, + "num_tokens": 587583335.0, + "step": 15139 + }, + { + "epoch": 1.9259636178603232, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.431357383728027, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.882687509059906, + "num_tokens": 587620160.0, + "step": 15140 + }, + { + "epoch": 1.9260908281389137, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.429790496826172, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.877349317073822, + "num_tokens": 587657304.0, + "step": 15141 + }, + { + "epoch": 1.9262180384175043, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.362936973571777, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8925355076789856, + "num_tokens": 587696360.0, + "step": 15142 + }, + { + "epoch": 1.9263452486960948, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.389229774475098, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8874194622039795, + "num_tokens": 587734996.0, + "step": 15143 + }, + { + "epoch": 1.926472458974685, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.252634048461914, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8982692360877991, + "num_tokens": 587779701.0, + "step": 15144 + }, + { + "epoch": 1.9265996692532756, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.299103736877441, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8816760182380676, + "num_tokens": 587824725.0, + "step": 15145 + }, + { + "epoch": 1.9267268795318662, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.388620376586914, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8953049182891846, + "num_tokens": 587863547.0, + "step": 15146 + }, + { + "epoch": 1.9268540898104567, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.346776962280273, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8870319724082947, + "num_tokens": 587908060.0, + "step": 15147 + }, + { + "epoch": 1.9269813000890472, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.41061019897461, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8894906044006348, + "num_tokens": 587942785.0, + "step": 15148 + }, + { + "epoch": 1.9271085103676378, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.407419204711914, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8923854827880859, + "num_tokens": 587978334.0, + "step": 15149 + }, + { + "epoch": 1.927235720646228, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.368042945861816, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8826441764831543, + "num_tokens": 588019706.0, + "step": 15150 + }, + { + "epoch": 1.9273629309248186, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.354581832885742, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8861000537872314, + "num_tokens": 588062343.0, + "step": 15151 + }, + { + "epoch": 1.9274901412034091, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.234729766845703, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9019067287445068, + "num_tokens": 588104121.0, + "step": 15152 + }, + { + "epoch": 1.9276173514819996, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.37819766998291, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8863842487335205, + "num_tokens": 588143707.0, + "step": 15153 + }, + { + "epoch": 1.9277445617605902, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.354333877563477, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8909344673156738, + "num_tokens": 588187546.0, + "step": 15154 + }, + { + "epoch": 1.9278717720391807, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.347295761108398, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8965882062911987, + "num_tokens": 588223303.0, + "step": 15155 + }, + { + "epoch": 1.9279989823177712, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.405571937561035, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8880449533462524, + "num_tokens": 588256617.0, + "step": 15156 + }, + { + "epoch": 1.9281261925963618, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.47403621673584, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8788145780563354, + "num_tokens": 588290547.0, + "step": 15157 + }, + { + "epoch": 1.9282534028749523, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.42393684387207, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8827322125434875, + "num_tokens": 588326320.0, + "step": 15158 + }, + { + "epoch": 1.9283806131535428, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.400294303894043, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8917807340621948, + "num_tokens": 588362368.0, + "step": 15159 + }, + { + "epoch": 1.9285078234321333, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.30946159362793, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8866933584213257, + "num_tokens": 588404013.0, + "step": 15160 + }, + { + "epoch": 1.9286350337107239, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.252676963806152, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8941343426704407, + "num_tokens": 588445646.0, + "step": 15161 + }, + { + "epoch": 1.9287622439893144, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.40415096282959, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8791964054107666, + "num_tokens": 588487401.0, + "step": 15162 + }, + { + "epoch": 1.928889454267905, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.382946014404297, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8793002367019653, + "num_tokens": 588523409.0, + "step": 15163 + }, + { + "epoch": 1.9290166645464955, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.334593772888184, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8831859827041626, + "num_tokens": 588560891.0, + "step": 15164 + }, + { + "epoch": 1.929143874825086, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.399785041809082, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8867740631103516, + "num_tokens": 588598929.0, + "step": 15165 + }, + { + "epoch": 1.9292710851036765, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.317391395568848, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8766449689865112, + "num_tokens": 588638967.0, + "step": 15166 + }, + { + "epoch": 1.929398295382267, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.320067405700684, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8825526833534241, + "num_tokens": 588680415.0, + "step": 15167 + }, + { + "epoch": 1.9295255056608573, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.388910293579102, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.8685076236724854, + "num_tokens": 588717494.0, + "step": 15168 + }, + { + "epoch": 1.9296527159394479, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.320313453674316, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9063191413879395, + "num_tokens": 588754552.0, + "step": 15169 + }, + { + "epoch": 1.9297799262180384, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.441027641296387, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8861948251724243, + "num_tokens": 588790140.0, + "step": 15170 + }, + { + "epoch": 1.929907136496629, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.368179321289062, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8851187229156494, + "num_tokens": 588829320.0, + "step": 15171 + }, + { + "epoch": 1.9300343467752195, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.200369834899902, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8981505632400513, + "num_tokens": 588872380.0, + "step": 15172 + }, + { + "epoch": 1.93016155705381, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.439896583557129, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8907706141471863, + "num_tokens": 588906964.0, + "step": 15173 + }, + { + "epoch": 1.9302887673324005, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.45805549621582, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.8703114986419678, + "num_tokens": 588941542.0, + "step": 15174 + }, + { + "epoch": 1.9304159776109908, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.291014671325684, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8917434215545654, + "num_tokens": 588982864.0, + "step": 15175 + }, + { + "epoch": 1.9305431878895813, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.302651405334473, + "learning_rate": 1e-06, + "loss": 0.3806, + "mean_token_accuracy": 0.8728510141372681, + "num_tokens": 589027090.0, + "step": 15176 + }, + { + "epoch": 1.9306703981681719, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.398921966552734, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8810485005378723, + "num_tokens": 589069263.0, + "step": 15177 + }, + { + "epoch": 1.9307976084467624, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.2654390335083, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8925085067749023, + "num_tokens": 589116234.0, + "step": 15178 + }, + { + "epoch": 1.930924818725353, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.36822509765625, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8828033804893494, + "num_tokens": 589155709.0, + "step": 15179 + }, + { + "epoch": 1.9310520290039435, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.407217979431152, + "learning_rate": 1e-06, + "loss": 0.4014, + "mean_token_accuracy": 0.867283284664154, + "num_tokens": 589199081.0, + "step": 15180 + }, + { + "epoch": 1.931179239282534, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.30235481262207, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.9075222015380859, + "num_tokens": 589234899.0, + "step": 15181 + }, + { + "epoch": 1.9313064495611245, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.353926658630371, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8727540373802185, + "num_tokens": 589278097.0, + "step": 15182 + }, + { + "epoch": 1.931433659839715, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.413820266723633, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8867570757865906, + "num_tokens": 589316638.0, + "step": 15183 + }, + { + "epoch": 1.9315608701183056, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.338059425354004, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8983062505722046, + "num_tokens": 589355320.0, + "step": 15184 + }, + { + "epoch": 1.931688080396896, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.39635181427002, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.8801339268684387, + "num_tokens": 589395054.0, + "step": 15185 + }, + { + "epoch": 1.9318152906754866, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.445821762084961, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8825491666793823, + "num_tokens": 589427569.0, + "step": 15186 + }, + { + "epoch": 1.9319425009540772, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.37934684753418, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8913231492042542, + "num_tokens": 589460400.0, + "step": 15187 + }, + { + "epoch": 1.9320697112326677, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.395835876464844, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8943490982055664, + "num_tokens": 589496069.0, + "step": 15188 + }, + { + "epoch": 1.9321969215112582, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.229238510131836, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8983291387557983, + "num_tokens": 589535793.0, + "step": 15189 + }, + { + "epoch": 1.9323241317898487, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.377364158630371, + "learning_rate": 1e-06, + "loss": 0.3886, + "mean_token_accuracy": 0.8735491037368774, + "num_tokens": 589581310.0, + "step": 15190 + }, + { + "epoch": 1.9324513420684393, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.351483345031738, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8857629299163818, + "num_tokens": 589621833.0, + "step": 15191 + }, + { + "epoch": 1.9325785523470298, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.436112403869629, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8855031132698059, + "num_tokens": 589660637.0, + "step": 15192 + }, + { + "epoch": 1.93270576262562, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.328889846801758, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8909508585929871, + "num_tokens": 589699208.0, + "step": 15193 + }, + { + "epoch": 1.9328329729042106, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.412410736083984, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8891137838363647, + "num_tokens": 589734577.0, + "step": 15194 + }, + { + "epoch": 1.9329601831828012, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.354435920715332, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8808940649032593, + "num_tokens": 589775554.0, + "step": 15195 + }, + { + "epoch": 1.9330873934613917, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.297618865966797, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9006556272506714, + "num_tokens": 589813644.0, + "step": 15196 + }, + { + "epoch": 1.9332146037399822, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.349119186401367, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9013681411743164, + "num_tokens": 589847068.0, + "step": 15197 + }, + { + "epoch": 1.9333418140185727, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.344639778137207, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8848133683204651, + "num_tokens": 589889982.0, + "step": 15198 + }, + { + "epoch": 1.933469024297163, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.435367584228516, + "learning_rate": 1e-06, + "loss": 0.3732, + "mean_token_accuracy": 0.8734734058380127, + "num_tokens": 589926971.0, + "step": 15199 + }, + { + "epoch": 1.9335962345757536, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.384379386901855, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.882012128829956, + "num_tokens": 589966558.0, + "step": 15200 + }, + { + "epoch": 1.933723444854344, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.363911628723145, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8761792778968811, + "num_tokens": 590008002.0, + "step": 15201 + }, + { + "epoch": 1.9338506551329346, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.350409507751465, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.903893232345581, + "num_tokens": 590042249.0, + "step": 15202 + }, + { + "epoch": 1.9339778654115252, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.342153549194336, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8865794539451599, + "num_tokens": 590084663.0, + "step": 15203 + }, + { + "epoch": 1.9341050756901157, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.289438247680664, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8931564092636108, + "num_tokens": 590127404.0, + "step": 15204 + }, + { + "epoch": 1.9342322859687062, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.260575294494629, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8985121250152588, + "num_tokens": 590171181.0, + "step": 15205 + }, + { + "epoch": 1.9343594962472968, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.352892875671387, + "learning_rate": 1e-06, + "loss": 0.3716, + "mean_token_accuracy": 0.8744996786117554, + "num_tokens": 590217057.0, + "step": 15206 + }, + { + "epoch": 1.9344867065258873, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.309961318969727, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8956303000450134, + "num_tokens": 590257817.0, + "step": 15207 + }, + { + "epoch": 1.9346139168044778, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.492072105407715, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8780032992362976, + "num_tokens": 590292651.0, + "step": 15208 + }, + { + "epoch": 1.9347411270830683, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.340903282165527, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8853062391281128, + "num_tokens": 590331070.0, + "step": 15209 + }, + { + "epoch": 1.9348683373616589, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.352855682373047, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8847470879554749, + "num_tokens": 590368320.0, + "step": 15210 + }, + { + "epoch": 1.9349955476402494, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.31143569946289, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8853533864021301, + "num_tokens": 590410968.0, + "step": 15211 + }, + { + "epoch": 1.93512275791884, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.409880638122559, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8769676685333252, + "num_tokens": 590446042.0, + "step": 15212 + }, + { + "epoch": 1.9352499681974304, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.367877006530762, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8920828104019165, + "num_tokens": 590485787.0, + "step": 15213 + }, + { + "epoch": 1.935377178476021, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.473043441772461, + "learning_rate": 1e-06, + "loss": 0.3985, + "mean_token_accuracy": 0.8695639371871948, + "num_tokens": 590523450.0, + "step": 15214 + }, + { + "epoch": 1.9355043887546115, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.512293815612793, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8860449194908142, + "num_tokens": 590557396.0, + "step": 15215 + }, + { + "epoch": 1.935631599033202, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.34135627746582, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8909801244735718, + "num_tokens": 590590083.0, + "step": 15216 + }, + { + "epoch": 1.9357588093117923, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.30795955657959, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.9003254175186157, + "num_tokens": 590631386.0, + "step": 15217 + }, + { + "epoch": 1.9358860195903829, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.317693710327148, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8904246687889099, + "num_tokens": 590671040.0, + "step": 15218 + }, + { + "epoch": 1.9360132298689734, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.393031120300293, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8980921506881714, + "num_tokens": 590704982.0, + "step": 15219 + }, + { + "epoch": 1.936140440147564, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.417126655578613, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8784200549125671, + "num_tokens": 590744068.0, + "step": 15220 + }, + { + "epoch": 1.9362676504261545, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.403839111328125, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8773830533027649, + "num_tokens": 590784023.0, + "step": 15221 + }, + { + "epoch": 1.936394860704745, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.343210220336914, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9077028036117554, + "num_tokens": 590816365.0, + "step": 15222 + }, + { + "epoch": 1.9365220709833355, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.329082489013672, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.895356297492981, + "num_tokens": 590858814.0, + "step": 15223 + }, + { + "epoch": 1.9366492812619258, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.288630485534668, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8901978135108948, + "num_tokens": 590903161.0, + "step": 15224 + }, + { + "epoch": 1.9367764915405163, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.278258323669434, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.9015154242515564, + "num_tokens": 590943876.0, + "step": 15225 + }, + { + "epoch": 1.9369037018191069, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.333616256713867, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8937565088272095, + "num_tokens": 590985098.0, + "step": 15226 + }, + { + "epoch": 1.9370309120976974, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.356849670410156, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8927329778671265, + "num_tokens": 591025892.0, + "step": 15227 + }, + { + "epoch": 1.937158122376288, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.289352416992188, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8905630111694336, + "num_tokens": 591064172.0, + "step": 15228 + }, + { + "epoch": 1.9372853326548785, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.427774429321289, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8870377540588379, + "num_tokens": 591103438.0, + "step": 15229 + }, + { + "epoch": 1.937412542933469, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.369609832763672, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8819907903671265, + "num_tokens": 591141520.0, + "step": 15230 + }, + { + "epoch": 1.9375397532120595, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.319792747497559, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8904313445091248, + "num_tokens": 591183696.0, + "step": 15231 + }, + { + "epoch": 1.93766696349065, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.290197372436523, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8934721946716309, + "num_tokens": 591228865.0, + "step": 15232 + }, + { + "epoch": 1.9377941737692406, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.39071273803711, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8873392343521118, + "num_tokens": 591266998.0, + "step": 15233 + }, + { + "epoch": 1.937921384047831, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.425025939941406, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8971526622772217, + "num_tokens": 591301284.0, + "step": 15234 + }, + { + "epoch": 1.9380485943264216, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.33160400390625, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8847617506980896, + "num_tokens": 591344330.0, + "step": 15235 + }, + { + "epoch": 1.9381758046050122, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.390606880187988, + "learning_rate": 1e-06, + "loss": 0.3754, + "mean_token_accuracy": 0.8706744313240051, + "num_tokens": 591386504.0, + "step": 15236 + }, + { + "epoch": 1.9383030148836027, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.291605949401855, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8831808567047119, + "num_tokens": 591430485.0, + "step": 15237 + }, + { + "epoch": 1.9384302251621932, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.410613059997559, + "learning_rate": 1e-06, + "loss": 0.4005, + "mean_token_accuracy": 0.8651265501976013, + "num_tokens": 591471999.0, + "step": 15238 + }, + { + "epoch": 1.9385574354407837, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.382037162780762, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.8748656511306763, + "num_tokens": 591511838.0, + "step": 15239 + }, + { + "epoch": 1.9386846457193743, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.348456382751465, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8910173177719116, + "num_tokens": 591549655.0, + "step": 15240 + }, + { + "epoch": 1.9388118559979648, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.306559562683105, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8877471089363098, + "num_tokens": 591592843.0, + "step": 15241 + }, + { + "epoch": 1.938939066276555, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.314363479614258, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8910071849822998, + "num_tokens": 591631487.0, + "step": 15242 + }, + { + "epoch": 1.9390662765551456, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.438267707824707, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8958082795143127, + "num_tokens": 591664390.0, + "step": 15243 + }, + { + "epoch": 1.9391934868337362, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.384211540222168, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8761342763900757, + "num_tokens": 591707126.0, + "step": 15244 + }, + { + "epoch": 1.9393206971123267, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.321452140808105, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8931559324264526, + "num_tokens": 591744957.0, + "step": 15245 + }, + { + "epoch": 1.9394479073909172, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.35041332244873, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8864319324493408, + "num_tokens": 591778339.0, + "step": 15246 + }, + { + "epoch": 1.9395751176695077, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.349664688110352, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9050576686859131, + "num_tokens": 591811465.0, + "step": 15247 + }, + { + "epoch": 1.939702327948098, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.27873420715332, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8821053504943848, + "num_tokens": 591852566.0, + "step": 15248 + }, + { + "epoch": 1.9398295382266886, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.335362434387207, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8860617876052856, + "num_tokens": 591894493.0, + "step": 15249 + }, + { + "epoch": 1.939956748505279, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.44083309173584, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8809762001037598, + "num_tokens": 591932287.0, + "step": 15250 + }, + { + "epoch": 1.9400839587838696, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.358539581298828, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8955831527709961, + "num_tokens": 591971613.0, + "step": 15251 + }, + { + "epoch": 1.9402111690624602, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.30222225189209, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.896043062210083, + "num_tokens": 592011490.0, + "step": 15252 + }, + { + "epoch": 1.9403383793410507, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.382763862609863, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.895444929599762, + "num_tokens": 592050289.0, + "step": 15253 + }, + { + "epoch": 1.9404655896196412, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.344985008239746, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8745882511138916, + "num_tokens": 592091353.0, + "step": 15254 + }, + { + "epoch": 1.9405927998982317, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.44239616394043, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8971778154373169, + "num_tokens": 592122074.0, + "step": 15255 + }, + { + "epoch": 1.9407200101768223, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.36158561706543, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.889525294303894, + "num_tokens": 592164892.0, + "step": 15256 + }, + { + "epoch": 1.9408472204554128, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.307708740234375, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8849215507507324, + "num_tokens": 592206686.0, + "step": 15257 + }, + { + "epoch": 1.9409744307340033, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.378313064575195, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8914695978164673, + "num_tokens": 592247048.0, + "step": 15258 + }, + { + "epoch": 1.9411016410125939, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.283347129821777, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8953226804733276, + "num_tokens": 592286912.0, + "step": 15259 + }, + { + "epoch": 1.9412288512911844, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.443391799926758, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8867029547691345, + "num_tokens": 592323983.0, + "step": 15260 + }, + { + "epoch": 1.941356061569775, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.381364822387695, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8920094966888428, + "num_tokens": 592363215.0, + "step": 15261 + }, + { + "epoch": 1.9414832718483654, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.290184020996094, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.8770368695259094, + "num_tokens": 592406857.0, + "step": 15262 + }, + { + "epoch": 1.941610482126956, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.432555198669434, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8781077861785889, + "num_tokens": 592447091.0, + "step": 15263 + }, + { + "epoch": 1.9417376924055465, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.406597137451172, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8806834816932678, + "num_tokens": 592483998.0, + "step": 15264 + }, + { + "epoch": 1.941864902684137, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.35780143737793, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8756788372993469, + "num_tokens": 592524641.0, + "step": 15265 + }, + { + "epoch": 1.9419921129627273, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.410289764404297, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8812716007232666, + "num_tokens": 592559097.0, + "step": 15266 + }, + { + "epoch": 1.9421193232413179, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.269593238830566, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.9029586315155029, + "num_tokens": 592598766.0, + "step": 15267 + }, + { + "epoch": 1.9422465335199084, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.389530181884766, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8892185688018799, + "num_tokens": 592636583.0, + "step": 15268 + }, + { + "epoch": 1.942373743798499, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.323617935180664, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8919208645820618, + "num_tokens": 592679838.0, + "step": 15269 + }, + { + "epoch": 1.9425009540770894, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.396682739257812, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8746967315673828, + "num_tokens": 592720902.0, + "step": 15270 + }, + { + "epoch": 1.94262816435568, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.396069526672363, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8940794467926025, + "num_tokens": 592757342.0, + "step": 15271 + }, + { + "epoch": 1.9427553746342705, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.339509010314941, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.890918493270874, + "num_tokens": 592793388.0, + "step": 15272 + }, + { + "epoch": 1.9428825849128608, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.337422370910645, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8847191333770752, + "num_tokens": 592836489.0, + "step": 15273 + }, + { + "epoch": 1.9430097951914513, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.34582233428955, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8863420486450195, + "num_tokens": 592877787.0, + "step": 15274 + }, + { + "epoch": 1.9431370054700419, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.317682266235352, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8850682973861694, + "num_tokens": 592923660.0, + "step": 15275 + }, + { + "epoch": 1.9432642157486324, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.300318717956543, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8904380202293396, + "num_tokens": 592964045.0, + "step": 15276 + }, + { + "epoch": 1.943391426027223, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.400200843811035, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8787655234336853, + "num_tokens": 593004519.0, + "step": 15277 + }, + { + "epoch": 1.9435186363058135, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.303210258483887, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8923327326774597, + "num_tokens": 593046325.0, + "step": 15278 + }, + { + "epoch": 1.943645846584404, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.343390464782715, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8936192393302917, + "num_tokens": 593083528.0, + "step": 15279 + }, + { + "epoch": 1.9437730568629945, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.290569305419922, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8925143480300903, + "num_tokens": 593126032.0, + "step": 15280 + }, + { + "epoch": 1.943900267141585, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.314207077026367, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8855828642845154, + "num_tokens": 593177520.0, + "step": 15281 + }, + { + "epoch": 1.9440274774201756, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.273187637329102, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8826873898506165, + "num_tokens": 593225247.0, + "step": 15282 + }, + { + "epoch": 1.944154687698766, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.357439041137695, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8957668542861938, + "num_tokens": 593263405.0, + "step": 15283 + }, + { + "epoch": 1.9442818979773566, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.430191993713379, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8842201232910156, + "num_tokens": 593297410.0, + "step": 15284 + }, + { + "epoch": 1.9444091082559471, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.303136825561523, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.885606050491333, + "num_tokens": 593339623.0, + "step": 15285 + }, + { + "epoch": 1.9445363185345377, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.364439964294434, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.896064281463623, + "num_tokens": 593372433.0, + "step": 15286 + }, + { + "epoch": 1.9446635288131282, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.391284942626953, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8858262300491333, + "num_tokens": 593408084.0, + "step": 15287 + }, + { + "epoch": 1.9447907390917187, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.322957992553711, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.890017032623291, + "num_tokens": 593445889.0, + "step": 15288 + }, + { + "epoch": 1.9449179493703093, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.44973087310791, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8920000195503235, + "num_tokens": 593479574.0, + "step": 15289 + }, + { + "epoch": 1.9450451596488998, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.323600769042969, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8858140707015991, + "num_tokens": 593522734.0, + "step": 15290 + }, + { + "epoch": 1.94517236992749, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.413142204284668, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.880303144454956, + "num_tokens": 593561289.0, + "step": 15291 + }, + { + "epoch": 1.9452995802060806, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.376450538635254, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8882948756217957, + "num_tokens": 593600019.0, + "step": 15292 + }, + { + "epoch": 1.9454267904846712, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.46456527709961, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8956800699234009, + "num_tokens": 593630396.0, + "step": 15293 + }, + { + "epoch": 1.9455540007632617, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.310951232910156, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8984884023666382, + "num_tokens": 593667959.0, + "step": 15294 + }, + { + "epoch": 1.9456812110418522, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.493316650390625, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.8755855560302734, + "num_tokens": 593702980.0, + "step": 15295 + }, + { + "epoch": 1.9458084213204427, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.426401138305664, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8851231336593628, + "num_tokens": 593742871.0, + "step": 15296 + }, + { + "epoch": 1.945935631599033, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.292521476745605, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8757951259613037, + "num_tokens": 593787950.0, + "step": 15297 + }, + { + "epoch": 1.9460628418776236, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.36292552947998, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8907293081283569, + "num_tokens": 593825387.0, + "step": 15298 + }, + { + "epoch": 1.946190052156214, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.288082122802734, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8886315226554871, + "num_tokens": 593867535.0, + "step": 15299 + }, + { + "epoch": 1.9463172624348046, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.347810745239258, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8948829174041748, + "num_tokens": 593914157.0, + "step": 15300 + }, + { + "epoch": 1.9464444727133952, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.385725021362305, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8761812448501587, + "num_tokens": 593960572.0, + "step": 15301 + }, + { + "epoch": 1.9465716829919857, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.455252647399902, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8961223363876343, + "num_tokens": 593990546.0, + "step": 15302 + }, + { + "epoch": 1.9466988932705762, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.374579429626465, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8872962594032288, + "num_tokens": 594032881.0, + "step": 15303 + }, + { + "epoch": 1.9468261035491667, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.38323974609375, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8936105966567993, + "num_tokens": 594069660.0, + "step": 15304 + }, + { + "epoch": 1.9469533138277573, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.348678588867188, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8929516673088074, + "num_tokens": 594105348.0, + "step": 15305 + }, + { + "epoch": 1.9470805241063478, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.480796813964844, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8759011030197144, + "num_tokens": 594140691.0, + "step": 15306 + }, + { + "epoch": 1.9472077343849383, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.40087604522705, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8956857919692993, + "num_tokens": 594173265.0, + "step": 15307 + }, + { + "epoch": 1.9473349446635289, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.357632637023926, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8881566524505615, + "num_tokens": 594208672.0, + "step": 15308 + }, + { + "epoch": 1.9474621549421194, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.283026695251465, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8861802816390991, + "num_tokens": 594253092.0, + "step": 15309 + }, + { + "epoch": 1.94758936522071, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.256200790405273, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8896504640579224, + "num_tokens": 594301946.0, + "step": 15310 + }, + { + "epoch": 1.9477165754993004, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.462722778320312, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.889782190322876, + "num_tokens": 594332968.0, + "step": 15311 + }, + { + "epoch": 1.947843785777891, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.374302864074707, + "learning_rate": 1e-06, + "loss": 0.3768, + "mean_token_accuracy": 0.8745912313461304, + "num_tokens": 594375075.0, + "step": 15312 + }, + { + "epoch": 1.9479709960564815, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.343536376953125, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.9003689289093018, + "num_tokens": 594412687.0, + "step": 15313 + }, + { + "epoch": 1.948098206335072, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.363673210144043, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8818265199661255, + "num_tokens": 594453474.0, + "step": 15314 + }, + { + "epoch": 1.9482254166136623, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.380789756774902, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8881779909133911, + "num_tokens": 594492284.0, + "step": 15315 + }, + { + "epoch": 1.9483526268922529, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.392963409423828, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8983821868896484, + "num_tokens": 594527839.0, + "step": 15316 + }, + { + "epoch": 1.9484798371708434, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.38167953491211, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8762907981872559, + "num_tokens": 594572730.0, + "step": 15317 + }, + { + "epoch": 1.948607047449434, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.311270713806152, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8901488780975342, + "num_tokens": 594610759.0, + "step": 15318 + }, + { + "epoch": 1.9487342577280244, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.306439399719238, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8891112208366394, + "num_tokens": 594649230.0, + "step": 15319 + }, + { + "epoch": 1.948861468006615, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.480419158935547, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.887008011341095, + "num_tokens": 594679867.0, + "step": 15320 + }, + { + "epoch": 1.9489886782852053, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.361634254455566, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8955446481704712, + "num_tokens": 594713091.0, + "step": 15321 + }, + { + "epoch": 1.9491158885637958, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.514134407043457, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.885316014289856, + "num_tokens": 594744274.0, + "step": 15322 + }, + { + "epoch": 1.9492430988423863, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.351298332214355, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8986890912055969, + "num_tokens": 594778025.0, + "step": 15323 + }, + { + "epoch": 1.9493703091209769, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.509429931640625, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.888492226600647, + "num_tokens": 594811939.0, + "step": 15324 + }, + { + "epoch": 1.9494975193995674, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.313815116882324, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8966773152351379, + "num_tokens": 594856587.0, + "step": 15325 + }, + { + "epoch": 1.949624729678158, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.357571601867676, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8894213438034058, + "num_tokens": 594890487.0, + "step": 15326 + }, + { + "epoch": 1.9497519399567484, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.35575008392334, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8937093019485474, + "num_tokens": 594934801.0, + "step": 15327 + }, + { + "epoch": 1.949879150235339, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.338082313537598, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8779367208480835, + "num_tokens": 594977278.0, + "step": 15328 + }, + { + "epoch": 1.9500063605139295, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.398205757141113, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8852320313453674, + "num_tokens": 595014526.0, + "step": 15329 + }, + { + "epoch": 1.95013357079252, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.281342506408691, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.9025977849960327, + "num_tokens": 595053858.0, + "step": 15330 + }, + { + "epoch": 1.9502607810711106, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.359384536743164, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.877623438835144, + "num_tokens": 595097574.0, + "step": 15331 + }, + { + "epoch": 1.950387991349701, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.406038284301758, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8904461860656738, + "num_tokens": 595134786.0, + "step": 15332 + }, + { + "epoch": 1.9505152016282916, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.35181999206543, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8971983790397644, + "num_tokens": 595171330.0, + "step": 15333 + }, + { + "epoch": 1.9506424119068821, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.413362503051758, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9000399112701416, + "num_tokens": 595205413.0, + "step": 15334 + }, + { + "epoch": 1.9507696221854727, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.305734634399414, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.893742024898529, + "num_tokens": 595246257.0, + "step": 15335 + }, + { + "epoch": 1.9508968324640632, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.480807304382324, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8927738070487976, + "num_tokens": 595276602.0, + "step": 15336 + }, + { + "epoch": 1.9510240427426537, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.456124305725098, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8886884450912476, + "num_tokens": 595310050.0, + "step": 15337 + }, + { + "epoch": 1.9511512530212443, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.3612642288208, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8875144720077515, + "num_tokens": 595347636.0, + "step": 15338 + }, + { + "epoch": 1.9512784632998348, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.355263710021973, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.887710690498352, + "num_tokens": 595388519.0, + "step": 15339 + }, + { + "epoch": 1.951405673578425, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.323384284973145, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.890118420124054, + "num_tokens": 595423440.0, + "step": 15340 + }, + { + "epoch": 1.9515328838570156, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.316025733947754, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8896811008453369, + "num_tokens": 595465974.0, + "step": 15341 + }, + { + "epoch": 1.9516600941356061, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.262591361999512, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.9002645015716553, + "num_tokens": 595506072.0, + "step": 15342 + }, + { + "epoch": 1.9517873044141967, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.381098747253418, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8903735280036926, + "num_tokens": 595550340.0, + "step": 15343 + }, + { + "epoch": 1.9519145146927872, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.3562650680542, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.889032244682312, + "num_tokens": 595589528.0, + "step": 15344 + }, + { + "epoch": 1.9520417249713777, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.305399894714355, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8847516775131226, + "num_tokens": 595632850.0, + "step": 15345 + }, + { + "epoch": 1.952168935249968, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.365283966064453, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8954648971557617, + "num_tokens": 595670067.0, + "step": 15346 + }, + { + "epoch": 1.9522961455285586, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.285287857055664, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8883376121520996, + "num_tokens": 595715351.0, + "step": 15347 + }, + { + "epoch": 1.952423355807149, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.43393325805664, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8879410028457642, + "num_tokens": 595750752.0, + "step": 15348 + }, + { + "epoch": 1.9525505660857396, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.36123275756836, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.9038857221603394, + "num_tokens": 595783910.0, + "step": 15349 + }, + { + "epoch": 1.9526777763643302, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.4099702835083, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8894869089126587, + "num_tokens": 595820276.0, + "step": 15350 + }, + { + "epoch": 1.9528049866429207, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.366537094116211, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8911048173904419, + "num_tokens": 595856718.0, + "step": 15351 + }, + { + "epoch": 1.9529321969215112, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.416698455810547, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8779059648513794, + "num_tokens": 595895835.0, + "step": 15352 + }, + { + "epoch": 1.9530594072001017, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.381049156188965, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8769255876541138, + "num_tokens": 595936078.0, + "step": 15353 + }, + { + "epoch": 1.9531866174786923, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.35901927947998, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8901098370552063, + "num_tokens": 595973592.0, + "step": 15354 + }, + { + "epoch": 1.9533138277572828, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.451281547546387, + "learning_rate": 1e-06, + "loss": 0.4039, + "mean_token_accuracy": 0.8629753589630127, + "num_tokens": 596011962.0, + "step": 15355 + }, + { + "epoch": 1.9534410380358733, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.369573593139648, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8823248147964478, + "num_tokens": 596049470.0, + "step": 15356 + }, + { + "epoch": 1.9535682483144639, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.422188758850098, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8943749070167542, + "num_tokens": 596084774.0, + "step": 15357 + }, + { + "epoch": 1.9536954585930544, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.44662857055664, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.877414345741272, + "num_tokens": 596119560.0, + "step": 15358 + }, + { + "epoch": 1.953822668871645, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.4259672164917, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8874220848083496, + "num_tokens": 596159685.0, + "step": 15359 + }, + { + "epoch": 1.9539498791502354, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.325444221496582, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8961698412895203, + "num_tokens": 596198326.0, + "step": 15360 + }, + { + "epoch": 1.954077089428826, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.431930541992188, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8832510709762573, + "num_tokens": 596234479.0, + "step": 15361 + }, + { + "epoch": 1.9542042997074165, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.309776306152344, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8937082290649414, + "num_tokens": 596279495.0, + "step": 15362 + }, + { + "epoch": 1.954331509986007, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.29928207397461, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8943329453468323, + "num_tokens": 596323861.0, + "step": 15363 + }, + { + "epoch": 1.9544587202645973, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.395051956176758, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8944882154464722, + "num_tokens": 596359870.0, + "step": 15364 + }, + { + "epoch": 1.9545859305431879, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.504230499267578, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8817369341850281, + "num_tokens": 596397133.0, + "step": 15365 + }, + { + "epoch": 1.9547131408217784, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.367024421691895, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8878953456878662, + "num_tokens": 596436279.0, + "step": 15366 + }, + { + "epoch": 1.954840351100369, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.333954811096191, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8962894678115845, + "num_tokens": 596475244.0, + "step": 15367 + }, + { + "epoch": 1.9549675613789594, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.397306442260742, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8920010328292847, + "num_tokens": 596508069.0, + "step": 15368 + }, + { + "epoch": 1.95509477165755, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.462359428405762, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8869718909263611, + "num_tokens": 596546348.0, + "step": 15369 + }, + { + "epoch": 1.9552219819361403, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.313632011413574, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8854410648345947, + "num_tokens": 596583982.0, + "step": 15370 + }, + { + "epoch": 1.9553491922147308, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.368722915649414, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9085955619812012, + "num_tokens": 596614988.0, + "step": 15371 + }, + { + "epoch": 1.9554764024933213, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.356473922729492, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8860267400741577, + "num_tokens": 596655651.0, + "step": 15372 + }, + { + "epoch": 1.9556036127719119, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.32878589630127, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8892251253128052, + "num_tokens": 596691644.0, + "step": 15373 + }, + { + "epoch": 1.9557308230505024, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.427178382873535, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8796255588531494, + "num_tokens": 596727870.0, + "step": 15374 + }, + { + "epoch": 1.955858033329093, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.351578712463379, + "learning_rate": 1e-06, + "loss": 0.3786, + "mean_token_accuracy": 0.8746503591537476, + "num_tokens": 596771452.0, + "step": 15375 + }, + { + "epoch": 1.9559852436076834, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.424164772033691, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8868733644485474, + "num_tokens": 596810781.0, + "step": 15376 + }, + { + "epoch": 1.956112453886274, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.390040397644043, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8943353891372681, + "num_tokens": 596849206.0, + "step": 15377 + }, + { + "epoch": 1.9562396641648645, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.291411399841309, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8917558193206787, + "num_tokens": 596889746.0, + "step": 15378 + }, + { + "epoch": 1.956366874443455, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.394920349121094, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.9001197814941406, + "num_tokens": 596926505.0, + "step": 15379 + }, + { + "epoch": 1.9564940847220456, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.405983924865723, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8908798694610596, + "num_tokens": 596965704.0, + "step": 15380 + }, + { + "epoch": 1.956621295000636, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.473333358764648, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8947796821594238, + "num_tokens": 596997801.0, + "step": 15381 + }, + { + "epoch": 1.9567485052792266, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.290428161621094, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.902144193649292, + "num_tokens": 597037523.0, + "step": 15382 + }, + { + "epoch": 1.9568757155578171, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.388309478759766, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8817341923713684, + "num_tokens": 597080911.0, + "step": 15383 + }, + { + "epoch": 1.9570029258364077, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.287041664123535, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8956516981124878, + "num_tokens": 597122326.0, + "step": 15384 + }, + { + "epoch": 1.9571301361149982, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.422279357910156, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8863127827644348, + "num_tokens": 597163277.0, + "step": 15385 + }, + { + "epoch": 1.9572573463935887, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.448890686035156, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8725554943084717, + "num_tokens": 597202552.0, + "step": 15386 + }, + { + "epoch": 1.9573845566721793, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.414785385131836, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8980503082275391, + "num_tokens": 597237448.0, + "step": 15387 + }, + { + "epoch": 1.9575117669507698, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.46252155303955, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8774561882019043, + "num_tokens": 597272346.0, + "step": 15388 + }, + { + "epoch": 1.95763897722936, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.425904273986816, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8952112197875977, + "num_tokens": 597310587.0, + "step": 15389 + }, + { + "epoch": 1.9577661875079506, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.633974075317383, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.883527934551239, + "num_tokens": 597339153.0, + "step": 15390 + }, + { + "epoch": 1.9578933977865411, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.272191047668457, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.9070091843605042, + "num_tokens": 597372615.0, + "step": 15391 + }, + { + "epoch": 1.9580206080651317, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.47194766998291, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.9042628407478333, + "num_tokens": 597402263.0, + "step": 15392 + }, + { + "epoch": 1.9581478183437222, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.376802444458008, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8977360725402832, + "num_tokens": 597439109.0, + "step": 15393 + }, + { + "epoch": 1.9582750286223127, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.361710548400879, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.9029538035392761, + "num_tokens": 597471477.0, + "step": 15394 + }, + { + "epoch": 1.958402238900903, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.744586944580078, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8874239921569824, + "num_tokens": 597503089.0, + "step": 15395 + }, + { + "epoch": 1.9585294491794936, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.37595272064209, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8865465521812439, + "num_tokens": 597539565.0, + "step": 15396 + }, + { + "epoch": 1.958656659458084, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.353965759277344, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.882314145565033, + "num_tokens": 597581430.0, + "step": 15397 + }, + { + "epoch": 1.9587838697366746, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.338944435119629, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8947428464889526, + "num_tokens": 597619862.0, + "step": 15398 + }, + { + "epoch": 1.9589110800152651, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.32291030883789, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8938708305358887, + "num_tokens": 597660556.0, + "step": 15399 + }, + { + "epoch": 1.9590382902938557, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.441025733947754, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8947770595550537, + "num_tokens": 597696349.0, + "step": 15400 + }, + { + "epoch": 1.9591655005724462, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.345747947692871, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.900437593460083, + "num_tokens": 597739249.0, + "step": 15401 + }, + { + "epoch": 1.9592927108510367, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.321609497070312, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8896802067756653, + "num_tokens": 597781452.0, + "step": 15402 + }, + { + "epoch": 1.9594199211296273, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.48271656036377, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8820489645004272, + "num_tokens": 597818306.0, + "step": 15403 + }, + { + "epoch": 1.9595471314082178, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.334756851196289, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8801860809326172, + "num_tokens": 597859158.0, + "step": 15404 + }, + { + "epoch": 1.9596743416868083, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.39169692993164, + "learning_rate": 1e-06, + "loss": 0.3771, + "mean_token_accuracy": 0.8721532821655273, + "num_tokens": 597904189.0, + "step": 15405 + }, + { + "epoch": 1.9598015519653988, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.416617393493652, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8859843611717224, + "num_tokens": 597940946.0, + "step": 15406 + }, + { + "epoch": 1.9599287622439894, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.323811531066895, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8930585980415344, + "num_tokens": 597982966.0, + "step": 15407 + }, + { + "epoch": 1.96005597252258, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.473204612731934, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.883968710899353, + "num_tokens": 598017249.0, + "step": 15408 + }, + { + "epoch": 1.9601831828011704, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.363473892211914, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8909831643104553, + "num_tokens": 598057471.0, + "step": 15409 + }, + { + "epoch": 1.960310393079761, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.488340377807617, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8829672932624817, + "num_tokens": 598092705.0, + "step": 15410 + }, + { + "epoch": 1.9604376033583515, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.373645782470703, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8870663642883301, + "num_tokens": 598136309.0, + "step": 15411 + }, + { + "epoch": 1.960564813636942, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.285979270935059, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8994093537330627, + "num_tokens": 598176100.0, + "step": 15412 + }, + { + "epoch": 1.9606920239155323, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.430925369262695, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8908342123031616, + "num_tokens": 598212613.0, + "step": 15413 + }, + { + "epoch": 1.9608192341941229, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.373623847961426, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8978532552719116, + "num_tokens": 598251266.0, + "step": 15414 + }, + { + "epoch": 1.9609464444727134, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.373552322387695, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8935332298278809, + "num_tokens": 598289212.0, + "step": 15415 + }, + { + "epoch": 1.961073654751304, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.47225284576416, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8917311429977417, + "num_tokens": 598326245.0, + "step": 15416 + }, + { + "epoch": 1.9612008650298944, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.400472640991211, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8825453519821167, + "num_tokens": 598369362.0, + "step": 15417 + }, + { + "epoch": 1.961328075308485, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.370817184448242, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8891850709915161, + "num_tokens": 598415026.0, + "step": 15418 + }, + { + "epoch": 1.9614552855870753, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.483983993530273, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8860180974006653, + "num_tokens": 598448742.0, + "step": 15419 + }, + { + "epoch": 1.9615824958656658, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.604689598083496, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8938682079315186, + "num_tokens": 598493452.0, + "step": 15420 + }, + { + "epoch": 1.9617097061442563, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.382702827453613, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8894557356834412, + "num_tokens": 598529595.0, + "step": 15421 + }, + { + "epoch": 1.9618369164228469, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0788440704345703e-05, + "grad_norm": 9.472407341003418, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8822311162948608, + "num_tokens": 598567653.0, + "step": 15422 + }, + { + "epoch": 1.9619641267014374, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.41280460357666, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8880817294120789, + "num_tokens": 598605350.0, + "step": 15423 + }, + { + "epoch": 1.962091336980028, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.336846351623535, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8981944918632507, + "num_tokens": 598646584.0, + "step": 15424 + }, + { + "epoch": 1.9622185472586184, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.431986808776855, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.884161651134491, + "num_tokens": 598685946.0, + "step": 15425 + }, + { + "epoch": 1.962345757537209, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.44354248046875, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8800016045570374, + "num_tokens": 598724039.0, + "step": 15426 + }, + { + "epoch": 1.9624729678157995, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.393089294433594, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8986276984214783, + "num_tokens": 598765612.0, + "step": 15427 + }, + { + "epoch": 1.96260017809439, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.389982223510742, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8954795002937317, + "num_tokens": 598803335.0, + "step": 15428 + }, + { + "epoch": 1.9627273883729806, + "ewc_loss": 0.017333984375, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.399918556213379, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8939787745475769, + "num_tokens": 598838498.0, + "step": 15429 + }, + { + "epoch": 1.962854598651571, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.532669067382812e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.36962890625, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8901407718658447, + "num_tokens": 598886576.0, + "step": 15430 + }, + { + "epoch": 1.9629818089301616, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.394868850708008, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8891882300376892, + "num_tokens": 598921475.0, + "step": 15431 + }, + { + "epoch": 1.9631090192087521, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.369258880615234, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8790240287780762, + "num_tokens": 598964465.0, + "step": 15432 + }, + { + "epoch": 1.9632362294873427, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.395135879516602, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8910025358200073, + "num_tokens": 599004302.0, + "step": 15433 + }, + { + "epoch": 1.9633634397659332, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.348674774169922, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8884562253952026, + "num_tokens": 599046154.0, + "step": 15434 + }, + { + "epoch": 1.9634906500445237, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.414942741394043, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8784797787666321, + "num_tokens": 599082502.0, + "step": 15435 + }, + { + "epoch": 1.9636178603231143, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.400778770446777, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8844671249389648, + "num_tokens": 599122229.0, + "step": 15436 + }, + { + "epoch": 1.9637450706017048, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.388036727905273, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8908748626708984, + "num_tokens": 599159239.0, + "step": 15437 + }, + { + "epoch": 1.963872280880295, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.448882102966309, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8880563378334045, + "num_tokens": 599195066.0, + "step": 15438 + }, + { + "epoch": 1.9639994911588856, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.320771217346191, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8987215757369995, + "num_tokens": 599239464.0, + "step": 15439 + }, + { + "epoch": 1.9641267014374761, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.35580062866211, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8940461874008179, + "num_tokens": 599278794.0, + "step": 15440 + }, + { + "epoch": 1.9642539117160667, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.472335815429688, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8934727311134338, + "num_tokens": 599315252.0, + "step": 15441 + }, + { + "epoch": 1.9643811219946572, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.394699096679688, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.895600438117981, + "num_tokens": 599352145.0, + "step": 15442 + }, + { + "epoch": 1.9645083322732477, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.425593376159668, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.892720103263855, + "num_tokens": 599386369.0, + "step": 15443 + }, + { + "epoch": 1.964635542551838, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.396257400512695, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8849191665649414, + "num_tokens": 599430513.0, + "step": 15444 + }, + { + "epoch": 1.9647627528304286, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.380537033081055, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8898411393165588, + "num_tokens": 599471780.0, + "step": 15445 + }, + { + "epoch": 1.964889963109019, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.386635780334473, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8979869484901428, + "num_tokens": 599504965.0, + "step": 15446 + }, + { + "epoch": 1.9650171733876096, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.382555961608887, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8896675705909729, + "num_tokens": 599544092.0, + "step": 15447 + }, + { + "epoch": 1.9651443836662001, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.447271347045898, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8799774050712585, + "num_tokens": 599586107.0, + "step": 15448 + }, + { + "epoch": 1.9652715939447907, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.407934188842773, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8874632120132446, + "num_tokens": 599622725.0, + "step": 15449 + }, + { + "epoch": 1.9653988042233812, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.416646003723145, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8860198259353638, + "num_tokens": 599661732.0, + "step": 15450 + }, + { + "epoch": 1.9655260145019717, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.42505931854248, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8872107863426208, + "num_tokens": 599695899.0, + "step": 15451 + }, + { + "epoch": 1.9656532247805623, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.412055015563965, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8861216306686401, + "num_tokens": 599737373.0, + "step": 15452 + }, + { + "epoch": 1.9657804350591528, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.321270942687988, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8931111693382263, + "num_tokens": 599778823.0, + "step": 15453 + }, + { + "epoch": 1.9659076453377433, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.40481948852539, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8806095719337463, + "num_tokens": 599818036.0, + "step": 15454 + }, + { + "epoch": 1.9660348556163338, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.370006561279297, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8878456354141235, + "num_tokens": 599862028.0, + "step": 15455 + }, + { + "epoch": 1.9661620658949244, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.371747970581055, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8869304656982422, + "num_tokens": 599905812.0, + "step": 15456 + }, + { + "epoch": 1.966289276173515, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.456546783447266, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8935106992721558, + "num_tokens": 599940848.0, + "step": 15457 + }, + { + "epoch": 1.9664164864521054, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.323945045471191, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9059754610061646, + "num_tokens": 599981308.0, + "step": 15458 + }, + { + "epoch": 1.966543696730696, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.455819129943848, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8888159990310669, + "num_tokens": 600018708.0, + "step": 15459 + }, + { + "epoch": 1.9666709070092865, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.375555038452148, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8999711275100708, + "num_tokens": 600059771.0, + "step": 15460 + }, + { + "epoch": 1.966798117287877, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.359219551086426, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.879569411277771, + "num_tokens": 600101080.0, + "step": 15461 + }, + { + "epoch": 1.9669253275664673, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.391289710998535, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8872812986373901, + "num_tokens": 600142712.0, + "step": 15462 + }, + { + "epoch": 1.9670525378450578, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.43383502960205, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8780014514923096, + "num_tokens": 600184382.0, + "step": 15463 + }, + { + "epoch": 1.9671797481236484, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.328712463378906, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8918721675872803, + "num_tokens": 600227790.0, + "step": 15464 + }, + { + "epoch": 1.967306958402239, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.468135833740234, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8811713457107544, + "num_tokens": 600266826.0, + "step": 15465 + }, + { + "epoch": 1.9674341686808294, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.387818336486816, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8757874369621277, + "num_tokens": 600308631.0, + "step": 15466 + }, + { + "epoch": 1.96756137895942, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.387808799743652, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.8712306022644043, + "num_tokens": 600351799.0, + "step": 15467 + }, + { + "epoch": 1.9676885892380103, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.453330039978027, + "learning_rate": 1e-06, + "loss": 0.3738, + "mean_token_accuracy": 0.8748663663864136, + "num_tokens": 600390847.0, + "step": 15468 + }, + { + "epoch": 1.9678157995166008, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.272912979125977, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8860487937927246, + "num_tokens": 600436250.0, + "step": 15469 + }, + { + "epoch": 1.9679430097951913, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.429023742675781, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.888708770275116, + "num_tokens": 600470906.0, + "step": 15470 + }, + { + "epoch": 1.9680702200737819, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.295821189880371, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8857790231704712, + "num_tokens": 600517907.0, + "step": 15471 + }, + { + "epoch": 1.9681974303523724, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.310394287109375, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.9011396765708923, + "num_tokens": 600556493.0, + "step": 15472 + }, + { + "epoch": 1.968324640630963, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.373393058776855, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8917791843414307, + "num_tokens": 600599870.0, + "step": 15473 + }, + { + "epoch": 1.9684518509095534, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.364241600036621, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8838029503822327, + "num_tokens": 600640889.0, + "step": 15474 + }, + { + "epoch": 1.968579061188144, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.409809112548828, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8837233185768127, + "num_tokens": 600681709.0, + "step": 15475 + }, + { + "epoch": 1.9687062714667345, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.385724067687988, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8883076310157776, + "num_tokens": 600717656.0, + "step": 15476 + }, + { + "epoch": 1.968833481745325, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.422412872314453, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8860765099525452, + "num_tokens": 600758280.0, + "step": 15477 + }, + { + "epoch": 1.9689606920239155, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.408941268920898, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8820077180862427, + "num_tokens": 600802414.0, + "step": 15478 + }, + { + "epoch": 1.969087902302506, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.376399040222168, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8839502334594727, + "num_tokens": 600843174.0, + "step": 15479 + }, + { + "epoch": 1.9692151125810966, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.475598335266113, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8815115690231323, + "num_tokens": 600882469.0, + "step": 15480 + }, + { + "epoch": 1.9693423228596871, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.41556167602539, + "learning_rate": 1e-06, + "loss": 0.3775, + "mean_token_accuracy": 0.8731247782707214, + "num_tokens": 600921833.0, + "step": 15481 + }, + { + "epoch": 1.9694695331382777, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.395060539245605, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8934270143508911, + "num_tokens": 600964374.0, + "step": 15482 + }, + { + "epoch": 1.9695967434168682, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.486095428466797, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8938515186309814, + "num_tokens": 600995541.0, + "step": 15483 + }, + { + "epoch": 1.9697239536954587, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.431245803833008, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.897947371006012, + "num_tokens": 601032072.0, + "step": 15484 + }, + { + "epoch": 1.9698511639740492, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.389134407043457, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.890110969543457, + "num_tokens": 601071847.0, + "step": 15485 + }, + { + "epoch": 1.9699783742526398, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.407240867614746, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.9013745188713074, + "num_tokens": 601112593.0, + "step": 15486 + }, + { + "epoch": 1.97010558453123, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.387808799743652, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8890100717544556, + "num_tokens": 601149038.0, + "step": 15487 + }, + { + "epoch": 1.9702327948098206, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.428845405578613, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8797690272331238, + "num_tokens": 601192467.0, + "step": 15488 + }, + { + "epoch": 1.9703600050884111, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.399051666259766, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8784299492835999, + "num_tokens": 601232568.0, + "step": 15489 + }, + { + "epoch": 1.9704872153670017, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.390819549560547, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8947492837905884, + "num_tokens": 601269166.0, + "step": 15490 + }, + { + "epoch": 1.9706144256455922, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.406190872192383, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8917769193649292, + "num_tokens": 601306793.0, + "step": 15491 + }, + { + "epoch": 1.9707416359241827, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.367566108703613, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8920459747314453, + "num_tokens": 601343685.0, + "step": 15492 + }, + { + "epoch": 1.970868846202773, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.418769836425781, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.9008184671401978, + "num_tokens": 601379293.0, + "step": 15493 + }, + { + "epoch": 1.9709960564813636, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.489188194274902, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8745331168174744, + "num_tokens": 601418103.0, + "step": 15494 + }, + { + "epoch": 1.971123266759954, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.489253044128418, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8716418743133545, + "num_tokens": 601451574.0, + "step": 15495 + }, + { + "epoch": 1.9712504770385446, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.26150131225586, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.887236475944519, + "num_tokens": 601493295.0, + "step": 15496 + }, + { + "epoch": 1.9713776873171351, + "ewc_loss": 0.0174560546875, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.510075569152832, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8786667585372925, + "num_tokens": 601527664.0, + "step": 15497 + }, + { + "epoch": 1.9715048975957257, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.406834602355957, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8813909888267517, + "num_tokens": 601562057.0, + "step": 15498 + }, + { + "epoch": 1.9716321078743162, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.351899147033691, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.9005658626556396, + "num_tokens": 601601691.0, + "step": 15499 + }, + { + "epoch": 1.9717593181529067, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.389182090759277, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8857473731040955, + "num_tokens": 601635088.0, + "step": 15500 + }, + { + "epoch": 1.9718865284314973, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.507722854614258, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.882024884223938, + "num_tokens": 601668705.0, + "step": 15501 + }, + { + "epoch": 1.9720137387100878, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.34088134765625, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8873897790908813, + "num_tokens": 601709332.0, + "step": 15502 + }, + { + "epoch": 1.9721409489886783, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.441874504089355, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8912838101387024, + "num_tokens": 601745159.0, + "step": 15503 + }, + { + "epoch": 1.9722681592672688, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.32189655303955, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8921592235565186, + "num_tokens": 601783876.0, + "step": 15504 + }, + { + "epoch": 1.9723953695458594, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.372610092163086, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8941776752471924, + "num_tokens": 601821109.0, + "step": 15505 + }, + { + "epoch": 1.97252257982445, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.481063842773438, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8847380876541138, + "num_tokens": 601858031.0, + "step": 15506 + }, + { + "epoch": 1.9726497901030404, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.376059532165527, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8850476741790771, + "num_tokens": 601897688.0, + "step": 15507 + }, + { + "epoch": 1.972777000381631, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.376283645629883, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.885725736618042, + "num_tokens": 601935164.0, + "step": 15508 + }, + { + "epoch": 1.9729042106602215, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.38719367980957, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8844165802001953, + "num_tokens": 601972147.0, + "step": 15509 + }, + { + "epoch": 1.973031420938812, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.453778266906738, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8857163786888123, + "num_tokens": 602009874.0, + "step": 15510 + }, + { + "epoch": 1.9731586312174023, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.435399055480957, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8917132019996643, + "num_tokens": 602047569.0, + "step": 15511 + }, + { + "epoch": 1.9732858414959928, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.451581001281738, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8921480178833008, + "num_tokens": 602085522.0, + "step": 15512 + }, + { + "epoch": 1.9734130517745834, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.380552291870117, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8886094093322754, + "num_tokens": 602127249.0, + "step": 15513 + }, + { + "epoch": 1.973540262053174, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.42906665802002, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8866230249404907, + "num_tokens": 602163380.0, + "step": 15514 + }, + { + "epoch": 1.9736674723317644, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.356355667114258, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8844605684280396, + "num_tokens": 602201411.0, + "step": 15515 + }, + { + "epoch": 1.973794682610355, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.390259742736816, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8934988379478455, + "num_tokens": 602236288.0, + "step": 15516 + }, + { + "epoch": 1.9739218928889453, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.412698745727539, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8904406428337097, + "num_tokens": 602277482.0, + "step": 15517 + }, + { + "epoch": 1.9740491031675358, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.422431945800781, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8929085731506348, + "num_tokens": 602311233.0, + "step": 15518 + }, + { + "epoch": 1.9741763134461263, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.462464332580566, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8845949172973633, + "num_tokens": 602350746.0, + "step": 15519 + }, + { + "epoch": 1.9743035237247168, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.267104148864746, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8912159204483032, + "num_tokens": 602395924.0, + "step": 15520 + }, + { + "epoch": 1.9744307340033074, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.433588027954102, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.886541485786438, + "num_tokens": 602438570.0, + "step": 15521 + }, + { + "epoch": 1.974557944281898, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.373919486999512, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8761379718780518, + "num_tokens": 602484505.0, + "step": 15522 + }, + { + "epoch": 1.9746851545604884, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.432921409606934, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8845298290252686, + "num_tokens": 602524963.0, + "step": 15523 + }, + { + "epoch": 1.974812364839079, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.433916091918945, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8825958967208862, + "num_tokens": 602564240.0, + "step": 15524 + }, + { + "epoch": 1.9749395751176695, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.412954330444336, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8899643421173096, + "num_tokens": 602605654.0, + "step": 15525 + }, + { + "epoch": 1.97506678539626, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.490849494934082, + "learning_rate": 1e-06, + "loss": 0.3865, + "mean_token_accuracy": 0.8680102229118347, + "num_tokens": 602649111.0, + "step": 15526 + }, + { + "epoch": 1.9751939956748505, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.281414985656738, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.9093698859214783, + "num_tokens": 602695103.0, + "step": 15527 + }, + { + "epoch": 1.975321205953441, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.38471508026123, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.9029250741004944, + "num_tokens": 602730207.0, + "step": 15528 + }, + { + "epoch": 1.9754484162320316, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.491707801818848, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8918836116790771, + "num_tokens": 602762866.0, + "step": 15529 + }, + { + "epoch": 1.9755756265106221, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.339475631713867, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8918541669845581, + "num_tokens": 602804245.0, + "step": 15530 + }, + { + "epoch": 1.9757028367892127, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.40516185760498, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.900650680065155, + "num_tokens": 602847169.0, + "step": 15531 + }, + { + "epoch": 1.9758300470678032, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.580352783203125e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.529458045959473, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8737379312515259, + "num_tokens": 602882740.0, + "step": 15532 + }, + { + "epoch": 1.9759572573463937, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.398836135864258, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8978516459465027, + "num_tokens": 602919218.0, + "step": 15533 + }, + { + "epoch": 1.9760844676249842, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.399028778076172, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8748660683631897, + "num_tokens": 602960741.0, + "step": 15534 + }, + { + "epoch": 1.9762116779035748, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.351252555847168, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8988227844238281, + "num_tokens": 602999350.0, + "step": 15535 + }, + { + "epoch": 1.976338888182165, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.395801544189453, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8806052207946777, + "num_tokens": 603041319.0, + "step": 15536 + }, + { + "epoch": 1.9764660984607556, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.449113845825195, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.886295735836029, + "num_tokens": 603077025.0, + "step": 15537 + }, + { + "epoch": 1.9765933087393461, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.501757621765137, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.887382984161377, + "num_tokens": 603110978.0, + "step": 15538 + }, + { + "epoch": 1.9767205190179367, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.431572914123535, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.898480236530304, + "num_tokens": 603145222.0, + "step": 15539 + }, + { + "epoch": 1.9768477292965272, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.392352104187012, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8925129175186157, + "num_tokens": 603183097.0, + "step": 15540 + }, + { + "epoch": 1.9769749395751177, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.458765029907227, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8930916786193848, + "num_tokens": 603223016.0, + "step": 15541 + }, + { + "epoch": 1.977102149853708, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.395771980285645, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8931312561035156, + "num_tokens": 603262710.0, + "step": 15542 + }, + { + "epoch": 1.9772293601322986, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.370890617370605, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9036943316459656, + "num_tokens": 603300692.0, + "step": 15543 + }, + { + "epoch": 1.977356570410889, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.3709135055542, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8901787996292114, + "num_tokens": 603341435.0, + "step": 15544 + }, + { + "epoch": 1.9774837806894796, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.470587730407715, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8901301026344299, + "num_tokens": 603374521.0, + "step": 15545 + }, + { + "epoch": 1.9776109909680701, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.396690368652344, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8877100348472595, + "num_tokens": 603416174.0, + "step": 15546 + }, + { + "epoch": 1.9777382012466607, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.410688400268555, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.878888726234436, + "num_tokens": 603456373.0, + "step": 15547 + }, + { + "epoch": 1.9778654115252512, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.356576919555664, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8806502223014832, + "num_tokens": 603502731.0, + "step": 15548 + }, + { + "epoch": 1.9779926218038417, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.378698348999023, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8814161419868469, + "num_tokens": 603542116.0, + "step": 15549 + }, + { + "epoch": 1.9781198320824323, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.403976440429688, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8909586668014526, + "num_tokens": 603581244.0, + "step": 15550 + }, + { + "epoch": 1.9782470423610228, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.41862964630127, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8749659657478333, + "num_tokens": 603617696.0, + "step": 15551 + }, + { + "epoch": 1.9783742526396133, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.374736785888672, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8937978744506836, + "num_tokens": 603660723.0, + "step": 15552 + }, + { + "epoch": 1.9785014629182038, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.394915580749512, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.885037899017334, + "num_tokens": 603699153.0, + "step": 15553 + }, + { + "epoch": 1.9786286731967944, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.545763969421387, + "learning_rate": 1e-06, + "loss": 0.4004, + "mean_token_accuracy": 0.8708266019821167, + "num_tokens": 603735096.0, + "step": 15554 + }, + { + "epoch": 1.978755883475385, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.35247802734375, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8811400532722473, + "num_tokens": 603780346.0, + "step": 15555 + }, + { + "epoch": 1.9788830937539754, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.451969146728516, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.8717359304428101, + "num_tokens": 603825634.0, + "step": 15556 + }, + { + "epoch": 1.979010304032566, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.447311401367188, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8885867595672607, + "num_tokens": 603859814.0, + "step": 15557 + }, + { + "epoch": 1.9791375143111565, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.37324333190918, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8982387781143188, + "num_tokens": 603896724.0, + "step": 15558 + }, + { + "epoch": 1.979264724589747, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.351988792419434, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8847136497497559, + "num_tokens": 603941897.0, + "step": 15559 + }, + { + "epoch": 1.9793919348683373, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.40691089630127, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8847149610519409, + "num_tokens": 603983709.0, + "step": 15560 + }, + { + "epoch": 1.9795191451469278, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.552980422973633, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.897124171257019, + "num_tokens": 604010952.0, + "step": 15561 + }, + { + "epoch": 1.9796463554255184, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.424094200134277, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8881366848945618, + "num_tokens": 604045412.0, + "step": 15562 + }, + { + "epoch": 1.979773565704109, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.316497802734375, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8864078521728516, + "num_tokens": 604085196.0, + "step": 15563 + }, + { + "epoch": 1.9799007759826994, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.497892379760742, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8828762769699097, + "num_tokens": 604124425.0, + "step": 15564 + }, + { + "epoch": 1.98002798626129, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.345771789550781, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8932433128356934, + "num_tokens": 604161725.0, + "step": 15565 + }, + { + "epoch": 1.9801551965398803, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.4827241897583, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8782070875167847, + "num_tokens": 604201603.0, + "step": 15566 + }, + { + "epoch": 1.9802824068184708, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.376824378967285, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8953917026519775, + "num_tokens": 604239084.0, + "step": 15567 + }, + { + "epoch": 1.9804096170970613, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.363274574279785, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.892392098903656, + "num_tokens": 604279162.0, + "step": 15568 + }, + { + "epoch": 1.9805368273756518, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.468461990356445, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8774126768112183, + "num_tokens": 604317913.0, + "step": 15569 + }, + { + "epoch": 1.9806640376542424, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.309544563293457, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8906142711639404, + "num_tokens": 604359622.0, + "step": 15570 + }, + { + "epoch": 1.980791247932833, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.410576820373535, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8963587880134583, + "num_tokens": 604397023.0, + "step": 15571 + }, + { + "epoch": 1.9809184582114234, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.393285751342773, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8915059566497803, + "num_tokens": 604439569.0, + "step": 15572 + }, + { + "epoch": 1.981045668490014, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.280592918395996, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8998009562492371, + "num_tokens": 604477009.0, + "step": 15573 + }, + { + "epoch": 1.9811728787686045, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.474832534790039, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8963412046432495, + "num_tokens": 604509262.0, + "step": 15574 + }, + { + "epoch": 1.981300089047195, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.304947853088379, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8889396786689758, + "num_tokens": 604556144.0, + "step": 15575 + }, + { + "epoch": 1.9814272993257855, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.385308265686035, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8864843845367432, + "num_tokens": 604594289.0, + "step": 15576 + }, + { + "epoch": 1.981554509604376, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.427949905395508, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8939087390899658, + "num_tokens": 604630210.0, + "step": 15577 + }, + { + "epoch": 1.9816817198829666, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.33248233795166, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.886894702911377, + "num_tokens": 604674035.0, + "step": 15578 + }, + { + "epoch": 1.9818089301615571, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.48185920715332, + "learning_rate": 1e-06, + "loss": 0.3885, + "mean_token_accuracy": 0.8706302642822266, + "num_tokens": 604720778.0, + "step": 15579 + }, + { + "epoch": 1.9819361404401477, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.39523696899414, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8998222351074219, + "num_tokens": 604756691.0, + "step": 15580 + }, + { + "epoch": 1.9820633507187382, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.339089393615723, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8945547342300415, + "num_tokens": 604794910.0, + "step": 15581 + }, + { + "epoch": 1.9821905609973287, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.462149620056152, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8831202983856201, + "num_tokens": 604836425.0, + "step": 15582 + }, + { + "epoch": 1.9823177712759192, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.511516571044922, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8859308958053589, + "num_tokens": 604866027.0, + "step": 15583 + }, + { + "epoch": 1.9824449815545098, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.362424850463867, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8813437223434448, + "num_tokens": 604913752.0, + "step": 15584 + }, + { + "epoch": 1.9825721918331, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.325028419494629, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8937559127807617, + "num_tokens": 604958617.0, + "step": 15585 + }, + { + "epoch": 1.9826994021116906, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.433747291564941, + "learning_rate": 1e-06, + "loss": 0.3901, + "mean_token_accuracy": 0.8698153495788574, + "num_tokens": 605001526.0, + "step": 15586 + }, + { + "epoch": 1.9828266123902811, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.50419807434082, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8883665800094604, + "num_tokens": 605039093.0, + "step": 15587 + }, + { + "epoch": 1.9829538226688717, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.307116508483887, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.9011004567146301, + "num_tokens": 605076229.0, + "step": 15588 + }, + { + "epoch": 1.9830810329474622, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.49485969543457, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.893964946269989, + "num_tokens": 605117556.0, + "step": 15589 + }, + { + "epoch": 1.9832082432260527, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.443368911743164, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.8728074431419373, + "num_tokens": 605158761.0, + "step": 15590 + }, + { + "epoch": 1.983335453504643, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.46327018737793, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8805047273635864, + "num_tokens": 605198463.0, + "step": 15591 + }, + { + "epoch": 1.9834626637832335, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.422951698303223, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8846286535263062, + "num_tokens": 605233509.0, + "step": 15592 + }, + { + "epoch": 1.983589874061824, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.430456161499023, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8942500948905945, + "num_tokens": 605268093.0, + "step": 15593 + }, + { + "epoch": 1.9837170843404146, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.282522201538086, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9012766480445862, + "num_tokens": 605304940.0, + "step": 15594 + }, + { + "epoch": 1.9838442946190051, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.441108703613281, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8697863817214966, + "num_tokens": 605343919.0, + "step": 15595 + }, + { + "epoch": 1.9839715048975957, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.426973342895508, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8926840424537659, + "num_tokens": 605375088.0, + "step": 15596 + }, + { + "epoch": 1.9840987151761862, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.379197120666504, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8909386396408081, + "num_tokens": 605421759.0, + "step": 15597 + }, + { + "epoch": 1.9842259254547767, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.376884460449219, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8765315413475037, + "num_tokens": 605464606.0, + "step": 15598 + }, + { + "epoch": 1.9843531357333672, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.455543518066406, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.882185697555542, + "num_tokens": 605499210.0, + "step": 15599 + }, + { + "epoch": 1.9844803460119578, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.374011039733887, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8957077264785767, + "num_tokens": 605538006.0, + "step": 15600 + }, + { + "epoch": 1.9846075562905483, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.32989501953125, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.891355037689209, + "num_tokens": 605579835.0, + "step": 15601 + }, + { + "epoch": 1.9847347665691388, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.401005744934082, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8911805152893066, + "num_tokens": 605618646.0, + "step": 15602 + }, + { + "epoch": 1.9848619768477294, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.447524070739746, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8908829092979431, + "num_tokens": 605653517.0, + "step": 15603 + }, + { + "epoch": 1.9849891871263199, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.42605209350586, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.889426589012146, + "num_tokens": 605687586.0, + "step": 15604 + }, + { + "epoch": 1.9851163974049104, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.46424388885498, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8896776437759399, + "num_tokens": 605725171.0, + "step": 15605 + }, + { + "epoch": 1.985243607683501, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.377182960510254, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8976254463195801, + "num_tokens": 605765827.0, + "step": 15606 + }, + { + "epoch": 1.9853708179620915, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.452542304992676, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8865230679512024, + "num_tokens": 605801765.0, + "step": 15607 + }, + { + "epoch": 1.985498028240682, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.465229988098145, + "learning_rate": 1e-06, + "loss": 0.3664, + "mean_token_accuracy": 0.8794171810150146, + "num_tokens": 605840101.0, + "step": 15608 + }, + { + "epoch": 1.9856252385192723, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.360620498657227, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.895455002784729, + "num_tokens": 605877814.0, + "step": 15609 + }, + { + "epoch": 1.9857524487978628, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.39931583404541, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8967625498771667, + "num_tokens": 605913145.0, + "step": 15610 + }, + { + "epoch": 1.9858796590764534, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.419900894165039, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8915046453475952, + "num_tokens": 605947719.0, + "step": 15611 + }, + { + "epoch": 1.986006869355044, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.433775901794434, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8844430446624756, + "num_tokens": 605991266.0, + "step": 15612 + }, + { + "epoch": 1.9861340796336344, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.334251403808594, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8886522054672241, + "num_tokens": 606031548.0, + "step": 15613 + }, + { + "epoch": 1.986261289912225, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.406235694885254, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8925813436508179, + "num_tokens": 606074754.0, + "step": 15614 + }, + { + "epoch": 1.9863885001908153, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.541836738586426, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8916034698486328, + "num_tokens": 606106917.0, + "step": 15615 + }, + { + "epoch": 1.9865157104694058, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.357906341552734, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.9044222831726074, + "num_tokens": 606140940.0, + "step": 15616 + }, + { + "epoch": 1.9866429207479963, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.413941383361816, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.894708514213562, + "num_tokens": 606175539.0, + "step": 15617 + }, + { + "epoch": 1.9867701310265868, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.346587181091309, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8854770660400391, + "num_tokens": 606218803.0, + "step": 15618 + }, + { + "epoch": 1.9868973413051774, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.355554580688477, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8837466239929199, + "num_tokens": 606265478.0, + "step": 15619 + }, + { + "epoch": 1.987024551583768, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.432321548461914, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.872878909111023, + "num_tokens": 606309137.0, + "step": 15620 + }, + { + "epoch": 1.9871517618623584, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.319571495056152, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8903759717941284, + "num_tokens": 606352367.0, + "step": 15621 + }, + { + "epoch": 1.987278972140949, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.317426681518555, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8828796148300171, + "num_tokens": 606401698.0, + "step": 15622 + }, + { + "epoch": 1.9874061824195395, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.462461471557617, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8847029209136963, + "num_tokens": 606438472.0, + "step": 15623 + }, + { + "epoch": 1.98753339269813, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.430889129638672, + "learning_rate": 1e-06, + "loss": 0.3667, + "mean_token_accuracy": 0.8753538131713867, + "num_tokens": 606475746.0, + "step": 15624 + }, + { + "epoch": 1.9876606029767205, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.419576644897461, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8818010091781616, + "num_tokens": 606514096.0, + "step": 15625 + }, + { + "epoch": 1.987787813255311, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.380929946899414, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8948400020599365, + "num_tokens": 606554493.0, + "step": 15626 + }, + { + "epoch": 1.9879150235339016, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.35212230682373, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8997830748558044, + "num_tokens": 606594217.0, + "step": 15627 + }, + { + "epoch": 1.9880422338124921, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.414918899536133, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8920901417732239, + "num_tokens": 606633957.0, + "step": 15628 + }, + { + "epoch": 1.9881694440910826, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.479433059692383, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.8719342947006226, + "num_tokens": 606674448.0, + "step": 15629 + }, + { + "epoch": 1.9882966543696732, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.3361234664917, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8957021236419678, + "num_tokens": 606722939.0, + "step": 15630 + }, + { + "epoch": 1.9884238646482637, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.439027786254883, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.895056962966919, + "num_tokens": 606763348.0, + "step": 15631 + }, + { + "epoch": 1.9885510749268542, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.351964950561523, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8973168730735779, + "num_tokens": 606802410.0, + "step": 15632 + }, + { + "epoch": 1.9886782852054448, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.411882400512695, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8891302943229675, + "num_tokens": 606843295.0, + "step": 15633 + }, + { + "epoch": 1.988805495484035, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.365194320678711, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8923475742340088, + "num_tokens": 606882026.0, + "step": 15634 + }, + { + "epoch": 1.9889327057626256, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.354621887207031, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8955769538879395, + "num_tokens": 606921045.0, + "step": 15635 + }, + { + "epoch": 1.9890599160412161, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.433197975158691, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8825121521949768, + "num_tokens": 606964718.0, + "step": 15636 + }, + { + "epoch": 1.9891871263198067, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.37755012512207, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8774145841598511, + "num_tokens": 607001878.0, + "step": 15637 + }, + { + "epoch": 1.9893143365983972, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.426613807678223, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8869754672050476, + "num_tokens": 607042686.0, + "step": 15638 + }, + { + "epoch": 1.9894415468769877, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.422128677368164, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8950684070587158, + "num_tokens": 607075527.0, + "step": 15639 + }, + { + "epoch": 1.989568757155578, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.385380744934082, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.9064220190048218, + "num_tokens": 607106693.0, + "step": 15640 + }, + { + "epoch": 1.9896959674341685, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.39844036102295, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.9032280445098877, + "num_tokens": 607143213.0, + "step": 15641 + }, + { + "epoch": 1.989823177712759, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.39454460144043, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8837348222732544, + "num_tokens": 607185165.0, + "step": 15642 + }, + { + "epoch": 1.9899503879913496, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.425436973571777, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8899784684181213, + "num_tokens": 607222261.0, + "step": 15643 + }, + { + "epoch": 1.9900775982699401, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.40285873413086, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8921409249305725, + "num_tokens": 607254949.0, + "step": 15644 + }, + { + "epoch": 1.9902048085485307, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.397268295288086, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8998857140541077, + "num_tokens": 607288829.0, + "step": 15645 + }, + { + "epoch": 1.9903320188271212, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.67572021484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.330694198608398, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.884765625, + "num_tokens": 607334213.0, + "step": 15646 + }, + { + "epoch": 1.9904592291057117, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.48525333404541, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.8744765520095825, + "num_tokens": 607373349.0, + "step": 15647 + }, + { + "epoch": 1.9905864393843022, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.324076652526855, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8914986848831177, + "num_tokens": 607414152.0, + "step": 15648 + }, + { + "epoch": 1.9907136496628928, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.395402908325195, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8971152901649475, + "num_tokens": 607455738.0, + "step": 15649 + }, + { + "epoch": 1.9908408599414833, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.375753402709961, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8831377029418945, + "num_tokens": 607494037.0, + "step": 15650 + }, + { + "epoch": 1.9909680702200738, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.383130073547363, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8881333470344543, + "num_tokens": 607530258.0, + "step": 15651 + }, + { + "epoch": 1.9910952804986644, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.37677001953125, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8984996676445007, + "num_tokens": 607573785.0, + "step": 15652 + }, + { + "epoch": 1.9912224907772549, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.335945129394531, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8838940858840942, + "num_tokens": 607618505.0, + "step": 15653 + }, + { + "epoch": 1.9913497010558454, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.483209609985352, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8946095108985901, + "num_tokens": 607654974.0, + "step": 15654 + }, + { + "epoch": 1.991476911334436, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.361196517944336, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8977952599525452, + "num_tokens": 607691655.0, + "step": 15655 + }, + { + "epoch": 1.9916041216130265, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.4638090133667, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.883100152015686, + "num_tokens": 607726173.0, + "step": 15656 + }, + { + "epoch": 1.991731331891617, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.423882484436035, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8847540020942688, + "num_tokens": 607767049.0, + "step": 15657 + }, + { + "epoch": 1.9918585421702073, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.31092357635498, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8995547890663147, + "num_tokens": 607805179.0, + "step": 15658 + }, + { + "epoch": 1.9919857524487978, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.394207954406738, + "learning_rate": 1e-06, + "loss": 0.3796, + "mean_token_accuracy": 0.874559760093689, + "num_tokens": 607850093.0, + "step": 15659 + }, + { + "epoch": 1.9921129627273884, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.406976699829102, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8857548236846924, + "num_tokens": 607884525.0, + "step": 15660 + }, + { + "epoch": 1.9922401730059789, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.334266662597656, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8778887391090393, + "num_tokens": 607929760.0, + "step": 15661 + }, + { + "epoch": 1.9923673832845694, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.42398738861084, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8709076642990112, + "num_tokens": 607972604.0, + "step": 15662 + }, + { + "epoch": 1.99249459356316, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.379240989685059, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8857474327087402, + "num_tokens": 608011263.0, + "step": 15663 + }, + { + "epoch": 1.9926218038417502, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.426466941833496, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8737909197807312, + "num_tokens": 608052583.0, + "step": 15664 + }, + { + "epoch": 1.9927490141203408, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.400310516357422, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8991419672966003, + "num_tokens": 608087538.0, + "step": 15665 + }, + { + "epoch": 1.9928762243989313, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.344085693359375, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8844671249389648, + "num_tokens": 608130511.0, + "step": 15666 + }, + { + "epoch": 1.9930034346775218, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.368768692016602, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8947556614875793, + "num_tokens": 608173755.0, + "step": 15667 + }, + { + "epoch": 1.9931306449561124, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.313450813293457, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8901337385177612, + "num_tokens": 608216828.0, + "step": 15668 + }, + { + "epoch": 1.993257855234703, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.476872444152832, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8842297196388245, + "num_tokens": 608254252.0, + "step": 15669 + }, + { + "epoch": 1.9933850655132934, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.434172630310059, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8764368891716003, + "num_tokens": 608293094.0, + "step": 15670 + }, + { + "epoch": 1.993512275791884, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.410421371459961, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8920396566390991, + "num_tokens": 608327505.0, + "step": 15671 + }, + { + "epoch": 1.9936394860704745, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.399788856506348, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8824231028556824, + "num_tokens": 608364423.0, + "step": 15672 + }, + { + "epoch": 1.993766696349065, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.383169174194336, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8911817073822021, + "num_tokens": 608403191.0, + "step": 15673 + }, + { + "epoch": 1.9938939066276555, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.41372299194336, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8880847692489624, + "num_tokens": 608441952.0, + "step": 15674 + }, + { + "epoch": 1.994021116906246, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.334491729736328, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.9026281833648682, + "num_tokens": 608479034.0, + "step": 15675 + }, + { + "epoch": 1.9941483271848366, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.447918891906738, + "learning_rate": 1e-06, + "loss": 0.3948, + "mean_token_accuracy": 0.8668715953826904, + "num_tokens": 608518499.0, + "step": 15676 + }, + { + "epoch": 1.9942755374634271, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.31558609008789, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.886986255645752, + "num_tokens": 608562552.0, + "step": 15677 + }, + { + "epoch": 1.9944027477420176, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.434344291687012, + "learning_rate": 1e-06, + "loss": 0.3928, + "mean_token_accuracy": 0.8688057661056519, + "num_tokens": 608601803.0, + "step": 15678 + }, + { + "epoch": 1.9945299580206082, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.432271957397461, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8844559788703918, + "num_tokens": 608640224.0, + "step": 15679 + }, + { + "epoch": 1.9946571682991987, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.451292037963867, + "learning_rate": 1e-06, + "loss": 0.3729, + "mean_token_accuracy": 0.8753447532653809, + "num_tokens": 608675303.0, + "step": 15680 + }, + { + "epoch": 1.9947843785777892, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.372114181518555, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8887437582015991, + "num_tokens": 608717946.0, + "step": 15681 + }, + { + "epoch": 1.9949115888563798, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.349730491638184, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8806766271591187, + "num_tokens": 608762458.0, + "step": 15682 + }, + { + "epoch": 1.99503879913497, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.378521919250488, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8885708451271057, + "num_tokens": 608803929.0, + "step": 15683 + }, + { + "epoch": 1.9951660094135606, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.405296325683594, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8922785520553589, + "num_tokens": 608845223.0, + "step": 15684 + }, + { + "epoch": 1.9952932196921511, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.359980583190918, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8835869431495667, + "num_tokens": 608887831.0, + "step": 15685 + }, + { + "epoch": 1.9954204299707416, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.434625625610352, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8868280649185181, + "num_tokens": 608931028.0, + "step": 15686 + }, + { + "epoch": 1.9955476402493322, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.388371467590332, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8769633769989014, + "num_tokens": 608972646.0, + "step": 15687 + }, + { + "epoch": 1.9956748505279227, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.424535751342773, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8779604434967041, + "num_tokens": 609015521.0, + "step": 15688 + }, + { + "epoch": 1.995802060806513, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.420619010925293, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8871488571166992, + "num_tokens": 609052814.0, + "step": 15689 + }, + { + "epoch": 1.9959292710851035, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.40921688079834, + "learning_rate": 1e-06, + "loss": 0.3714, + "mean_token_accuracy": 0.8757624626159668, + "num_tokens": 609094046.0, + "step": 15690 + }, + { + "epoch": 1.996056481363694, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.403383255004883, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8793258666992188, + "num_tokens": 609132221.0, + "step": 15691 + }, + { + "epoch": 1.9961836916422846, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.458611488342285, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8858069181442261, + "num_tokens": 609168379.0, + "step": 15692 + }, + { + "epoch": 1.9963109019208751, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.4440336227417, + "learning_rate": 1e-06, + "loss": 0.3775, + "mean_token_accuracy": 0.8703229427337646, + "num_tokens": 609205449.0, + "step": 15693 + }, + { + "epoch": 1.9964381121994657, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.334037780761719, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8864743709564209, + "num_tokens": 609246322.0, + "step": 15694 + }, + { + "epoch": 1.9965653224780562, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.455915451049805, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8943602442741394, + "num_tokens": 609284921.0, + "step": 15695 + }, + { + "epoch": 1.9966925327566467, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.482538223266602, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.88249671459198, + "num_tokens": 609315568.0, + "step": 15696 + }, + { + "epoch": 1.9968197430352372, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.326004981994629, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8959102630615234, + "num_tokens": 609359160.0, + "step": 15697 + }, + { + "epoch": 1.9969469533138278, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.495699882507324, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8874044418334961, + "num_tokens": 609391822.0, + "step": 15698 + }, + { + "epoch": 1.9970741635924183, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.495153427124023, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8796013593673706, + "num_tokens": 609425589.0, + "step": 15699 + }, + { + "epoch": 1.9972013738710088, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.335930824279785, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8977709412574768, + "num_tokens": 609465316.0, + "step": 15700 + }, + { + "epoch": 1.9973285841495994, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.441253662109375, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.897928774356842, + "num_tokens": 609505913.0, + "step": 15701 + }, + { + "epoch": 1.9974557944281899, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.430373191833496, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8890377879142761, + "num_tokens": 609546732.0, + "step": 15702 + }, + { + "epoch": 1.9975830047067804, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.438508033752441, + "learning_rate": 1e-06, + "loss": 0.3736, + "mean_token_accuracy": 0.8769356608390808, + "num_tokens": 609589585.0, + "step": 15703 + }, + { + "epoch": 1.997710214985371, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.504316329956055, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8720982670783997, + "num_tokens": 609625468.0, + "step": 15704 + }, + { + "epoch": 1.9978374252639615, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.321900367736816, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8987647294998169, + "num_tokens": 609665361.0, + "step": 15705 + }, + { + "epoch": 1.997964635542552, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.416461944580078, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.889252245426178, + "num_tokens": 609707269.0, + "step": 15706 + }, + { + "epoch": 1.9980918458211423, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.432084083557129, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8759202361106873, + "num_tokens": 609749040.0, + "step": 15707 + }, + { + "epoch": 1.9982190560997328, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.420052528381348, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8911298513412476, + "num_tokens": 609787871.0, + "step": 15708 + }, + { + "epoch": 1.9983462663783234, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.445029258728027, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8825875520706177, + "num_tokens": 609823089.0, + "step": 15709 + }, + { + "epoch": 1.9984734766569139, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.407088279724121, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8842567205429077, + "num_tokens": 609864274.0, + "step": 15710 + }, + { + "epoch": 1.9986006869355044, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.358353614807129, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.896292507648468, + "num_tokens": 609904193.0, + "step": 15711 + }, + { + "epoch": 1.998727897214095, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.530082702636719, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8793441653251648, + "num_tokens": 609937559.0, + "step": 15712 + }, + { + "epoch": 1.9988551074926852, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.390421867370605, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8980862498283386, + "num_tokens": 609975168.0, + "step": 15713 + }, + { + "epoch": 1.9989823177712758, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.355865478515625, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8965421915054321, + "num_tokens": 610017370.0, + "step": 15714 + }, + { + "epoch": 1.9991095280498663, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.499516487121582, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8881251215934753, + "num_tokens": 610051325.0, + "step": 15715 + }, + { + "epoch": 1.9992367383284568, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.47037410736084, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8871191143989563, + "num_tokens": 610088857.0, + "step": 15716 + }, + { + "epoch": 1.9993639486070474, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.331053733825684, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.9000200629234314, + "num_tokens": 610129657.0, + "step": 15717 + }, + { + "epoch": 1.9994911588856379, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.45580005645752, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.899367094039917, + "num_tokens": 610164932.0, + "step": 15718 + }, + { + "epoch": 1.9996183691642284, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.469364166259766, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8717662692070007, + "num_tokens": 610204264.0, + "step": 15719 + }, + { + "epoch": 1.999745579442819, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.463301658630371, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.886620044708252, + "num_tokens": 610241303.0, + "step": 15720 + }, + { + "epoch": 1.9998727897214095, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.423998832702637, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8783093094825745, + "num_tokens": 610280933.0, + "step": 15721 + }, + { + "epoch": 2.0, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.37667465209961, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8974926471710205, + "num_tokens": 610320398.0, + "step": 15722 + }, + { + "epoch": 2.0001272102785905, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.368072509765625, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.9030537009239197, + "num_tokens": 610362092.0, + "step": 15723 + }, + { + "epoch": 2.000254420557181, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.351617813110352, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8984602093696594, + "num_tokens": 610402257.0, + "step": 15724 + }, + { + "epoch": 2.0003816308357716, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.33275318145752, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.9040716290473938, + "num_tokens": 610443455.0, + "step": 15725 + }, + { + "epoch": 2.000508841114362, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.424238204956055, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.902227520942688, + "num_tokens": 610480461.0, + "step": 15726 + }, + { + "epoch": 2.0006360513929526, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.382246971130371, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8961001634597778, + "num_tokens": 610518905.0, + "step": 15727 + }, + { + "epoch": 2.000763261671543, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.420352935791016, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9033461213111877, + "num_tokens": 610560745.0, + "step": 15728 + }, + { + "epoch": 2.0008904719501337, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.408624649047852, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8973479866981506, + "num_tokens": 610604598.0, + "step": 15729 + }, + { + "epoch": 2.0010176822287242, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.471830368041992, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9054034948348999, + "num_tokens": 610641830.0, + "step": 15730 + }, + { + "epoch": 2.0011448925073148, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.38111686706543, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9017704129219055, + "num_tokens": 610685970.0, + "step": 15731 + }, + { + "epoch": 2.0012721027859053, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0967254638671875e-05, + "grad_norm": 9.428299903869629, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8980081677436829, + "num_tokens": 610728066.0, + "step": 15732 + }, + { + "epoch": 2.001399313064496, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.455484390258789, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.904457688331604, + "num_tokens": 610766735.0, + "step": 15733 + }, + { + "epoch": 2.0015265233430863, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.471627235412598, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8945838212966919, + "num_tokens": 610806423.0, + "step": 15734 + }, + { + "epoch": 2.0016537336216764, + "ewc_loss": 0.017578125, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0848045349121094e-05, + "grad_norm": 9.460083961486816, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8881148099899292, + "num_tokens": 610847472.0, + "step": 15735 + }, + { + "epoch": 2.001780943900267, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.398523330688477, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9049952030181885, + "num_tokens": 610890666.0, + "step": 15736 + }, + { + "epoch": 2.0019081541788575, + "ewc_loss": 0.0177001953125, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.0907649993896484e-05, + "grad_norm": 9.522004127502441, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8982061147689819, + "num_tokens": 610928515.0, + "step": 15737 + }, + { + "epoch": 2.002035364457448, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.446850776672363, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8967169523239136, + "num_tokens": 610968114.0, + "step": 15738 + }, + { + "epoch": 2.0021625747360385, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.462547302246094, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9066805243492126, + "num_tokens": 611002540.0, + "step": 15739 + }, + { + "epoch": 2.002289785014629, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.60705280303955, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9037749767303467, + "num_tokens": 611031903.0, + "step": 15740 + }, + { + "epoch": 2.0024169952932196, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.395698547363281, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9010671377182007, + "num_tokens": 611070094.0, + "step": 15741 + }, + { + "epoch": 2.00254420557181, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1086463928222656e-05, + "grad_norm": 9.50594711303711, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8914284706115723, + "num_tokens": 611106436.0, + "step": 15742 + }, + { + "epoch": 2.0026714158504006, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.356511116027832, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9020792245864868, + "num_tokens": 611149467.0, + "step": 15743 + }, + { + "epoch": 2.002798626128991, + "ewc_loss": 0.017822265625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1026859283447266e-05, + "grad_norm": 9.451122283935547, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.9080018997192383, + "num_tokens": 611188043.0, + "step": 15744 + }, + { + "epoch": 2.0029258364075817, + "ewc_loss": 0.0179443359375, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1146068572998047e-05, + "grad_norm": 9.39601993560791, + "learning_rate": 1e-06, + "loss": 0.2461, + "mean_token_accuracy": 0.9163702726364136, + "num_tokens": 611227777.0, + "step": 15745 + }, + { + "epoch": 2.0030530466861722, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.564396858215332, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8868617415428162, + "num_tokens": 611264694.0, + "step": 15746 + }, + { + "epoch": 2.0031802569647628, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.467422485351562, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8928052186965942, + "num_tokens": 611305134.0, + "step": 15747 + }, + { + "epoch": 2.0033074672433533, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.414748191833496, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8983361721038818, + "num_tokens": 611346503.0, + "step": 15748 + }, + { + "epoch": 2.003434677521944, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.453363418579102, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.897334098815918, + "num_tokens": 611388176.0, + "step": 15749 + }, + { + "epoch": 2.0035618878005343, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.426993370056152, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9104291796684265, + "num_tokens": 611429071.0, + "step": 15750 + }, + { + "epoch": 2.003689098079125, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.444056510925293, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8905601501464844, + "num_tokens": 611470027.0, + "step": 15751 + }, + { + "epoch": 2.0038163083577154, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.408342361450195, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.896560549736023, + "num_tokens": 611509512.0, + "step": 15752 + }, + { + "epoch": 2.003943518636306, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.470129013061523, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.9044939875602722, + "num_tokens": 611547330.0, + "step": 15753 + }, + { + "epoch": 2.0040707289148965, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.473843574523926, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8976500034332275, + "num_tokens": 611586055.0, + "step": 15754 + }, + { + "epoch": 2.004197939193487, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.4562349319458, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.898169755935669, + "num_tokens": 611622675.0, + "step": 15755 + }, + { + "epoch": 2.0043251494720775, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.448391914367676, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8998646140098572, + "num_tokens": 611664440.0, + "step": 15756 + }, + { + "epoch": 2.004452359750668, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.456029891967773, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.9053634405136108, + "num_tokens": 611703789.0, + "step": 15757 + }, + { + "epoch": 2.0045795700292586, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.52795124053955, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.9003559947013855, + "num_tokens": 611739584.0, + "step": 15758 + }, + { + "epoch": 2.0047067803078487, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.535771369934082, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8973080515861511, + "num_tokens": 611780118.0, + "step": 15759 + }, + { + "epoch": 2.004833990586439, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.402572631835938, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8990886211395264, + "num_tokens": 611825977.0, + "step": 15760 + }, + { + "epoch": 2.0049612008650297, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.458171844482422, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8994223475456238, + "num_tokens": 611867476.0, + "step": 15761 + }, + { + "epoch": 2.0050884111436202, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.50368595123291, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8936417102813721, + "num_tokens": 611901285.0, + "step": 15762 + }, + { + "epoch": 2.0052156214222108, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.40388298034668, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.9027603268623352, + "num_tokens": 611941754.0, + "step": 15763 + }, + { + "epoch": 2.0053428317008013, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.477325439453125, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9082597494125366, + "num_tokens": 611980646.0, + "step": 15764 + }, + { + "epoch": 2.005470041979392, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.426389694213867, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.90981125831604, + "num_tokens": 612020251.0, + "step": 15765 + }, + { + "epoch": 2.0055972522579824, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.500618934631348, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8935478925704956, + "num_tokens": 612055906.0, + "step": 15766 + }, + { + "epoch": 2.005724462536573, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.504027366638184, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9041368961334229, + "num_tokens": 612092323.0, + "step": 15767 + }, + { + "epoch": 2.0058516728151634, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.426259994506836, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.915029764175415, + "num_tokens": 612127822.0, + "step": 15768 + }, + { + "epoch": 2.005978883093754, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.412631034851074, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8885549902915955, + "num_tokens": 612170241.0, + "step": 15769 + }, + { + "epoch": 2.0061060933723445, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.455769538879395, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9050486087799072, + "num_tokens": 612207634.0, + "step": 15770 + }, + { + "epoch": 2.006233303650935, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.539658546447754, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9022379517555237, + "num_tokens": 612243122.0, + "step": 15771 + }, + { + "epoch": 2.0063605139295255, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.44472885131836, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9072476625442505, + "num_tokens": 612281439.0, + "step": 15772 + }, + { + "epoch": 2.006487724208116, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.429831504821777, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8951309323310852, + "num_tokens": 612323580.0, + "step": 15773 + }, + { + "epoch": 2.0066149344867066, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.415363311767578, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9122937917709351, + "num_tokens": 612361589.0, + "step": 15774 + }, + { + "epoch": 2.006742144765297, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.401978492736816, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.9028534889221191, + "num_tokens": 612403025.0, + "step": 15775 + }, + { + "epoch": 2.0068693550438876, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.392672538757324, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.903997540473938, + "num_tokens": 612442057.0, + "step": 15776 + }, + { + "epoch": 2.006996565322478, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.45296573638916, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.9011789560317993, + "num_tokens": 612486772.0, + "step": 15777 + }, + { + "epoch": 2.0071237756010687, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.426342010498047, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9152385592460632, + "num_tokens": 612521040.0, + "step": 15778 + }, + { + "epoch": 2.007250985879659, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.446303367614746, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.9023633003234863, + "num_tokens": 612562248.0, + "step": 15779 + }, + { + "epoch": 2.0073781961582498, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.424946784973145, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8982692956924438, + "num_tokens": 612602641.0, + "step": 15780 + }, + { + "epoch": 2.0075054064368403, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.557339668273926, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8857868909835815, + "num_tokens": 612639262.0, + "step": 15781 + }, + { + "epoch": 2.007632616715431, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.456506729125977, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.9036374092102051, + "num_tokens": 612678379.0, + "step": 15782 + }, + { + "epoch": 2.0077598269940213, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.553976058959961, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8979591131210327, + "num_tokens": 612713023.0, + "step": 15783 + }, + { + "epoch": 2.0078870372726114, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.39319896697998, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9043485522270203, + "num_tokens": 612754175.0, + "step": 15784 + }, + { + "epoch": 2.008014247551202, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1205673217773438e-05, + "grad_norm": 9.45687198638916, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8944742679595947, + "num_tokens": 612794131.0, + "step": 15785 + }, + { + "epoch": 2.0081414578297925, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.452067375183105, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.9035778045654297, + "num_tokens": 612832959.0, + "step": 15786 + }, + { + "epoch": 2.008268668108383, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.560125350952148, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.9004868268966675, + "num_tokens": 612866645.0, + "step": 15787 + }, + { + "epoch": 2.0083958783869735, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.383047103881836, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9029516577720642, + "num_tokens": 612907143.0, + "step": 15788 + }, + { + "epoch": 2.008523088665564, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.482808113098145, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8953132629394531, + "num_tokens": 612946183.0, + "step": 15789 + }, + { + "epoch": 2.0086502989441546, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.385516166687012, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9139684438705444, + "num_tokens": 612989932.0, + "step": 15790 + }, + { + "epoch": 2.008777509222745, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.461318969726562, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9082479476928711, + "num_tokens": 613027693.0, + "step": 15791 + }, + { + "epoch": 2.0089047195013356, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.460500717163086, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9076751470565796, + "num_tokens": 613062849.0, + "step": 15792 + }, + { + "epoch": 2.009031929779926, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.400425910949707, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9065643548965454, + "num_tokens": 613105165.0, + "step": 15793 + }, + { + "epoch": 2.0091591400585167, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.479432106018066, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8978832960128784, + "num_tokens": 613147155.0, + "step": 15794 + }, + { + "epoch": 2.0092863503371072, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.560785293579102, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9022152423858643, + "num_tokens": 613183747.0, + "step": 15795 + }, + { + "epoch": 2.0094135606156978, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.493424415588379, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8967655897140503, + "num_tokens": 613220735.0, + "step": 15796 + }, + { + "epoch": 2.0095407708942883, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.447822570800781, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9091375470161438, + "num_tokens": 613256994.0, + "step": 15797 + }, + { + "epoch": 2.009667981172879, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.493124961853027, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.9027930498123169, + "num_tokens": 613296093.0, + "step": 15798 + }, + { + "epoch": 2.0097951914514693, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.597347259521484, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8873247504234314, + "num_tokens": 613330042.0, + "step": 15799 + }, + { + "epoch": 2.00992240173006, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.477441787719727, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.9021381139755249, + "num_tokens": 613369422.0, + "step": 15800 + }, + { + "epoch": 2.0100496120086504, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1265277862548828e-05, + "grad_norm": 9.361084938049316, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8962465524673462, + "num_tokens": 613412607.0, + "step": 15801 + }, + { + "epoch": 2.010176822287241, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.453845977783203, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.9015383720397949, + "num_tokens": 613449619.0, + "step": 15802 + }, + { + "epoch": 2.0103040325658315, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.61339282989502, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.899391233921051, + "num_tokens": 613492188.0, + "step": 15803 + }, + { + "epoch": 2.010431242844422, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.422940254211426, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9059003591537476, + "num_tokens": 613534638.0, + "step": 15804 + }, + { + "epoch": 2.0105584531230125, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.48129940032959, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9091768264770508, + "num_tokens": 613566892.0, + "step": 15805 + }, + { + "epoch": 2.010685663401603, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.427406311035156, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.9000595808029175, + "num_tokens": 613606156.0, + "step": 15806 + }, + { + "epoch": 2.0108128736801936, + "ewc_loss": 0.01806640625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.416748046875, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8999505043029785, + "num_tokens": 613644996.0, + "step": 15807 + }, + { + "epoch": 2.0109400839587837, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.42740535736084, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.89029860496521, + "num_tokens": 613689276.0, + "step": 15808 + }, + { + "epoch": 2.011067294237374, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.481954574584961, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8902819156646729, + "num_tokens": 613726358.0, + "step": 15809 + }, + { + "epoch": 2.0111945045159647, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.515416145324707, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.9031569361686707, + "num_tokens": 613763733.0, + "step": 15810 + }, + { + "epoch": 2.0113217147945552, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.598675727844238, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.9021575450897217, + "num_tokens": 613791276.0, + "step": 15811 + }, + { + "epoch": 2.0114489250731458, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.491436004638672, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8800157904624939, + "num_tokens": 613830479.0, + "step": 15812 + }, + { + "epoch": 2.0115761353517363, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.771087646484375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.523112297058105, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8853533267974854, + "num_tokens": 613871896.0, + "step": 15813 + }, + { + "epoch": 2.011703345630327, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.574965476989746, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8838608264923096, + "num_tokens": 613904852.0, + "step": 15814 + }, + { + "epoch": 2.0118305559089174, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.480855941772461, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.9046145677566528, + "num_tokens": 613937834.0, + "step": 15815 + }, + { + "epoch": 2.011957766187508, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.333897590637207, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9040639400482178, + "num_tokens": 613979974.0, + "step": 15816 + }, + { + "epoch": 2.0120849764660984, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.649945259094238, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8996920585632324, + "num_tokens": 614008327.0, + "step": 15817 + }, + { + "epoch": 2.012212186744689, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.42204475402832, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8993675708770752, + "num_tokens": 614045075.0, + "step": 15818 + }, + { + "epoch": 2.0123393970232795, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.46274471282959, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8941107988357544, + "num_tokens": 614080413.0, + "step": 15819 + }, + { + "epoch": 2.01246660730187, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.537517547607422, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.907955527305603, + "num_tokens": 614114909.0, + "step": 15820 + }, + { + "epoch": 2.0125938175804605, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.478520393371582, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8874680399894714, + "num_tokens": 614153802.0, + "step": 15821 + }, + { + "epoch": 2.012721027859051, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.460437774658203, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8972403407096863, + "num_tokens": 614192073.0, + "step": 15822 + }, + { + "epoch": 2.0128482381376416, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.434598922729492, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9080722332000732, + "num_tokens": 614229937.0, + "step": 15823 + }, + { + "epoch": 2.012975448416232, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.463658332824707, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8886287212371826, + "num_tokens": 614271636.0, + "step": 15824 + }, + { + "epoch": 2.0131026586948226, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.47166633605957, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.900660514831543, + "num_tokens": 614310461.0, + "step": 15825 + }, + { + "epoch": 2.013229868973413, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.573500633239746, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8914626836776733, + "num_tokens": 614345317.0, + "step": 15826 + }, + { + "epoch": 2.0133570792520037, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.416388511657715, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9062932729721069, + "num_tokens": 614385389.0, + "step": 15827 + }, + { + "epoch": 2.013484289530594, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.492779731750488, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.891950249671936, + "num_tokens": 614424772.0, + "step": 15828 + }, + { + "epoch": 2.0136114998091847, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.458715438842773, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8915244936943054, + "num_tokens": 614467371.0, + "step": 15829 + }, + { + "epoch": 2.0137387100877753, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.51963996887207, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8966484665870667, + "num_tokens": 614508942.0, + "step": 15830 + }, + { + "epoch": 2.013865920366366, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.490236282348633, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8986262679100037, + "num_tokens": 614545563.0, + "step": 15831 + }, + { + "epoch": 2.0139931306449563, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.494180679321289, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9048161506652832, + "num_tokens": 614581901.0, + "step": 15832 + }, + { + "epoch": 2.0141203409235464, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.410866737365723, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8973112106323242, + "num_tokens": 614627193.0, + "step": 15833 + }, + { + "epoch": 2.014247551202137, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.625764846801758, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8930483460426331, + "num_tokens": 614662016.0, + "step": 15834 + }, + { + "epoch": 2.0143747614807275, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.441668510437012, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.9052338004112244, + "num_tokens": 614696317.0, + "step": 15835 + }, + { + "epoch": 2.014501971759318, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.422924041748047, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9086858034133911, + "num_tokens": 614731298.0, + "step": 15836 + }, + { + "epoch": 2.0146291820379085, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.462366104125977, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9118642210960388, + "num_tokens": 614769256.0, + "step": 15837 + }, + { + "epoch": 2.014756392316499, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.441858291625977, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.9032309651374817, + "num_tokens": 614809706.0, + "step": 15838 + }, + { + "epoch": 2.0148836025950896, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.479290008544922, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8971306085586548, + "num_tokens": 614854106.0, + "step": 15839 + }, + { + "epoch": 2.01501081287368, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.465471267700195, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9032547473907471, + "num_tokens": 614887631.0, + "step": 15840 + }, + { + "epoch": 2.0151380231522706, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.461383819580078, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9100459814071655, + "num_tokens": 614927872.0, + "step": 15841 + }, + { + "epoch": 2.015265233430861, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.404990196228027, + "learning_rate": 1e-06, + "loss": 0.2329, + "mean_token_accuracy": 0.9202303886413574, + "num_tokens": 614964916.0, + "step": 15842 + }, + { + "epoch": 2.0153924437094517, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.493802070617676, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.902182936668396, + "num_tokens": 615002418.0, + "step": 15843 + }, + { + "epoch": 2.0155196539880422, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.45223617553711, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9067988395690918, + "num_tokens": 615042641.0, + "step": 15844 + }, + { + "epoch": 2.0156468642666328, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.438766479492188, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9100574254989624, + "num_tokens": 615079935.0, + "step": 15845 + }, + { + "epoch": 2.0157740745452233, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.521327018737793, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9044463634490967, + "num_tokens": 615114677.0, + "step": 15846 + }, + { + "epoch": 2.015901284823814, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.566648483276367, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8954922556877136, + "num_tokens": 615149415.0, + "step": 15847 + }, + { + "epoch": 2.0160284951024043, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.39243221282959, + "learning_rate": 1e-06, + "loss": 0.2445, + "mean_token_accuracy": 0.9134728908538818, + "num_tokens": 615189095.0, + "step": 15848 + }, + { + "epoch": 2.016155705380995, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.526009559631348, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9130645394325256, + "num_tokens": 615223755.0, + "step": 15849 + }, + { + "epoch": 2.0162829156595854, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.509943962097168, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8927603363990784, + "num_tokens": 615260927.0, + "step": 15850 + }, + { + "epoch": 2.016410125938176, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.530579566955566, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8870803713798523, + "num_tokens": 615296705.0, + "step": 15851 + }, + { + "epoch": 2.0165373362167665, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.46980094909668, + "learning_rate": 1e-06, + "loss": 0.2455, + "mean_token_accuracy": 0.9150065183639526, + "num_tokens": 615332836.0, + "step": 15852 + }, + { + "epoch": 2.016664546495357, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.454303741455078, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.9066895246505737, + "num_tokens": 615369778.0, + "step": 15853 + }, + { + "epoch": 2.0167917567739475, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.427408218383789, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9041132926940918, + "num_tokens": 615412957.0, + "step": 15854 + }, + { + "epoch": 2.016918967052538, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.485590934753418, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8978517055511475, + "num_tokens": 615454386.0, + "step": 15855 + }, + { + "epoch": 2.0170461773311286, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.3856840133667, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9095594882965088, + "num_tokens": 615499335.0, + "step": 15856 + }, + { + "epoch": 2.0171733876097186, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.5067777633667, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.896405041217804, + "num_tokens": 615541208.0, + "step": 15857 + }, + { + "epoch": 2.017300597888309, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.421101570129395, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.90484619140625, + "num_tokens": 615579892.0, + "step": 15858 + }, + { + "epoch": 2.0174278081668997, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.493412971496582, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8996536135673523, + "num_tokens": 615618991.0, + "step": 15859 + }, + { + "epoch": 2.0175550184454902, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.455577850341797, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8885525465011597, + "num_tokens": 615663934.0, + "step": 15860 + }, + { + "epoch": 2.0176822287240808, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.387229919433594, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9054751992225647, + "num_tokens": 615708156.0, + "step": 15861 + }, + { + "epoch": 2.0178094390026713, + "ewc_loss": 0.0181884765625, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.435961723327637, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8923138380050659, + "num_tokens": 615751518.0, + "step": 15862 + }, + { + "epoch": 2.017936649281262, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.566617965698242, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.9028934240341187, + "num_tokens": 615783830.0, + "step": 15863 + }, + { + "epoch": 2.0180638595598523, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.540704727172852, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8948123455047607, + "num_tokens": 615819597.0, + "step": 15864 + }, + { + "epoch": 2.018191069838443, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.571744918823242, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9036300182342529, + "num_tokens": 615853015.0, + "step": 15865 + }, + { + "epoch": 2.0183182801170334, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.428930282592773, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9025060534477234, + "num_tokens": 615893975.0, + "step": 15866 + }, + { + "epoch": 2.018445490395624, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.408876419067383, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8951742649078369, + "num_tokens": 615935505.0, + "step": 15867 + }, + { + "epoch": 2.0185727006742145, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.40992546081543, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.9092419147491455, + "num_tokens": 615976570.0, + "step": 15868 + }, + { + "epoch": 2.018699910952805, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.448381423950195, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9028447270393372, + "num_tokens": 616015896.0, + "step": 15869 + }, + { + "epoch": 2.0188271212313955, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.449474334716797, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9023880958557129, + "num_tokens": 616055931.0, + "step": 15870 + }, + { + "epoch": 2.018954331509986, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.489531517028809, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8983793258666992, + "num_tokens": 616098630.0, + "step": 15871 + }, + { + "epoch": 2.0190815417885766, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.434009552001953, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8963123559951782, + "num_tokens": 616138896.0, + "step": 15872 + }, + { + "epoch": 2.019208752067167, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.482309341430664, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9015000462532043, + "num_tokens": 616175027.0, + "step": 15873 + }, + { + "epoch": 2.0193359623457576, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.440997123718262, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.9054105281829834, + "num_tokens": 616213608.0, + "step": 15874 + }, + { + "epoch": 2.019463172624348, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.496353149414062, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9084877967834473, + "num_tokens": 616247421.0, + "step": 15875 + }, + { + "epoch": 2.0195903829029387, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.818771362304688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.497513771057129, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8938151597976685, + "num_tokens": 616290254.0, + "step": 15876 + }, + { + "epoch": 2.019717593181529, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.428208351135254, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.9020296335220337, + "num_tokens": 616329491.0, + "step": 15877 + }, + { + "epoch": 2.0198448034601197, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.359376907348633, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9117170572280884, + "num_tokens": 616372733.0, + "step": 15878 + }, + { + "epoch": 2.0199720137387103, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.482772827148438, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8888901472091675, + "num_tokens": 616415442.0, + "step": 15879 + }, + { + "epoch": 2.020099224017301, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.42007064819336, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9075914025306702, + "num_tokens": 616454654.0, + "step": 15880 + }, + { + "epoch": 2.0202264342958913, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.609323501586914, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8857527375221252, + "num_tokens": 616490670.0, + "step": 15881 + }, + { + "epoch": 2.0203536445744814, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.438725471496582, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.9013038873672485, + "num_tokens": 616535948.0, + "step": 15882 + }, + { + "epoch": 2.020480854853072, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.362380027770996, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9072639346122742, + "num_tokens": 616574535.0, + "step": 15883 + }, + { + "epoch": 2.0206080651316625, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.514974594116211, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.9133924245834351, + "num_tokens": 616610433.0, + "step": 15884 + }, + { + "epoch": 2.020735275410253, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.407563209533691, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.903175413608551, + "num_tokens": 616650905.0, + "step": 15885 + }, + { + "epoch": 2.0208624856888435, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.511497497558594, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.899660587310791, + "num_tokens": 616691099.0, + "step": 15886 + }, + { + "epoch": 2.020989695967434, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.512290000915527, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.894561767578125, + "num_tokens": 616730635.0, + "step": 15887 + }, + { + "epoch": 2.0211169062460246, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.398284912109375, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8981022834777832, + "num_tokens": 616776379.0, + "step": 15888 + }, + { + "epoch": 2.021244116524615, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.532320022583008, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.9055122137069702, + "num_tokens": 616811457.0, + "step": 15889 + }, + { + "epoch": 2.0213713268032056, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.514683723449707, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8990570902824402, + "num_tokens": 616853399.0, + "step": 15890 + }, + { + "epoch": 2.021498537081796, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.527925491333008, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8984801769256592, + "num_tokens": 616890925.0, + "step": 15891 + }, + { + "epoch": 2.0216257473603867, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.489328384399414, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.9015404582023621, + "num_tokens": 616932245.0, + "step": 15892 + }, + { + "epoch": 2.021752957638977, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.575793266296387, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8937938213348389, + "num_tokens": 616964979.0, + "step": 15893 + }, + { + "epoch": 2.0218801679175677, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.417559623718262, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9061029553413391, + "num_tokens": 617006111.0, + "step": 15894 + }, + { + "epoch": 2.0220073781961583, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.478201866149902, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8892815113067627, + "num_tokens": 617045527.0, + "step": 15895 + }, + { + "epoch": 2.022134588474749, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.508316993713379, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.9006453156471252, + "num_tokens": 617081499.0, + "step": 15896 + }, + { + "epoch": 2.0222617987533393, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.500633239746094, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.9057726263999939, + "num_tokens": 617114443.0, + "step": 15897 + }, + { + "epoch": 2.02238900903193, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.427312850952148, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8986399173736572, + "num_tokens": 617153856.0, + "step": 15898 + }, + { + "epoch": 2.0225162193105204, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.457115173339844, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.8984126448631287, + "num_tokens": 617194296.0, + "step": 15899 + }, + { + "epoch": 2.022643429589111, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.553725242614746, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8988924026489258, + "num_tokens": 617232553.0, + "step": 15900 + }, + { + "epoch": 2.0227706398677014, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.494145393371582, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9029834270477295, + "num_tokens": 617267722.0, + "step": 15901 + }, + { + "epoch": 2.022897850146292, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.51845932006836, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8957424163818359, + "num_tokens": 617306561.0, + "step": 15902 + }, + { + "epoch": 2.0230250604248825, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.460257530212402, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9038686752319336, + "num_tokens": 617347666.0, + "step": 15903 + }, + { + "epoch": 2.023152270703473, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.496335983276367, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8984811305999756, + "num_tokens": 617386107.0, + "step": 15904 + }, + { + "epoch": 2.0232794809820636, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.509297370910645, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8988795280456543, + "num_tokens": 617428522.0, + "step": 15905 + }, + { + "epoch": 2.0234066912606536, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.431573867797852, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.9018053412437439, + "num_tokens": 617475405.0, + "step": 15906 + }, + { + "epoch": 2.023533901539244, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.49108600616455, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8955050706863403, + "num_tokens": 617517408.0, + "step": 15907 + }, + { + "epoch": 2.0236611118178347, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.44532585144043, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.896202564239502, + "num_tokens": 617564984.0, + "step": 15908 + }, + { + "epoch": 2.0237883220964252, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.597373962402344, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8970385789871216, + "num_tokens": 617604119.0, + "step": 15909 + }, + { + "epoch": 2.0239155323750158, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.539153099060059, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8970134854316711, + "num_tokens": 617640530.0, + "step": 15910 + }, + { + "epoch": 2.0240427426536063, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.40796184539795, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.9002789258956909, + "num_tokens": 617686407.0, + "step": 15911 + }, + { + "epoch": 2.024169952932197, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.520834922790527, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8998227119445801, + "num_tokens": 617727912.0, + "step": 15912 + }, + { + "epoch": 2.0242971632107873, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.601292610168457, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8910405039787292, + "num_tokens": 617766036.0, + "step": 15913 + }, + { + "epoch": 2.024424373489378, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.563444137573242, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9117708206176758, + "num_tokens": 617797893.0, + "step": 15914 + }, + { + "epoch": 2.0245515837679684, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.576248168945312, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.895051121711731, + "num_tokens": 617835259.0, + "step": 15915 + }, + { + "epoch": 2.024678794046559, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.46805477142334, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.9073683023452759, + "num_tokens": 617877347.0, + "step": 15916 + }, + { + "epoch": 2.0248060043251495, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.480888366699219, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.9004178643226624, + "num_tokens": 617917527.0, + "step": 15917 + }, + { + "epoch": 2.02493321460374, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.463920593261719, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8961487412452698, + "num_tokens": 617963500.0, + "step": 15918 + }, + { + "epoch": 2.0250604248823305, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.60954475402832, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.9012365341186523, + "num_tokens": 617996245.0, + "step": 15919 + }, + { + "epoch": 2.025187635160921, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.49269962310791, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8991996049880981, + "num_tokens": 618035594.0, + "step": 15920 + }, + { + "epoch": 2.0253148454395116, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.463833808898926, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.9006568193435669, + "num_tokens": 618074985.0, + "step": 15921 + }, + { + "epoch": 2.025442055718102, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.485368728637695, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9123276472091675, + "num_tokens": 618111105.0, + "step": 15922 + }, + { + "epoch": 2.0255692659966926, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.52794075012207, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8955913782119751, + "num_tokens": 618152643.0, + "step": 15923 + }, + { + "epoch": 2.025696476275283, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.563838005065918, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.887661337852478, + "num_tokens": 618187813.0, + "step": 15924 + }, + { + "epoch": 2.0258236865538737, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.424288749694824, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8948873281478882, + "num_tokens": 618231322.0, + "step": 15925 + }, + { + "epoch": 2.025950896832464, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.549560546875, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9060512185096741, + "num_tokens": 618263864.0, + "step": 15926 + }, + { + "epoch": 2.0260781071110547, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.551039695739746, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9095818996429443, + "num_tokens": 618297051.0, + "step": 15927 + }, + { + "epoch": 2.0262053173896453, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.49881649017334, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8944178223609924, + "num_tokens": 618334634.0, + "step": 15928 + }, + { + "epoch": 2.026332527668236, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.531081199645996, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9004453420639038, + "num_tokens": 618372367.0, + "step": 15929 + }, + { + "epoch": 2.0264597379468263, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.40290355682373, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9126020669937134, + "num_tokens": 618411877.0, + "step": 15930 + }, + { + "epoch": 2.0265869482254164, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.578890800476074, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8966569900512695, + "num_tokens": 618448685.0, + "step": 15931 + }, + { + "epoch": 2.026714158504007, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.463788032531738, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.9008457064628601, + "num_tokens": 618487114.0, + "step": 15932 + }, + { + "epoch": 2.0268413687825975, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.519928932189941, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8937742710113525, + "num_tokens": 618524933.0, + "step": 15933 + }, + { + "epoch": 2.026968579061188, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.516461372375488, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.9082614183425903, + "num_tokens": 618559175.0, + "step": 15934 + }, + { + "epoch": 2.0270957893397785, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.402931213378906, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.900294840335846, + "num_tokens": 618601160.0, + "step": 15935 + }, + { + "epoch": 2.027222999618369, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.516698837280273, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.9005792140960693, + "num_tokens": 618639117.0, + "step": 15936 + }, + { + "epoch": 2.0273502098969596, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.393133163452148, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9044944047927856, + "num_tokens": 618680432.0, + "step": 15937 + }, + { + "epoch": 2.02747742017555, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.511602401733398, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8968049883842468, + "num_tokens": 618715907.0, + "step": 15938 + }, + { + "epoch": 2.0276046304541406, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.529035568237305, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.9005864858627319, + "num_tokens": 618754861.0, + "step": 15939 + }, + { + "epoch": 2.027731840732731, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.42711067199707, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.895322859287262, + "num_tokens": 618796273.0, + "step": 15940 + }, + { + "epoch": 2.0278590510113217, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.478686332702637, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8980094194412231, + "num_tokens": 618838726.0, + "step": 15941 + }, + { + "epoch": 2.027986261289912, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.383082389831543, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.9011915922164917, + "num_tokens": 618881067.0, + "step": 15942 + }, + { + "epoch": 2.0281134715685027, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.415143966674805, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8925004601478577, + "num_tokens": 618923755.0, + "step": 15943 + }, + { + "epoch": 2.0282406818470933, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.481477737426758, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.909106969833374, + "num_tokens": 618962507.0, + "step": 15944 + }, + { + "epoch": 2.028367892125684, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.417020797729492, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9111800193786621, + "num_tokens": 619002578.0, + "step": 15945 + }, + { + "epoch": 2.0284951024042743, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.49719524383545, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.9065269231796265, + "num_tokens": 619038029.0, + "step": 15946 + }, + { + "epoch": 2.028622312682865, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.469432830810547, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.9011551737785339, + "num_tokens": 619077786.0, + "step": 15947 + }, + { + "epoch": 2.0287495229614554, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.6632719039917, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.898634672164917, + "num_tokens": 619113648.0, + "step": 15948 + }, + { + "epoch": 2.028876733240046, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.453951835632324, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.891826868057251, + "num_tokens": 619154286.0, + "step": 15949 + }, + { + "epoch": 2.0290039435186364, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.458141326904297, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.9018498659133911, + "num_tokens": 619194534.0, + "step": 15950 + }, + { + "epoch": 2.029131153797227, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.411613464355469, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8990299105644226, + "num_tokens": 619237664.0, + "step": 15951 + }, + { + "epoch": 2.0292583640758175, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.446868896484375, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.9078916311264038, + "num_tokens": 619272982.0, + "step": 15952 + }, + { + "epoch": 2.029385574354408, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.43564510345459, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8970962762832642, + "num_tokens": 619315542.0, + "step": 15953 + }, + { + "epoch": 2.0295127846329986, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.507737159729004, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8991061449050903, + "num_tokens": 619353496.0, + "step": 15954 + }, + { + "epoch": 2.0296399949115886, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.58373737335205, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8912156820297241, + "num_tokens": 619390807.0, + "step": 15955 + }, + { + "epoch": 2.029767205190179, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.406570434570312, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.90114426612854, + "num_tokens": 619433123.0, + "step": 15956 + }, + { + "epoch": 2.0298944154687697, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.533172607421875, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8990175724029541, + "num_tokens": 619468589.0, + "step": 15957 + }, + { + "epoch": 2.0300216257473602, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.47099494934082, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9034073352813721, + "num_tokens": 619510011.0, + "step": 15958 + }, + { + "epoch": 2.0301488360259508, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.382616996765137, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.9044981002807617, + "num_tokens": 619545596.0, + "step": 15959 + }, + { + "epoch": 2.0302760463045413, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.56406021118164, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9043114185333252, + "num_tokens": 619577178.0, + "step": 15960 + }, + { + "epoch": 2.030403256583132, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.466548919677734, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8810429573059082, + "num_tokens": 619616391.0, + "step": 15961 + }, + { + "epoch": 2.0305304668617223, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.469674110412598, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8941265940666199, + "num_tokens": 619658383.0, + "step": 15962 + }, + { + "epoch": 2.030657677140313, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.472890853881836, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.888797402381897, + "num_tokens": 619698732.0, + "step": 15963 + }, + { + "epoch": 2.0307848874189034, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.364855766296387, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9129631519317627, + "num_tokens": 619739918.0, + "step": 15964 + }, + { + "epoch": 2.030912097697494, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.436470031738281, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9105413556098938, + "num_tokens": 619782062.0, + "step": 15965 + }, + { + "epoch": 2.0310393079760845, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.464559555053711, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.9040495157241821, + "num_tokens": 619822249.0, + "step": 15966 + }, + { + "epoch": 2.031166518254675, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.470151901245117, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8971353769302368, + "num_tokens": 619861219.0, + "step": 15967 + }, + { + "epoch": 2.0312937285332655, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.409975051879883, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.9041832089424133, + "num_tokens": 619902184.0, + "step": 15968 + }, + { + "epoch": 2.031420938811856, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.45263385772705, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9080095887184143, + "num_tokens": 619942386.0, + "step": 15969 + }, + { + "epoch": 2.0315481490904466, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.433810234069824, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.9038292169570923, + "num_tokens": 619984452.0, + "step": 15970 + }, + { + "epoch": 2.031675359369037, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.464385032653809, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.9032938480377197, + "num_tokens": 620022392.0, + "step": 15971 + }, + { + "epoch": 2.0318025696476276, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.536778450012207, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8865770101547241, + "num_tokens": 620059278.0, + "step": 15972 + }, + { + "epoch": 2.031929779926218, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.866455078125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.49653434753418, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9091240167617798, + "num_tokens": 620094147.0, + "step": 15973 + }, + { + "epoch": 2.0320569902048087, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.549042701721191, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8989866375923157, + "num_tokens": 620128924.0, + "step": 15974 + }, + { + "epoch": 2.032184200483399, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.394960403442383, + "learning_rate": 1e-06, + "loss": 0.2454, + "mean_token_accuracy": 0.9148960709571838, + "num_tokens": 620167476.0, + "step": 15975 + }, + { + "epoch": 2.0323114107619897, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.510382652282715, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8987897634506226, + "num_tokens": 620204985.0, + "step": 15976 + }, + { + "epoch": 2.0324386210405803, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.517905235290527, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8933854103088379, + "num_tokens": 620247187.0, + "step": 15977 + }, + { + "epoch": 2.032565831319171, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.563138961791992, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.909233808517456, + "num_tokens": 620279904.0, + "step": 15978 + }, + { + "epoch": 2.032693041597761, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.472813606262207, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8975507616996765, + "num_tokens": 620321147.0, + "step": 15979 + }, + { + "epoch": 2.0328202518763514, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.534196853637695, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.890044093132019, + "num_tokens": 620358777.0, + "step": 15980 + }, + { + "epoch": 2.032947462154942, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.615189552307129, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8934957981109619, + "num_tokens": 620393477.0, + "step": 15981 + }, + { + "epoch": 2.0330746724335325, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.381423950195312, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8979999423027039, + "num_tokens": 620437354.0, + "step": 15982 + }, + { + "epoch": 2.033201882712123, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.47458267211914, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.9057543277740479, + "num_tokens": 620477572.0, + "step": 15983 + }, + { + "epoch": 2.0333290929907135, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.519920349121094, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.9012707471847534, + "num_tokens": 620516331.0, + "step": 15984 + }, + { + "epoch": 2.033456303269304, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.488605499267578, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8874699473381042, + "num_tokens": 620559473.0, + "step": 15985 + }, + { + "epoch": 2.0335835135478946, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.6375093460083, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8937474489212036, + "num_tokens": 620592192.0, + "step": 15986 + }, + { + "epoch": 2.033710723826485, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.460503578186035, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.9027529954910278, + "num_tokens": 620630476.0, + "step": 15987 + }, + { + "epoch": 2.0338379341050756, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.63192367553711, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.9009265899658203, + "num_tokens": 620663137.0, + "step": 15988 + }, + { + "epoch": 2.033965144383666, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.606613159179688, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8923627734184265, + "num_tokens": 620698034.0, + "step": 15989 + }, + { + "epoch": 2.0340923546622567, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.39205265045166, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8976142406463623, + "num_tokens": 620745030.0, + "step": 15990 + }, + { + "epoch": 2.034219564940847, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.476386070251465, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8972101807594299, + "num_tokens": 620787153.0, + "step": 15991 + }, + { + "epoch": 2.0343467752194377, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.498395919799805, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.9057189226150513, + "num_tokens": 620823598.0, + "step": 15992 + }, + { + "epoch": 2.0344739854980283, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.458935737609863, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8987347483634949, + "num_tokens": 620870141.0, + "step": 15993 + }, + { + "epoch": 2.034601195776619, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.462624549865723, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.9081252813339233, + "num_tokens": 620912441.0, + "step": 15994 + }, + { + "epoch": 2.0347284060552093, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.514212608337402, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8994830846786499, + "num_tokens": 620954261.0, + "step": 15995 + }, + { + "epoch": 2.0348556163338, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.544440269470215, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8848307728767395, + "num_tokens": 620993565.0, + "step": 15996 + }, + { + "epoch": 2.0349828266123904, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.4850492477417, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8935528993606567, + "num_tokens": 621031517.0, + "step": 15997 + }, + { + "epoch": 2.035110036890981, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.498238563537598, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8972994089126587, + "num_tokens": 621075200.0, + "step": 15998 + }, + { + "epoch": 2.0352372471695714, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.560088157653809, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8982869386672974, + "num_tokens": 621110177.0, + "step": 15999 + }, + { + "epoch": 2.035364457448162, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.478106498718262, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8989821672439575, + "num_tokens": 621156689.0, + "step": 16000 + }, + { + "epoch": 2.0354916677267525, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.446599006652832, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8920665979385376, + "num_tokens": 621195325.0, + "step": 16001 + }, + { + "epoch": 2.035618878005343, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.54985523223877, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8948639631271362, + "num_tokens": 621235786.0, + "step": 16002 + }, + { + "epoch": 2.0357460882839336, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.580763816833496, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8904272317886353, + "num_tokens": 621273305.0, + "step": 16003 + }, + { + "epoch": 2.0358732985625236, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.522958755493164, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9121525287628174, + "num_tokens": 621308313.0, + "step": 16004 + }, + { + "epoch": 2.036000508841114, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.484911918640137, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8885691165924072, + "num_tokens": 621347179.0, + "step": 16005 + }, + { + "epoch": 2.0361277191197047, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.491713523864746, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.9010221362113953, + "num_tokens": 621388937.0, + "step": 16006 + }, + { + "epoch": 2.036254929398295, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.461718559265137, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9112347364425659, + "num_tokens": 621429795.0, + "step": 16007 + }, + { + "epoch": 2.0363821396768857, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.475432395935059, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9002732038497925, + "num_tokens": 621472353.0, + "step": 16008 + }, + { + "epoch": 2.0365093499554763, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.724459648132324, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8984429836273193, + "num_tokens": 621504232.0, + "step": 16009 + }, + { + "epoch": 2.036636560234067, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.438878059387207, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8900594115257263, + "num_tokens": 621545786.0, + "step": 16010 + }, + { + "epoch": 2.0367637705126573, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.55469036102295, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8942707777023315, + "num_tokens": 621585272.0, + "step": 16011 + }, + { + "epoch": 2.036890980791248, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.474448204040527, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9048058986663818, + "num_tokens": 621625538.0, + "step": 16012 + }, + { + "epoch": 2.0370181910698384, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.526759147644043, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8986840844154358, + "num_tokens": 621661685.0, + "step": 16013 + }, + { + "epoch": 2.037145401348429, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.510062217712402, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.9104744791984558, + "num_tokens": 621698913.0, + "step": 16014 + }, + { + "epoch": 2.0372726116270194, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.515165328979492, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8919475674629211, + "num_tokens": 621738288.0, + "step": 16015 + }, + { + "epoch": 2.03739982190561, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.612505912780762, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8936443328857422, + "num_tokens": 621771697.0, + "step": 16016 + }, + { + "epoch": 2.0375270321842005, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.543795585632324, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8998705744743347, + "num_tokens": 621812196.0, + "step": 16017 + }, + { + "epoch": 2.037654242462791, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.491662979125977, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.903599202632904, + "num_tokens": 621849062.0, + "step": 16018 + }, + { + "epoch": 2.0377814527413816, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.572894096374512, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.895188570022583, + "num_tokens": 621884656.0, + "step": 16019 + }, + { + "epoch": 2.037908663019972, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.414865493774414, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8977950215339661, + "num_tokens": 621925742.0, + "step": 16020 + }, + { + "epoch": 2.0380358732985626, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.41472053527832, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9132528901100159, + "num_tokens": 621968554.0, + "step": 16021 + }, + { + "epoch": 2.038163083577153, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.455795288085938, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.910952091217041, + "num_tokens": 622007138.0, + "step": 16022 + }, + { + "epoch": 2.0382902938557437, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.413593292236328, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9079722166061401, + "num_tokens": 622049896.0, + "step": 16023 + }, + { + "epoch": 2.038417504134334, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.529802322387695, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.9007630348205566, + "num_tokens": 622085323.0, + "step": 16024 + }, + { + "epoch": 2.0385447144129247, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.513273239135742, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.9044294953346252, + "num_tokens": 622124821.0, + "step": 16025 + }, + { + "epoch": 2.0386719246915153, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.454423904418945, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.9023412466049194, + "num_tokens": 622166432.0, + "step": 16026 + }, + { + "epoch": 2.038799134970106, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.628484725952148, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8896989822387695, + "num_tokens": 622203724.0, + "step": 16027 + }, + { + "epoch": 2.0389263452486963, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.488401412963867, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8952777981758118, + "num_tokens": 622244655.0, + "step": 16028 + }, + { + "epoch": 2.0390535555272864, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.513230323791504, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9048354625701904, + "num_tokens": 622277416.0, + "step": 16029 + }, + { + "epoch": 2.039180765805877, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.492713928222656, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.890579879283905, + "num_tokens": 622316092.0, + "step": 16030 + }, + { + "epoch": 2.0393079760844675, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.43107795715332, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8978939652442932, + "num_tokens": 622356890.0, + "step": 16031 + }, + { + "epoch": 2.039435186363058, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.568164825439453, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8946369886398315, + "num_tokens": 622396330.0, + "step": 16032 + }, + { + "epoch": 2.0395623966416485, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.497788429260254, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9072126150131226, + "num_tokens": 622438592.0, + "step": 16033 + }, + { + "epoch": 2.039689606920239, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.465158462524414, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9024011492729187, + "num_tokens": 622480784.0, + "step": 16034 + }, + { + "epoch": 2.0398168171988296, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.51708984375, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8977158665657043, + "num_tokens": 622523685.0, + "step": 16035 + }, + { + "epoch": 2.03994402747742, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.009506225585938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.653642654418945, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.895209789276123, + "num_tokens": 622557788.0, + "step": 16036 + }, + { + "epoch": 2.0400712377560106, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.473936080932617, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.9075481295585632, + "num_tokens": 622597950.0, + "step": 16037 + }, + { + "epoch": 2.040198448034601, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.501134872436523, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8965433239936829, + "num_tokens": 622635783.0, + "step": 16038 + }, + { + "epoch": 2.0403256583131917, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.561537742614746, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8936973810195923, + "num_tokens": 622672980.0, + "step": 16039 + }, + { + "epoch": 2.040452868591782, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.533648490905762, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8849097490310669, + "num_tokens": 622716677.0, + "step": 16040 + }, + { + "epoch": 2.0405800788703727, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.521319389343262, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.896031379699707, + "num_tokens": 622751408.0, + "step": 16041 + }, + { + "epoch": 2.0407072891489633, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.614502906799316, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9053346514701843, + "num_tokens": 622784300.0, + "step": 16042 + }, + { + "epoch": 2.040834499427554, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.597990036010742, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8855271339416504, + "num_tokens": 622827996.0, + "step": 16043 + }, + { + "epoch": 2.0409617097061443, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.515687942504883, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8887076377868652, + "num_tokens": 622867853.0, + "step": 16044 + }, + { + "epoch": 2.041088919984735, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.376506805419922, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9016609191894531, + "num_tokens": 622908181.0, + "step": 16045 + }, + { + "epoch": 2.0412161302633254, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.549697875976562, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8843965530395508, + "num_tokens": 622951243.0, + "step": 16046 + }, + { + "epoch": 2.041343340541916, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.420001983642578, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9090304374694824, + "num_tokens": 622992343.0, + "step": 16047 + }, + { + "epoch": 2.0414705508205064, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.414460182189941, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.9011510610580444, + "num_tokens": 623036174.0, + "step": 16048 + }, + { + "epoch": 2.041597761099097, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.498312950134277, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.912041425704956, + "num_tokens": 623071960.0, + "step": 16049 + }, + { + "epoch": 2.0417249713776875, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.59574031829834, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8903522491455078, + "num_tokens": 623109653.0, + "step": 16050 + }, + { + "epoch": 2.041852181656278, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.451944351196289, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9086775779724121, + "num_tokens": 623144676.0, + "step": 16051 + }, + { + "epoch": 2.0419793919348685, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.914138793945312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.503315925598145, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.9071850776672363, + "num_tokens": 623182918.0, + "step": 16052 + }, + { + "epoch": 2.0421066022134586, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.437058448791504, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9073193669319153, + "num_tokens": 623221810.0, + "step": 16053 + }, + { + "epoch": 2.042233812492049, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.437667846679688, + "learning_rate": 1e-06, + "loss": 0.264, + "mean_token_accuracy": 0.9095736742019653, + "num_tokens": 623266338.0, + "step": 16054 + }, + { + "epoch": 2.0423610227706397, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.536092758178711, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8959062099456787, + "num_tokens": 623309346.0, + "step": 16055 + }, + { + "epoch": 2.04248823304923, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.535784721374512, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.9063758254051208, + "num_tokens": 623345378.0, + "step": 16056 + }, + { + "epoch": 2.0426154433278207, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.522297859191895, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.9045753479003906, + "num_tokens": 623382464.0, + "step": 16057 + }, + { + "epoch": 2.0427426536064113, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.505486488342285, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8895415663719177, + "num_tokens": 623425351.0, + "step": 16058 + }, + { + "epoch": 2.042869863885002, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.533025741577148, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8987835645675659, + "num_tokens": 623462459.0, + "step": 16059 + }, + { + "epoch": 2.0429970741635923, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.528948783874512, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8936278223991394, + "num_tokens": 623500785.0, + "step": 16060 + }, + { + "epoch": 2.043124284442183, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.464713096618652, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.9021003246307373, + "num_tokens": 623536778.0, + "step": 16061 + }, + { + "epoch": 2.0432514947207734, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.448897361755371, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.9064496755599976, + "num_tokens": 623578404.0, + "step": 16062 + }, + { + "epoch": 2.043378704999364, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.48989486694336, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.908251166343689, + "num_tokens": 623614680.0, + "step": 16063 + }, + { + "epoch": 2.0435059152779544, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.436002731323242, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9066992998123169, + "num_tokens": 623656386.0, + "step": 16064 + }, + { + "epoch": 2.043633125556545, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.47370719909668, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9053356051445007, + "num_tokens": 623693382.0, + "step": 16065 + }, + { + "epoch": 2.0437603358351355, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.475099563598633, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8990663886070251, + "num_tokens": 623735009.0, + "step": 16066 + }, + { + "epoch": 2.043887546113726, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.556049346923828, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9071910381317139, + "num_tokens": 623770414.0, + "step": 16067 + }, + { + "epoch": 2.0440147563923166, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.478300094604492, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9060791730880737, + "num_tokens": 623806548.0, + "step": 16068 + }, + { + "epoch": 2.044141966670907, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.540743827819824, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.9034631252288818, + "num_tokens": 623844723.0, + "step": 16069 + }, + { + "epoch": 2.0442691769494976, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.509617805480957, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.9003015160560608, + "num_tokens": 623881550.0, + "step": 16070 + }, + { + "epoch": 2.044396387228088, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.614737510681152, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.9033437967300415, + "num_tokens": 623913466.0, + "step": 16071 + }, + { + "epoch": 2.0445235975066787, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.513968467712402, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8944709300994873, + "num_tokens": 623947909.0, + "step": 16072 + }, + { + "epoch": 2.044650807785269, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.419490814208984, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9036229848861694, + "num_tokens": 623992672.0, + "step": 16073 + }, + { + "epoch": 2.0447780180638597, + "ewc_loss": 0.018310546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.469219207763672, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9064491987228394, + "num_tokens": 624031476.0, + "step": 16074 + }, + { + "epoch": 2.0449052283424503, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.380667686462402, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9050161838531494, + "num_tokens": 624073452.0, + "step": 16075 + }, + { + "epoch": 2.045032438621041, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.46265697479248, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9042896032333374, + "num_tokens": 624107971.0, + "step": 16076 + }, + { + "epoch": 2.045159648899631, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.489546775817871, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8995662331581116, + "num_tokens": 624146371.0, + "step": 16077 + }, + { + "epoch": 2.0452868591782214, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.546943664550781, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8920769691467285, + "num_tokens": 624183925.0, + "step": 16078 + }, + { + "epoch": 2.045414069456812, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.386799812316895, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8951042294502258, + "num_tokens": 624227196.0, + "step": 16079 + }, + { + "epoch": 2.0455412797354025, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.401628494262695, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.9034284353256226, + "num_tokens": 624266830.0, + "step": 16080 + }, + { + "epoch": 2.045668490013993, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.450090408325195, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8995107412338257, + "num_tokens": 624310018.0, + "step": 16081 + }, + { + "epoch": 2.0457957002925835, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.520263671875, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9028218984603882, + "num_tokens": 624346635.0, + "step": 16082 + }, + { + "epoch": 2.045922910571174, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 6.961822509765625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.488234519958496, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.898727297782898, + "num_tokens": 624387253.0, + "step": 16083 + }, + { + "epoch": 2.0460501208497646, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.009506225585938e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.540664672851562, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.898194432258606, + "num_tokens": 624420503.0, + "step": 16084 + }, + { + "epoch": 2.046177331128355, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.009506225585938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.54357624053955, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.9006858468055725, + "num_tokens": 624454686.0, + "step": 16085 + }, + { + "epoch": 2.0463045414069456, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.009506225585938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.568148612976074, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8906124234199524, + "num_tokens": 624493055.0, + "step": 16086 + }, + { + "epoch": 2.046431751685536, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.009506225585938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.392818450927734, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9054064154624939, + "num_tokens": 624535164.0, + "step": 16087 + }, + { + "epoch": 2.0465589619641267, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.009506225585938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.372161865234375, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9042445421218872, + "num_tokens": 624581005.0, + "step": 16088 + }, + { + "epoch": 2.046686172242717, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.009506225585938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.457590103149414, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9072509407997131, + "num_tokens": 624625043.0, + "step": 16089 + }, + { + "epoch": 2.0468133825213077, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.009506225585938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.449121475219727, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.907898485660553, + "num_tokens": 624664419.0, + "step": 16090 + }, + { + "epoch": 2.0469405927998983, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.009506225585938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.414457321166992, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.9026126861572266, + "num_tokens": 624708138.0, + "step": 16091 + }, + { + "epoch": 2.047067803078489, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.009506225585938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.503962516784668, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8946404457092285, + "num_tokens": 624753663.0, + "step": 16092 + }, + { + "epoch": 2.0471950133570793, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.488626480102539, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8964966535568237, + "num_tokens": 624793084.0, + "step": 16093 + }, + { + "epoch": 2.04732222363567, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.537740707397461, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9088356494903564, + "num_tokens": 624828963.0, + "step": 16094 + }, + { + "epoch": 2.0474494339142604, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.488424301147461, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8950644731521606, + "num_tokens": 624870210.0, + "step": 16095 + }, + { + "epoch": 2.047576644192851, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.534004211425781, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.901694655418396, + "num_tokens": 624909092.0, + "step": 16096 + }, + { + "epoch": 2.0477038544714414, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.439966201782227, + "learning_rate": 1e-06, + "loss": 0.2317, + "mean_token_accuracy": 0.919485330581665, + "num_tokens": 624947661.0, + "step": 16097 + }, + { + "epoch": 2.047831064750032, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.548009872436523, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9073401689529419, + "num_tokens": 624983876.0, + "step": 16098 + }, + { + "epoch": 2.0479582750286225, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.491141319274902, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8968116641044617, + "num_tokens": 625021873.0, + "step": 16099 + }, + { + "epoch": 2.048085485307213, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.552946090698242, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8755586743354797, + "num_tokens": 625067184.0, + "step": 16100 + }, + { + "epoch": 2.0482126955858035, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.574548721313477, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8910815119743347, + "num_tokens": 625103368.0, + "step": 16101 + }, + { + "epoch": 2.0483399058643936, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.601262092590332, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.9055209755897522, + "num_tokens": 625139328.0, + "step": 16102 + }, + { + "epoch": 2.048467116142984, + "ewc_loss": 0.0184326171875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.415897369384766, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.9038802981376648, + "num_tokens": 625178715.0, + "step": 16103 + }, + { + "epoch": 2.0485943264215747, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.600629806518555, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8969568014144897, + "num_tokens": 625216302.0, + "step": 16104 + }, + { + "epoch": 2.048721536700165, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.433125495910645, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9087489247322083, + "num_tokens": 625256172.0, + "step": 16105 + }, + { + "epoch": 2.0488487469787557, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.489782333374023, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9070715308189392, + "num_tokens": 625293115.0, + "step": 16106 + }, + { + "epoch": 2.0489759572573463, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.631025314331055, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8875657320022583, + "num_tokens": 625330252.0, + "step": 16107 + }, + { + "epoch": 2.049103167535937, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.474148750305176, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8992655277252197, + "num_tokens": 625371499.0, + "step": 16108 + }, + { + "epoch": 2.0492303778145273, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.669698715209961, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.9014837741851807, + "num_tokens": 625409880.0, + "step": 16109 + }, + { + "epoch": 2.049357588093118, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.634258270263672, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.9026920795440674, + "num_tokens": 625445460.0, + "step": 16110 + }, + { + "epoch": 2.0494847983717084, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.467241287231445, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8901252746582031, + "num_tokens": 625486467.0, + "step": 16111 + }, + { + "epoch": 2.049612008650299, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.612192153930664, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8921087980270386, + "num_tokens": 625520882.0, + "step": 16112 + }, + { + "epoch": 2.0497392189288894, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.490510940551758, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9042617082595825, + "num_tokens": 625560673.0, + "step": 16113 + }, + { + "epoch": 2.04986642920748, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.59794807434082, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8890149593353271, + "num_tokens": 625596179.0, + "step": 16114 + }, + { + "epoch": 2.0499936394860705, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.546659469604492, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8986160159111023, + "num_tokens": 625631537.0, + "step": 16115 + }, + { + "epoch": 2.050120849764661, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.53423023223877, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8933948278427124, + "num_tokens": 625668686.0, + "step": 16116 + }, + { + "epoch": 2.0502480600432516, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.539751052856445, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8997429609298706, + "num_tokens": 625706613.0, + "step": 16117 + }, + { + "epoch": 2.050375270321842, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.505253791809082, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.905907392501831, + "num_tokens": 625741868.0, + "step": 16118 + }, + { + "epoch": 2.0505024806004326, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.5668363571167, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.9007124900817871, + "num_tokens": 625776590.0, + "step": 16119 + }, + { + "epoch": 2.050629690879023, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.441884994506836, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.9006842374801636, + "num_tokens": 625818599.0, + "step": 16120 + }, + { + "epoch": 2.0507569011576137, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.580119132995605, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8927810788154602, + "num_tokens": 625855724.0, + "step": 16121 + }, + { + "epoch": 2.050884111436204, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.524714469909668, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8909338712692261, + "num_tokens": 625894649.0, + "step": 16122 + }, + { + "epoch": 2.0510113217147947, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.444559097290039, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.9082326889038086, + "num_tokens": 625934445.0, + "step": 16123 + }, + { + "epoch": 2.0511385319933853, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.559207916259766, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.9031070470809937, + "num_tokens": 625969609.0, + "step": 16124 + }, + { + "epoch": 2.051265742271976, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.557262420654297, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.905393660068512, + "num_tokens": 626004001.0, + "step": 16125 + }, + { + "epoch": 2.0513929525505663, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.51289176940918, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.9033713340759277, + "num_tokens": 626041477.0, + "step": 16126 + }, + { + "epoch": 2.0515201628291564, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.488783836364746, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8978635668754578, + "num_tokens": 626078020.0, + "step": 16127 + }, + { + "epoch": 2.051647373107747, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.537015914916992, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8998745679855347, + "num_tokens": 626113089.0, + "step": 16128 + }, + { + "epoch": 2.0517745833863374, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.444246292114258, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9068294167518616, + "num_tokens": 626152291.0, + "step": 16129 + }, + { + "epoch": 2.051901793664928, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.618945121765137, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8881900906562805, + "num_tokens": 626187356.0, + "step": 16130 + }, + { + "epoch": 2.0520290039435185, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.480687141418457, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.901139497756958, + "num_tokens": 626230736.0, + "step": 16131 + }, + { + "epoch": 2.052156214222109, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.636126518249512, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8900755643844604, + "num_tokens": 626265480.0, + "step": 16132 + }, + { + "epoch": 2.0522834245006996, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.455487251281738, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9096291661262512, + "num_tokens": 626303916.0, + "step": 16133 + }, + { + "epoch": 2.05241063477929, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.484435081481934, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8938783407211304, + "num_tokens": 626345002.0, + "step": 16134 + }, + { + "epoch": 2.0525378450578806, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.46525764465332, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.9081645607948303, + "num_tokens": 626384873.0, + "step": 16135 + }, + { + "epoch": 2.052665055336471, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.104873657226562e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.56003475189209, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.900074303150177, + "num_tokens": 626421113.0, + "step": 16136 + }, + { + "epoch": 2.0527922656150617, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.152557373046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.487385749816895, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8990169763565063, + "num_tokens": 626463025.0, + "step": 16137 + }, + { + "epoch": 2.052919475893652, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.152557373046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.4207763671875, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9045215845108032, + "num_tokens": 626511280.0, + "step": 16138 + }, + { + "epoch": 2.0530466861722427, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.152557373046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.622322082519531, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8951376676559448, + "num_tokens": 626546245.0, + "step": 16139 + }, + { + "epoch": 2.0531738964508333, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.152557373046875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.592524528503418, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8948714733123779, + "num_tokens": 626582570.0, + "step": 16140 + }, + { + "epoch": 2.053301106729424, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.152557373046875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.455846786499023, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9054141044616699, + "num_tokens": 626622256.0, + "step": 16141 + }, + { + "epoch": 2.0534283170080143, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.152557373046875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.550119400024414, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8895265460014343, + "num_tokens": 626665744.0, + "step": 16142 + }, + { + "epoch": 2.053555527286605, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.152557373046875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.596553802490234, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8992669582366943, + "num_tokens": 626701690.0, + "step": 16143 + }, + { + "epoch": 2.0536827375651954, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.434998512268066, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9078567028045654, + "num_tokens": 626738450.0, + "step": 16144 + }, + { + "epoch": 2.053809947843786, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.470715522766113, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9033926725387573, + "num_tokens": 626782045.0, + "step": 16145 + }, + { + "epoch": 2.0539371581223764, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.517718315124512, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.896629273891449, + "num_tokens": 626824531.0, + "step": 16146 + }, + { + "epoch": 2.054064368400967, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.50981330871582, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9019191861152649, + "num_tokens": 626863183.0, + "step": 16147 + }, + { + "epoch": 2.0541915786795575, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.512040138244629, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9050347805023193, + "num_tokens": 626901698.0, + "step": 16148 + }, + { + "epoch": 2.054318788958148, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.47445011138916, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9074691534042358, + "num_tokens": 626944470.0, + "step": 16149 + }, + { + "epoch": 2.0544459992367385, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.503544807434082, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8942630290985107, + "num_tokens": 626991991.0, + "step": 16150 + }, + { + "epoch": 2.0545732095153286, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.527104377746582, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8989914655685425, + "num_tokens": 627032231.0, + "step": 16151 + }, + { + "epoch": 2.054700419793919, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.502985000610352, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.9031782150268555, + "num_tokens": 627071816.0, + "step": 16152 + }, + { + "epoch": 2.0548276300725097, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.605265617370605, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9070918560028076, + "num_tokens": 627104515.0, + "step": 16153 + }, + { + "epoch": 2.0549548403511, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.547410011291504, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.896155595779419, + "num_tokens": 627144015.0, + "step": 16154 + }, + { + "epoch": 2.0550820506296907, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.435296058654785, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9056425094604492, + "num_tokens": 627180518.0, + "step": 16155 + }, + { + "epoch": 2.0552092609082813, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.639021873474121, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.90616375207901, + "num_tokens": 627212336.0, + "step": 16156 + }, + { + "epoch": 2.055336471186872, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.53704833984375, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9076739549636841, + "num_tokens": 627247420.0, + "step": 16157 + }, + { + "epoch": 2.0554636814654623, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.578863143920898, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8878119587898254, + "num_tokens": 627286241.0, + "step": 16158 + }, + { + "epoch": 2.055590891744053, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.530661582946777, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8942394256591797, + "num_tokens": 627327106.0, + "step": 16159 + }, + { + "epoch": 2.0557181020226434, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.505101203918457, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8890514969825745, + "num_tokens": 627368963.0, + "step": 16160 + }, + { + "epoch": 2.055845312301234, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.489537239074707, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8959635496139526, + "num_tokens": 627409367.0, + "step": 16161 + }, + { + "epoch": 2.0559725225798244, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.516797065734863, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.9049404263496399, + "num_tokens": 627444088.0, + "step": 16162 + }, + { + "epoch": 2.056099732858415, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.57709789276123, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8992105722427368, + "num_tokens": 627480886.0, + "step": 16163 + }, + { + "epoch": 2.0562269431370055, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.58807373046875, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.899086594581604, + "num_tokens": 627519089.0, + "step": 16164 + }, + { + "epoch": 2.056354153415596, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.522368431091309, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8990778923034668, + "num_tokens": 627559046.0, + "step": 16165 + }, + { + "epoch": 2.0564813636941865, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.512224197387695, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8983402848243713, + "num_tokens": 627601462.0, + "step": 16166 + }, + { + "epoch": 2.056608573972777, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.594820976257324, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.9052059650421143, + "num_tokens": 627632800.0, + "step": 16167 + }, + { + "epoch": 2.0567357842513676, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.652478218078613, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8759604692459106, + "num_tokens": 627669716.0, + "step": 16168 + }, + { + "epoch": 2.056862994529958, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.396293640136719, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.9026166200637817, + "num_tokens": 627711467.0, + "step": 16169 + }, + { + "epoch": 2.0569902048085487, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.518067359924316, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8920469284057617, + "num_tokens": 627754948.0, + "step": 16170 + }, + { + "epoch": 2.057117415087139, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.473136901855469, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.9006757736206055, + "num_tokens": 627797419.0, + "step": 16171 + }, + { + "epoch": 2.0572446253657297, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.643107414245605, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8934900760650635, + "num_tokens": 627830022.0, + "step": 16172 + }, + { + "epoch": 2.0573718356443202, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.545424461364746, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8971751928329468, + "num_tokens": 627865425.0, + "step": 16173 + }, + { + "epoch": 2.0574990459229108, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.461431503295898, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.9033588171005249, + "num_tokens": 627905808.0, + "step": 16174 + }, + { + "epoch": 2.057626256201501, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.474705696105957, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.9005401134490967, + "num_tokens": 627947220.0, + "step": 16175 + }, + { + "epoch": 2.0577534664800914, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.570143699645996, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9052761793136597, + "num_tokens": 627982965.0, + "step": 16176 + }, + { + "epoch": 2.057880676758682, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.55813217163086, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8950234651565552, + "num_tokens": 628022258.0, + "step": 16177 + }, + { + "epoch": 2.0580078870372724, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.551065444946289, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8954887986183167, + "num_tokens": 628065301.0, + "step": 16178 + }, + { + "epoch": 2.058135097315863, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.570099830627441, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9014413952827454, + "num_tokens": 628099709.0, + "step": 16179 + }, + { + "epoch": 2.0582623075944535, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.45787525177002, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9028618335723877, + "num_tokens": 628142363.0, + "step": 16180 + }, + { + "epoch": 2.058389517873044, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.573355674743652, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9047058820724487, + "num_tokens": 628180393.0, + "step": 16181 + }, + { + "epoch": 2.0585167281516346, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.479171752929688, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8866572976112366, + "num_tokens": 628230225.0, + "step": 16182 + }, + { + "epoch": 2.058643938430225, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.526819229125977, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9065898656845093, + "num_tokens": 628266268.0, + "step": 16183 + }, + { + "epoch": 2.0587711487088156, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.569174766540527, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8942122459411621, + "num_tokens": 628308969.0, + "step": 16184 + }, + { + "epoch": 2.058898358987406, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.461532592773438, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.9054364562034607, + "num_tokens": 628345790.0, + "step": 16185 + }, + { + "epoch": 2.0590255692659967, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.705167770385742, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8855975866317749, + "num_tokens": 628380616.0, + "step": 16186 + }, + { + "epoch": 2.059152779544587, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.559209823608398, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8935375213623047, + "num_tokens": 628419511.0, + "step": 16187 + }, + { + "epoch": 2.0592799898231777, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.396063804626465, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8973315954208374, + "num_tokens": 628465095.0, + "step": 16188 + }, + { + "epoch": 2.0594072001017683, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.529335975646973, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8968324661254883, + "num_tokens": 628501979.0, + "step": 16189 + }, + { + "epoch": 2.059534410380359, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.54047966003418, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.892660915851593, + "num_tokens": 628539093.0, + "step": 16190 + }, + { + "epoch": 2.0596616206589493, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.522573471069336, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8952627778053284, + "num_tokens": 628580087.0, + "step": 16191 + }, + { + "epoch": 2.05978883093754, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.508661270141602, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9041041135787964, + "num_tokens": 628619359.0, + "step": 16192 + }, + { + "epoch": 2.0599160412161304, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.422104835510254, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9038279056549072, + "num_tokens": 628658541.0, + "step": 16193 + }, + { + "epoch": 2.060043251494721, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.595458030700684, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8990309238433838, + "num_tokens": 628692819.0, + "step": 16194 + }, + { + "epoch": 2.0601704617733114, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.505661010742188, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8972164988517761, + "num_tokens": 628728951.0, + "step": 16195 + }, + { + "epoch": 2.060297672051902, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.458016395568848, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8943650722503662, + "num_tokens": 628769376.0, + "step": 16196 + }, + { + "epoch": 2.0604248823304925, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.512272834777832, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.8978205919265747, + "num_tokens": 628812923.0, + "step": 16197 + }, + { + "epoch": 2.060552092609083, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.392929077148438, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9025134444236755, + "num_tokens": 628858356.0, + "step": 16198 + }, + { + "epoch": 2.0606793028876735, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.469935417175293, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9094098210334778, + "num_tokens": 628901098.0, + "step": 16199 + }, + { + "epoch": 2.0608065131662636, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.516566276550293, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8986951112747192, + "num_tokens": 628940101.0, + "step": 16200 + }, + { + "epoch": 2.060933723444854, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.512908935546875, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.9061422348022461, + "num_tokens": 628976960.0, + "step": 16201 + }, + { + "epoch": 2.0610609337234447, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.540937423706055, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9027159214019775, + "num_tokens": 629019786.0, + "step": 16202 + }, + { + "epoch": 2.061188144002035, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.559669494628906, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8942923545837402, + "num_tokens": 629054829.0, + "step": 16203 + }, + { + "epoch": 2.0613153542806257, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.470404624938965, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9145852327346802, + "num_tokens": 629091951.0, + "step": 16204 + }, + { + "epoch": 2.0614425645592163, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.43125057220459, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9107363224029541, + "num_tokens": 629131468.0, + "step": 16205 + }, + { + "epoch": 2.061569774837807, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.452116012573242, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.911038875579834, + "num_tokens": 629165647.0, + "step": 16206 + }, + { + "epoch": 2.0616969851163973, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.574861526489258, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.890680193901062, + "num_tokens": 629205435.0, + "step": 16207 + }, + { + "epoch": 2.061824195394988, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.568754196166992, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.9039828181266785, + "num_tokens": 629241916.0, + "step": 16208 + }, + { + "epoch": 2.0619514056735784, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.449175834655762, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.9003973007202148, + "num_tokens": 629285566.0, + "step": 16209 + }, + { + "epoch": 2.062078615952169, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.598733901977539, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8899933099746704, + "num_tokens": 629322831.0, + "step": 16210 + }, + { + "epoch": 2.0622058262307594, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.632061004638672, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8905376195907593, + "num_tokens": 629358734.0, + "step": 16211 + }, + { + "epoch": 2.06233303650935, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.499415397644043, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.906524658203125, + "num_tokens": 629395162.0, + "step": 16212 + }, + { + "epoch": 2.0624602467879405, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.577982902526855, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8956767320632935, + "num_tokens": 629432659.0, + "step": 16213 + }, + { + "epoch": 2.062587457066531, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.428092956542969, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9063667058944702, + "num_tokens": 629470853.0, + "step": 16214 + }, + { + "epoch": 2.0627146673451215, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.515668869018555, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8914682865142822, + "num_tokens": 629508186.0, + "step": 16215 + }, + { + "epoch": 2.062841877623712, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.465302467346191, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.9060772657394409, + "num_tokens": 629545598.0, + "step": 16216 + }, + { + "epoch": 2.0629690879023026, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.631184577941895, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8913885354995728, + "num_tokens": 629580804.0, + "step": 16217 + }, + { + "epoch": 2.063096298180893, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.454609870910645, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9017183780670166, + "num_tokens": 629619147.0, + "step": 16218 + }, + { + "epoch": 2.0632235084594837, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.433720588684082, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.9072800874710083, + "num_tokens": 629661075.0, + "step": 16219 + }, + { + "epoch": 2.063350718738074, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.536795616149902, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.9019893407821655, + "num_tokens": 629702760.0, + "step": 16220 + }, + { + "epoch": 2.0634779290166647, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.493987083435059, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8992953896522522, + "num_tokens": 629746744.0, + "step": 16221 + }, + { + "epoch": 2.0636051392952552, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.419160842895508, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8948154449462891, + "num_tokens": 629792357.0, + "step": 16222 + }, + { + "epoch": 2.0637323495738458, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.616776466369629, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8979922533035278, + "num_tokens": 629828898.0, + "step": 16223 + }, + { + "epoch": 2.0638595598524363, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.511309623718262, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9123497009277344, + "num_tokens": 629865423.0, + "step": 16224 + }, + { + "epoch": 2.0639867701310264, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.427491188049316, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9014256596565247, + "num_tokens": 629908777.0, + "step": 16225 + }, + { + "epoch": 2.064113980409617, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.47947883605957, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9118890166282654, + "num_tokens": 629950725.0, + "step": 16226 + }, + { + "epoch": 2.0642411906882074, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.46618366241455, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9120213985443115, + "num_tokens": 629990200.0, + "step": 16227 + }, + { + "epoch": 2.064368400966798, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.547745704650879, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8982418775558472, + "num_tokens": 630025828.0, + "step": 16228 + }, + { + "epoch": 2.0644956112453885, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.573171615600586, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8896223902702332, + "num_tokens": 630066697.0, + "step": 16229 + }, + { + "epoch": 2.064622821523979, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.516989707946777, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8874331116676331, + "num_tokens": 630103791.0, + "step": 16230 + }, + { + "epoch": 2.0647500318025696, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.58203125, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9076548218727112, + "num_tokens": 630139633.0, + "step": 16231 + }, + { + "epoch": 2.06487724208116, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.552681922912598, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8982944488525391, + "num_tokens": 630174904.0, + "step": 16232 + }, + { + "epoch": 2.0650044523597506, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.537623405456543, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8847504258155823, + "num_tokens": 630216073.0, + "step": 16233 + }, + { + "epoch": 2.065131662638341, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.392889976501465, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9007563591003418, + "num_tokens": 630266327.0, + "step": 16234 + }, + { + "epoch": 2.0652588729169317, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.471936225891113, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.9004209637641907, + "num_tokens": 630305770.0, + "step": 16235 + }, + { + "epoch": 2.065386083195522, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.419209480285645, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9070401191711426, + "num_tokens": 630344294.0, + "step": 16236 + }, + { + "epoch": 2.0655132934741127, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.51321029663086, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9033606052398682, + "num_tokens": 630380492.0, + "step": 16237 + }, + { + "epoch": 2.0656405037527032, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.493024826049805, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9062144756317139, + "num_tokens": 630418938.0, + "step": 16238 + }, + { + "epoch": 2.065767714031294, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.454438209533691, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9027079343795776, + "num_tokens": 630453799.0, + "step": 16239 + }, + { + "epoch": 2.0658949243098843, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.445281028747559, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.9006701707839966, + "num_tokens": 630500304.0, + "step": 16240 + }, + { + "epoch": 2.066022134588475, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.200241088867188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.459314346313477, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9106361865997314, + "num_tokens": 630539867.0, + "step": 16241 + }, + { + "epoch": 2.0661493448670654, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.558647155761719, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9058730602264404, + "num_tokens": 630576918.0, + "step": 16242 + }, + { + "epoch": 2.066276555145656, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.58472728729248, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.9007635712623596, + "num_tokens": 630610669.0, + "step": 16243 + }, + { + "epoch": 2.0664037654242464, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.442898750305176, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.9011477828025818, + "num_tokens": 630655174.0, + "step": 16244 + }, + { + "epoch": 2.066530975702837, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.537087440490723, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8988164067268372, + "num_tokens": 630694852.0, + "step": 16245 + }, + { + "epoch": 2.0666581859814275, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.55930233001709, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8899900317192078, + "num_tokens": 630732642.0, + "step": 16246 + }, + { + "epoch": 2.066785396260018, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.508204460144043, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8974576592445374, + "num_tokens": 630770074.0, + "step": 16247 + }, + { + "epoch": 2.0669126065386085, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.487682342529297, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.9007031917572021, + "num_tokens": 630811320.0, + "step": 16248 + }, + { + "epoch": 2.0670398168171986, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.424159049987793, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.9061634540557861, + "num_tokens": 630853008.0, + "step": 16249 + }, + { + "epoch": 2.067167027095789, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.443022727966309, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9099371433258057, + "num_tokens": 630889924.0, + "step": 16250 + }, + { + "epoch": 2.0672942373743797, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.417853355407715, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9047784805297852, + "num_tokens": 630933876.0, + "step": 16251 + }, + { + "epoch": 2.06742144765297, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.458715438842773, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.9029404520988464, + "num_tokens": 630973997.0, + "step": 16252 + }, + { + "epoch": 2.0675486579315607, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.538531303405762, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.9008393287658691, + "num_tokens": 631010321.0, + "step": 16253 + }, + { + "epoch": 2.0676758682101513, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.485280990600586, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.909588098526001, + "num_tokens": 631045566.0, + "step": 16254 + }, + { + "epoch": 2.067803078488742, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.562931060791016, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.9001295566558838, + "num_tokens": 631086263.0, + "step": 16255 + }, + { + "epoch": 2.0679302887673323, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.469066619873047, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.9008886814117432, + "num_tokens": 631124387.0, + "step": 16256 + }, + { + "epoch": 2.068057499045923, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.507781028747559, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8937958478927612, + "num_tokens": 631163737.0, + "step": 16257 + }, + { + "epoch": 2.0681847093245134, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.520281791687012, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9010608792304993, + "num_tokens": 631201188.0, + "step": 16258 + }, + { + "epoch": 2.068311919603104, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.511992454528809, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8996655344963074, + "num_tokens": 631238891.0, + "step": 16259 + }, + { + "epoch": 2.0684391298816944, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.577999114990234, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9000132083892822, + "num_tokens": 631277837.0, + "step": 16260 + }, + { + "epoch": 2.068566340160285, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.429755210876465, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9053525328636169, + "num_tokens": 631318833.0, + "step": 16261 + }, + { + "epoch": 2.0686935504388755, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.421433448791504, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.9053475856781006, + "num_tokens": 631362214.0, + "step": 16262 + }, + { + "epoch": 2.068820760717466, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.487198829650879, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9035016894340515, + "num_tokens": 631397656.0, + "step": 16263 + }, + { + "epoch": 2.0689479709960565, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.495038032531738, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9121412038803101, + "num_tokens": 631436705.0, + "step": 16264 + }, + { + "epoch": 2.069075181274647, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.615569114685059, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8988145589828491, + "num_tokens": 631470390.0, + "step": 16265 + }, + { + "epoch": 2.0692023915532376, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.469805717468262, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8948467373847961, + "num_tokens": 631511698.0, + "step": 16266 + }, + { + "epoch": 2.069329601831828, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.62389087677002, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8913713097572327, + "num_tokens": 631547299.0, + "step": 16267 + }, + { + "epoch": 2.0694568121104187, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.56010913848877, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9058652520179749, + "num_tokens": 631580792.0, + "step": 16268 + }, + { + "epoch": 2.069584022389009, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.615838050842285, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8789269924163818, + "num_tokens": 631618500.0, + "step": 16269 + }, + { + "epoch": 2.0697112326675997, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.59355640411377, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8934616446495056, + "num_tokens": 631655272.0, + "step": 16270 + }, + { + "epoch": 2.0698384429461902, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.554595947265625, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8912522792816162, + "num_tokens": 631692827.0, + "step": 16271 + }, + { + "epoch": 2.0699656532247808, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.471830368041992, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.9071260690689087, + "num_tokens": 631729077.0, + "step": 16272 + }, + { + "epoch": 2.070092863503371, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.533172607421875, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8925303220748901, + "num_tokens": 631772292.0, + "step": 16273 + }, + { + "epoch": 2.0702200737819614, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.660886764526367, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8980481624603271, + "num_tokens": 631804188.0, + "step": 16274 + }, + { + "epoch": 2.070347284060552, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.454642295837402, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8980806469917297, + "num_tokens": 631848428.0, + "step": 16275 + }, + { + "epoch": 2.0704744943391424, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.648048400878906, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8943840265274048, + "num_tokens": 631880387.0, + "step": 16276 + }, + { + "epoch": 2.070601704617733, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.558409690856934, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.899169921875, + "num_tokens": 631921314.0, + "step": 16277 + }, + { + "epoch": 2.0707289148963235, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.560474395751953, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8987109661102295, + "num_tokens": 631956816.0, + "step": 16278 + }, + { + "epoch": 2.070856125174914, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.474681854248047, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.908377468585968, + "num_tokens": 632000212.0, + "step": 16279 + }, + { + "epoch": 2.0709833354535045, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.578096389770508, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9055721759796143, + "num_tokens": 632032889.0, + "step": 16280 + }, + { + "epoch": 2.071110545732095, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.58397388458252, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.9053977131843567, + "num_tokens": 632070527.0, + "step": 16281 + }, + { + "epoch": 2.0712377560106856, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.54823112487793, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.901231586933136, + "num_tokens": 632108337.0, + "step": 16282 + }, + { + "epoch": 2.071364966289276, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.576385498046875, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8998363018035889, + "num_tokens": 632143781.0, + "step": 16283 + }, + { + "epoch": 2.0714921765678667, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.592001914978027, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.9024548530578613, + "num_tokens": 632181172.0, + "step": 16284 + }, + { + "epoch": 2.071619386846457, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.707502365112305, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8793257474899292, + "num_tokens": 632212537.0, + "step": 16285 + }, + { + "epoch": 2.0717465971250477, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.596179962158203, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8844685554504395, + "num_tokens": 632252869.0, + "step": 16286 + }, + { + "epoch": 2.0718738074036382, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.414202690124512, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9073611497879028, + "num_tokens": 632290184.0, + "step": 16287 + }, + { + "epoch": 2.0720010176822288, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.479467391967773, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9117221236228943, + "num_tokens": 632331727.0, + "step": 16288 + }, + { + "epoch": 2.0721282279608193, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.483185768127441, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9138981103897095, + "num_tokens": 632376469.0, + "step": 16289 + }, + { + "epoch": 2.07225543823941, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.606627464294434, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9012216925621033, + "num_tokens": 632412749.0, + "step": 16290 + }, + { + "epoch": 2.0723826485180004, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.529231071472168, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.9034090042114258, + "num_tokens": 632448541.0, + "step": 16291 + }, + { + "epoch": 2.072509858796591, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.510285377502441, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8976006507873535, + "num_tokens": 632489955.0, + "step": 16292 + }, + { + "epoch": 2.0726370690751814, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.497318267822266, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9012219905853271, + "num_tokens": 632528351.0, + "step": 16293 + }, + { + "epoch": 2.072764279353772, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.55669116973877, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9092649817466736, + "num_tokens": 632566990.0, + "step": 16294 + }, + { + "epoch": 2.0728914896323625, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.51892375946045, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8987550139427185, + "num_tokens": 632608648.0, + "step": 16295 + }, + { + "epoch": 2.073018699910953, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.45082950592041, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9113836288452148, + "num_tokens": 632645829.0, + "step": 16296 + }, + { + "epoch": 2.0731459101895435, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.568771362304688, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9054465293884277, + "num_tokens": 632686024.0, + "step": 16297 + }, + { + "epoch": 2.0732731204681336, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.551025390625, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.9034286141395569, + "num_tokens": 632726408.0, + "step": 16298 + }, + { + "epoch": 2.073400330746724, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.534661293029785, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8985244035720825, + "num_tokens": 632766364.0, + "step": 16299 + }, + { + "epoch": 2.0735275410253147, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.634180068969727, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8875291347503662, + "num_tokens": 632806579.0, + "step": 16300 + }, + { + "epoch": 2.073654751303905, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.557498931884766, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8944776058197021, + "num_tokens": 632847603.0, + "step": 16301 + }, + { + "epoch": 2.0737819615824957, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.560944557189941, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.892505943775177, + "num_tokens": 632889743.0, + "step": 16302 + }, + { + "epoch": 2.0739091718610863, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.457990646362305, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.9086276292800903, + "num_tokens": 632932170.0, + "step": 16303 + }, + { + "epoch": 2.074036382139677, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.566640853881836, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.906285285949707, + "num_tokens": 632967842.0, + "step": 16304 + }, + { + "epoch": 2.0741635924182673, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.51376724243164, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9092154502868652, + "num_tokens": 633007188.0, + "step": 16305 + }, + { + "epoch": 2.074290802696858, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.513604164123535, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.896237850189209, + "num_tokens": 633047573.0, + "step": 16306 + }, + { + "epoch": 2.0744180129754484, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.653633117675781, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8936705589294434, + "num_tokens": 633080557.0, + "step": 16307 + }, + { + "epoch": 2.074545223254039, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.458498001098633, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9088371992111206, + "num_tokens": 633120788.0, + "step": 16308 + }, + { + "epoch": 2.0746724335326294, + "ewc_loss": 0.0185546875, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.475841522216797, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9081529378890991, + "num_tokens": 633161832.0, + "step": 16309 + }, + { + "epoch": 2.07479964381122, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.470891952514648, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.905490517616272, + "num_tokens": 633196185.0, + "step": 16310 + }, + { + "epoch": 2.0749268540898105, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.55627727508545, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8973218202590942, + "num_tokens": 633238497.0, + "step": 16311 + }, + { + "epoch": 2.075054064368401, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.611743927001953, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8967471718788147, + "num_tokens": 633272353.0, + "step": 16312 + }, + { + "epoch": 2.0751812746469915, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.55927848815918, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9005287289619446, + "num_tokens": 633303918.0, + "step": 16313 + }, + { + "epoch": 2.075308484925582, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.52656078338623, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.9004248380661011, + "num_tokens": 633337473.0, + "step": 16314 + }, + { + "epoch": 2.0754356952041726, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.555246353149414, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8959608674049377, + "num_tokens": 633375201.0, + "step": 16315 + }, + { + "epoch": 2.075562905482763, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.466578483581543, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8875247240066528, + "num_tokens": 633418030.0, + "step": 16316 + }, + { + "epoch": 2.0756901157613536, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.420674324035645, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9134389162063599, + "num_tokens": 633455135.0, + "step": 16317 + }, + { + "epoch": 2.075817326039944, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.482110023498535, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8976762890815735, + "num_tokens": 633493136.0, + "step": 16318 + }, + { + "epoch": 2.0759445363185347, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.69154167175293, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.898864209651947, + "num_tokens": 633533173.0, + "step": 16319 + }, + { + "epoch": 2.0760717465971252, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.44788646697998, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9050189256668091, + "num_tokens": 633570975.0, + "step": 16320 + }, + { + "epoch": 2.0761989568757158, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.437126159667969, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.9074263572692871, + "num_tokens": 633619650.0, + "step": 16321 + }, + { + "epoch": 2.0763261671543063, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.456470489501953, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9074896574020386, + "num_tokens": 633661098.0, + "step": 16322 + }, + { + "epoch": 2.0764533774328964, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.566485404968262, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.9018130302429199, + "num_tokens": 633701604.0, + "step": 16323 + }, + { + "epoch": 2.076580587711487, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.499344825744629, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.9050156474113464, + "num_tokens": 633744659.0, + "step": 16324 + }, + { + "epoch": 2.0767077979900774, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.570008277893066, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8952978849411011, + "num_tokens": 633784966.0, + "step": 16325 + }, + { + "epoch": 2.076835008268668, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.622367858886719, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.9047491550445557, + "num_tokens": 633817931.0, + "step": 16326 + }, + { + "epoch": 2.0769622185472585, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.549921989440918, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9103670120239258, + "num_tokens": 633852470.0, + "step": 16327 + }, + { + "epoch": 2.077089428825849, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.439041137695312, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9112330675125122, + "num_tokens": 633893461.0, + "step": 16328 + }, + { + "epoch": 2.0772166391044395, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.723304748535156, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8926533460617065, + "num_tokens": 633931581.0, + "step": 16329 + }, + { + "epoch": 2.07734384938303, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.536184310913086, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8943246603012085, + "num_tokens": 633970089.0, + "step": 16330 + }, + { + "epoch": 2.0774710596616206, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.52746868133545, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8974891901016235, + "num_tokens": 634009259.0, + "step": 16331 + }, + { + "epoch": 2.077598269940211, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.522751808166504, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.9018692374229431, + "num_tokens": 634050902.0, + "step": 16332 + }, + { + "epoch": 2.0777254802188017, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.554595947265625, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8997029066085815, + "num_tokens": 634086886.0, + "step": 16333 + }, + { + "epoch": 2.077852690497392, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.519225120544434, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.9007452726364136, + "num_tokens": 634125124.0, + "step": 16334 + }, + { + "epoch": 2.0779799007759827, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.520400047302246, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8944810032844543, + "num_tokens": 634164550.0, + "step": 16335 + }, + { + "epoch": 2.0781071110545732, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.553938865661621, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.903159499168396, + "num_tokens": 634200386.0, + "step": 16336 + }, + { + "epoch": 2.0782343213331638, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.520039558410645, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.9009712934494019, + "num_tokens": 634240011.0, + "step": 16337 + }, + { + "epoch": 2.0783615316117543, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.48469352722168, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.884386420249939, + "num_tokens": 634282199.0, + "step": 16338 + }, + { + "epoch": 2.078488741890345, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.56446361541748, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9014676809310913, + "num_tokens": 634318872.0, + "step": 16339 + }, + { + "epoch": 2.0786159521689354, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.622559547424316, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8841641545295715, + "num_tokens": 634354981.0, + "step": 16340 + }, + { + "epoch": 2.078743162447526, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.499972343444824, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9047209620475769, + "num_tokens": 634394408.0, + "step": 16341 + }, + { + "epoch": 2.0788703727261164, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.632586479187012, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8858686685562134, + "num_tokens": 634431994.0, + "step": 16342 + }, + { + "epoch": 2.078997583004707, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.463184356689453, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.9019632339477539, + "num_tokens": 634471233.0, + "step": 16343 + }, + { + "epoch": 2.0791247932832975, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.538350105285645, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8953612446784973, + "num_tokens": 634508702.0, + "step": 16344 + }, + { + "epoch": 2.079252003561888, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.2479248046875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.451491355895996, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9138095378875732, + "num_tokens": 634548255.0, + "step": 16345 + }, + { + "epoch": 2.0793792138404785, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.615222930908203, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8984249234199524, + "num_tokens": 634584326.0, + "step": 16346 + }, + { + "epoch": 2.0795064241190686, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.465478897094727, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9084570407867432, + "num_tokens": 634627136.0, + "step": 16347 + }, + { + "epoch": 2.079633634397659, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.44853687286377, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8873318433761597, + "num_tokens": 634671308.0, + "step": 16348 + }, + { + "epoch": 2.0797608446762497, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.551423072814941, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.901943564414978, + "num_tokens": 634710585.0, + "step": 16349 + }, + { + "epoch": 2.07988805495484, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.445531845092773, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.889193058013916, + "num_tokens": 634755852.0, + "step": 16350 + }, + { + "epoch": 2.0800152652334307, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.549948692321777, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8974910974502563, + "num_tokens": 634796697.0, + "step": 16351 + }, + { + "epoch": 2.0801424755120212, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.535307884216309, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8962582349777222, + "num_tokens": 634832910.0, + "step": 16352 + }, + { + "epoch": 2.0802696857906118, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.534101486206055, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.9004114270210266, + "num_tokens": 634872916.0, + "step": 16353 + }, + { + "epoch": 2.0803968960692023, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.574487686157227, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8871323466300964, + "num_tokens": 634908812.0, + "step": 16354 + }, + { + "epoch": 2.080524106347793, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.484224319458008, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.9022101163864136, + "num_tokens": 634948629.0, + "step": 16355 + }, + { + "epoch": 2.0806513166263834, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.59225082397461, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.892820417881012, + "num_tokens": 634985859.0, + "step": 16356 + }, + { + "epoch": 2.080778526904974, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.558370590209961, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.9006381630897522, + "num_tokens": 635023017.0, + "step": 16357 + }, + { + "epoch": 2.0809057371835644, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.472084999084473, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9072128534317017, + "num_tokens": 635064207.0, + "step": 16358 + }, + { + "epoch": 2.081032947462155, + "ewc_loss": 0.0186767578125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.506709098815918, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8980337381362915, + "num_tokens": 635106697.0, + "step": 16359 + }, + { + "epoch": 2.0811601577407455, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.543581008911133, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8976284265518188, + "num_tokens": 635144397.0, + "step": 16360 + }, + { + "epoch": 2.081287368019336, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.504088401794434, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8935727477073669, + "num_tokens": 635185050.0, + "step": 16361 + }, + { + "epoch": 2.0814145782979265, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.618903160095215, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8939272165298462, + "num_tokens": 635221352.0, + "step": 16362 + }, + { + "epoch": 2.081541788576517, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.607855796813965, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9040740728378296, + "num_tokens": 635254146.0, + "step": 16363 + }, + { + "epoch": 2.0816689988551076, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.542692184448242, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9026583433151245, + "num_tokens": 635292852.0, + "step": 16364 + }, + { + "epoch": 2.081796209133698, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.684083938598633, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8882217407226562, + "num_tokens": 635328610.0, + "step": 16365 + }, + { + "epoch": 2.0819234194122886, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.54959487915039, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8836177587509155, + "num_tokens": 635369663.0, + "step": 16366 + }, + { + "epoch": 2.082050629690879, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.484513282775879, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.906019389629364, + "num_tokens": 635411422.0, + "step": 16367 + }, + { + "epoch": 2.0821778399694697, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.562500953674316, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8933432698249817, + "num_tokens": 635446855.0, + "step": 16368 + }, + { + "epoch": 2.0823050502480602, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.552642822265625, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.9059590101242065, + "num_tokens": 635484498.0, + "step": 16369 + }, + { + "epoch": 2.0824322605266508, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.602984428405762, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8987723588943481, + "num_tokens": 635521193.0, + "step": 16370 + }, + { + "epoch": 2.082559470805241, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.516768455505371, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8999927639961243, + "num_tokens": 635562557.0, + "step": 16371 + }, + { + "epoch": 2.0826866810838314, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.67789363861084, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9030355215072632, + "num_tokens": 635596427.0, + "step": 16372 + }, + { + "epoch": 2.082813891362422, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.533722877502441, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.885500431060791, + "num_tokens": 635644918.0, + "step": 16373 + }, + { + "epoch": 2.0829411016410124, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.378918647766113, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9072420001029968, + "num_tokens": 635689693.0, + "step": 16374 + }, + { + "epoch": 2.083068311919603, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.714323043823242, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9085391759872437, + "num_tokens": 635720784.0, + "step": 16375 + }, + { + "epoch": 2.0831955221981935, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.539679527282715, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8936654329299927, + "num_tokens": 635765170.0, + "step": 16376 + }, + { + "epoch": 2.083322732476784, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.515653610229492, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.90000319480896, + "num_tokens": 635803078.0, + "step": 16377 + }, + { + "epoch": 2.0834499427553745, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.561622619628906, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8912879824638367, + "num_tokens": 635841763.0, + "step": 16378 + }, + { + "epoch": 2.083577153033965, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.577055931091309, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.9009342193603516, + "num_tokens": 635882977.0, + "step": 16379 + }, + { + "epoch": 2.0837043633125556, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.648524284362793, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9013457298278809, + "num_tokens": 635917758.0, + "step": 16380 + }, + { + "epoch": 2.083831573591146, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.57716178894043, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8931267261505127, + "num_tokens": 635956680.0, + "step": 16381 + }, + { + "epoch": 2.0839587838697367, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.51328182220459, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9111849069595337, + "num_tokens": 635993296.0, + "step": 16382 + }, + { + "epoch": 2.084085994148327, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.494685173034668, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9109523892402649, + "num_tokens": 636031973.0, + "step": 16383 + }, + { + "epoch": 2.0842132044269177, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.615677833557129, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8934174180030823, + "num_tokens": 636070527.0, + "step": 16384 + }, + { + "epoch": 2.0843404147055082, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.480785369873047, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9067668318748474, + "num_tokens": 636109531.0, + "step": 16385 + }, + { + "epoch": 2.0844676249840988, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.517768859863281, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.905457615852356, + "num_tokens": 636148627.0, + "step": 16386 + }, + { + "epoch": 2.0845948352626893, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.590911865234375, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.9070590734481812, + "num_tokens": 636186146.0, + "step": 16387 + }, + { + "epoch": 2.08472204554128, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.717470169067383, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9044597744941711, + "num_tokens": 636221176.0, + "step": 16388 + }, + { + "epoch": 2.0848492558198704, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.558028221130371, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8973366618156433, + "num_tokens": 636255828.0, + "step": 16389 + }, + { + "epoch": 2.084976466098461, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.581096649169922, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9080923199653625, + "num_tokens": 636290919.0, + "step": 16390 + }, + { + "epoch": 2.0851036763770514, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.50285530090332, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.9063459038734436, + "num_tokens": 636330601.0, + "step": 16391 + }, + { + "epoch": 2.085230886655642, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.438793182373047, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.9045856595039368, + "num_tokens": 636372363.0, + "step": 16392 + }, + { + "epoch": 2.0853580969342325, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.555526733398438, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8933550119400024, + "num_tokens": 636406040.0, + "step": 16393 + }, + { + "epoch": 2.085485307212823, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.63071346282959, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8882748484611511, + "num_tokens": 636447736.0, + "step": 16394 + }, + { + "epoch": 2.0856125174914135, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.511866569519043, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.9034502506256104, + "num_tokens": 636484194.0, + "step": 16395 + }, + { + "epoch": 2.0857397277700036, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.466670989990234, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9093204736709595, + "num_tokens": 636528610.0, + "step": 16396 + }, + { + "epoch": 2.085866938048594, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.606531143188477, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.9020583629608154, + "num_tokens": 636564330.0, + "step": 16397 + }, + { + "epoch": 2.0859941483271847, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.474891662597656, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9100184440612793, + "num_tokens": 636602752.0, + "step": 16398 + }, + { + "epoch": 2.086121358605775, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.553890228271484, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.895505428314209, + "num_tokens": 636640974.0, + "step": 16399 + }, + { + "epoch": 2.0862485688843657, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.715452194213867, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8866350650787354, + "num_tokens": 636680506.0, + "step": 16400 + }, + { + "epoch": 2.0863757791629562, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.554920196533203, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8980775475502014, + "num_tokens": 636721745.0, + "step": 16401 + }, + { + "epoch": 2.0865029894415468, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.563273429870605, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.9053617119789124, + "num_tokens": 636758674.0, + "step": 16402 + }, + { + "epoch": 2.0866301997201373, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.47398853302002, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.9012109637260437, + "num_tokens": 636800388.0, + "step": 16403 + }, + { + "epoch": 2.086757409998728, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.484541893005371, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.9023217558860779, + "num_tokens": 636841894.0, + "step": 16404 + }, + { + "epoch": 2.0868846202773184, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.659470558166504, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.898774266242981, + "num_tokens": 636877320.0, + "step": 16405 + }, + { + "epoch": 2.087011830555909, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.605999946594238, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9145326614379883, + "num_tokens": 636917722.0, + "step": 16406 + }, + { + "epoch": 2.0871390408344994, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.531900405883789, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8827171325683594, + "num_tokens": 636957626.0, + "step": 16407 + }, + { + "epoch": 2.08726625111309, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.59818172454834, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.883460521697998, + "num_tokens": 636998308.0, + "step": 16408 + }, + { + "epoch": 2.0873934613916805, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.715574264526367, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8972809314727783, + "num_tokens": 637033224.0, + "step": 16409 + }, + { + "epoch": 2.087520671670271, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.533926010131836, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8856961727142334, + "num_tokens": 637073298.0, + "step": 16410 + }, + { + "epoch": 2.0876478819488615, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.553098678588867, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8874720335006714, + "num_tokens": 637114474.0, + "step": 16411 + }, + { + "epoch": 2.087775092227452, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.499651908874512, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8854061365127563, + "num_tokens": 637152857.0, + "step": 16412 + }, + { + "epoch": 2.0879023025060426, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.725543975830078, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8877792358398438, + "num_tokens": 637184274.0, + "step": 16413 + }, + { + "epoch": 2.088029512784633, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.471614837646484, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.9024940729141235, + "num_tokens": 637224035.0, + "step": 16414 + }, + { + "epoch": 2.0881567230632236, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.346541404724121, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9138157963752747, + "num_tokens": 637268667.0, + "step": 16415 + }, + { + "epoch": 2.088283933341814, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.54232406616211, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9056053757667542, + "num_tokens": 637307147.0, + "step": 16416 + }, + { + "epoch": 2.0884111436204047, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.543058395385742, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9041085243225098, + "num_tokens": 637342066.0, + "step": 16417 + }, + { + "epoch": 2.0885383538989952, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.467473983764648, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8929822444915771, + "num_tokens": 637394748.0, + "step": 16418 + }, + { + "epoch": 2.0886655641775858, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.504623413085938, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8976291418075562, + "num_tokens": 637439735.0, + "step": 16419 + }, + { + "epoch": 2.0887927744561763, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.619754791259766, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8784564137458801, + "num_tokens": 637479923.0, + "step": 16420 + }, + { + "epoch": 2.0889199847347664, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.49448299407959, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8927391767501831, + "num_tokens": 637518144.0, + "step": 16421 + }, + { + "epoch": 2.089047195013357, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.4796142578125, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.908548891544342, + "num_tokens": 637556384.0, + "step": 16422 + }, + { + "epoch": 2.0891744052919474, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.517961502075195, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8947467803955078, + "num_tokens": 637598238.0, + "step": 16423 + }, + { + "epoch": 2.089301615570538, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.547645568847656, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.9017676115036011, + "num_tokens": 637638397.0, + "step": 16424 + }, + { + "epoch": 2.0894288258491285, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.538166046142578, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9047984480857849, + "num_tokens": 637675561.0, + "step": 16425 + }, + { + "epoch": 2.089556036127719, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.496103286743164, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8821467161178589, + "num_tokens": 637721225.0, + "step": 16426 + }, + { + "epoch": 2.0896832464063095, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.444364547729492, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8940714597702026, + "num_tokens": 637767993.0, + "step": 16427 + }, + { + "epoch": 2.0898104566849, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.569060325622559, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.9029010534286499, + "num_tokens": 637801665.0, + "step": 16428 + }, + { + "epoch": 2.0899376669634906, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.564574241638184, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8885233402252197, + "num_tokens": 637841729.0, + "step": 16429 + }, + { + "epoch": 2.090064877242081, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.452163696289062, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8914784789085388, + "num_tokens": 637886104.0, + "step": 16430 + }, + { + "epoch": 2.0901920875206716, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.647015571594238, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8893086910247803, + "num_tokens": 637922403.0, + "step": 16431 + }, + { + "epoch": 2.090319297799262, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.527152061462402, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8961877822875977, + "num_tokens": 637961224.0, + "step": 16432 + }, + { + "epoch": 2.0904465080778527, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.514249801635742, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8947547674179077, + "num_tokens": 638001620.0, + "step": 16433 + }, + { + "epoch": 2.0905737183564432, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.397547721862793, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.9028129577636719, + "num_tokens": 638046772.0, + "step": 16434 + }, + { + "epoch": 2.0907009286350338, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.54184341430664, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9064880609512329, + "num_tokens": 638082247.0, + "step": 16435 + }, + { + "epoch": 2.0908281389136243, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.550363540649414, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9007668495178223, + "num_tokens": 638117812.0, + "step": 16436 + }, + { + "epoch": 2.090955349192215, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.59414005279541, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9031099081039429, + "num_tokens": 638156970.0, + "step": 16437 + }, + { + "epoch": 2.0910825594708053, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.548344612121582, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.9065684080123901, + "num_tokens": 638193025.0, + "step": 16438 + }, + { + "epoch": 2.091209769749396, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.4812650680542, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9066177606582642, + "num_tokens": 638231287.0, + "step": 16439 + }, + { + "epoch": 2.0913369800279864, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.445236206054688, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.912135124206543, + "num_tokens": 638268260.0, + "step": 16440 + }, + { + "epoch": 2.091464190306577, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.694087982177734, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8944477438926697, + "num_tokens": 638301159.0, + "step": 16441 + }, + { + "epoch": 2.0915914005851675, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.470013618469238, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8956618905067444, + "num_tokens": 638346178.0, + "step": 16442 + }, + { + "epoch": 2.091718610863758, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.49752140045166, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.9023144841194153, + "num_tokens": 638389441.0, + "step": 16443 + }, + { + "epoch": 2.0918458211423485, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.488178253173828, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.907396674156189, + "num_tokens": 638427465.0, + "step": 16444 + }, + { + "epoch": 2.0919730314209386, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.597678184509277, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8931944370269775, + "num_tokens": 638470051.0, + "step": 16445 + }, + { + "epoch": 2.092100241699529, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.522260665893555, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9098410606384277, + "num_tokens": 638503958.0, + "step": 16446 + }, + { + "epoch": 2.0922274519781197, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.628386497497559, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9028285145759583, + "num_tokens": 638542578.0, + "step": 16447 + }, + { + "epoch": 2.09235466225671, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.582037925720215, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.9046946167945862, + "num_tokens": 638579080.0, + "step": 16448 + }, + { + "epoch": 2.0924818725353007, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.571752548217773, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.9012706875801086, + "num_tokens": 638620508.0, + "step": 16449 + }, + { + "epoch": 2.0926090828138912, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.590351104736328, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9066449403762817, + "num_tokens": 638658711.0, + "step": 16450 + }, + { + "epoch": 2.0927362930924818, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.589550018310547, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.9025866985321045, + "num_tokens": 638693754.0, + "step": 16451 + }, + { + "epoch": 2.0928635033710723, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.615774154663086, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.895999550819397, + "num_tokens": 638731119.0, + "step": 16452 + }, + { + "epoch": 2.092990713649663, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.511283874511719, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.9017770886421204, + "num_tokens": 638767827.0, + "step": 16453 + }, + { + "epoch": 2.0931179239282534, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.551158905029297, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8927900195121765, + "num_tokens": 638808203.0, + "step": 16454 + }, + { + "epoch": 2.093245134206844, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.589049339294434, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.9013873934745789, + "num_tokens": 638845157.0, + "step": 16455 + }, + { + "epoch": 2.0933723444854344, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.478896141052246, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9119033813476562, + "num_tokens": 638884508.0, + "step": 16456 + }, + { + "epoch": 2.093499554764025, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.452826499938965, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.9009565114974976, + "num_tokens": 638925166.0, + "step": 16457 + }, + { + "epoch": 2.0936267650426155, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.665801048278809, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8959591388702393, + "num_tokens": 638960110.0, + "step": 16458 + }, + { + "epoch": 2.093753975321206, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.559062004089355, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.9048458337783813, + "num_tokens": 638992864.0, + "step": 16459 + }, + { + "epoch": 2.0938811855997965, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.572609901428223, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8985500335693359, + "num_tokens": 639028351.0, + "step": 16460 + }, + { + "epoch": 2.094008395878387, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.53337574005127, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9065032601356506, + "num_tokens": 639070036.0, + "step": 16461 + }, + { + "epoch": 2.0941356061569776, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.546520233154297, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8978948593139648, + "num_tokens": 639106639.0, + "step": 16462 + }, + { + "epoch": 2.094262816435568, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.46118450164795, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.914935827255249, + "num_tokens": 639145362.0, + "step": 16463 + }, + { + "epoch": 2.0943900267141586, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.503456115722656, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.9043039083480835, + "num_tokens": 639180900.0, + "step": 16464 + }, + { + "epoch": 2.094517236992749, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.590052604675293, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8980700373649597, + "num_tokens": 639225944.0, + "step": 16465 + }, + { + "epoch": 2.0946444472713397, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.62519645690918, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8968397378921509, + "num_tokens": 639259728.0, + "step": 16466 + }, + { + "epoch": 2.09477165754993, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.542860984802246, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.9006864428520203, + "num_tokens": 639297773.0, + "step": 16467 + }, + { + "epoch": 2.0948988678285207, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.521188735961914, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.9014045596122742, + "num_tokens": 639337377.0, + "step": 16468 + }, + { + "epoch": 2.095026078107111, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.611503601074219, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.884770393371582, + "num_tokens": 639376678.0, + "step": 16469 + }, + { + "epoch": 2.0951532883857014, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.582841873168945, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.898141086101532, + "num_tokens": 639412344.0, + "step": 16470 + }, + { + "epoch": 2.095280498664292, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.549817085266113, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.9016618728637695, + "num_tokens": 639447325.0, + "step": 16471 + }, + { + "epoch": 2.0954077089428824, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.512207984924316, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8945631980895996, + "num_tokens": 639489271.0, + "step": 16472 + }, + { + "epoch": 2.095534919221473, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.458465576171875, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9019165635108948, + "num_tokens": 639529106.0, + "step": 16473 + }, + { + "epoch": 2.0956621295000635, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.614643096923828, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8846795558929443, + "num_tokens": 639567078.0, + "step": 16474 + }, + { + "epoch": 2.095789339778654, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.596789360046387, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.909805417060852, + "num_tokens": 639600406.0, + "step": 16475 + }, + { + "epoch": 2.0959165500572445, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.52517318725586, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9054611325263977, + "num_tokens": 639638945.0, + "step": 16476 + }, + { + "epoch": 2.096043760335835, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.525233268737793, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9006345272064209, + "num_tokens": 639676799.0, + "step": 16477 + }, + { + "epoch": 2.0961709706144256, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.591669082641602, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9137846231460571, + "num_tokens": 639710044.0, + "step": 16478 + }, + { + "epoch": 2.096298180893016, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.480355262756348, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8997129797935486, + "num_tokens": 639751989.0, + "step": 16479 + }, + { + "epoch": 2.0964253911716066, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.579183578491211, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8966392278671265, + "num_tokens": 639788408.0, + "step": 16480 + }, + { + "epoch": 2.096552601450197, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.568269729614258, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9052255749702454, + "num_tokens": 639823295.0, + "step": 16481 + }, + { + "epoch": 2.0966798117287877, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.573906898498535, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9120770692825317, + "num_tokens": 639859984.0, + "step": 16482 + }, + { + "epoch": 2.0968070220073782, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.642104148864746, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8972744941711426, + "num_tokens": 639897996.0, + "step": 16483 + }, + { + "epoch": 2.0969342322859688, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.525296211242676, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9105493426322937, + "num_tokens": 639934767.0, + "step": 16484 + }, + { + "epoch": 2.0970614425645593, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.557282447814941, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8866554498672485, + "num_tokens": 639975536.0, + "step": 16485 + }, + { + "epoch": 2.09718865284315, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.455073356628418, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9115127325057983, + "num_tokens": 640021061.0, + "step": 16486 + }, + { + "epoch": 2.0973158631217403, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.567997932434082, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9010372161865234, + "num_tokens": 640057448.0, + "step": 16487 + }, + { + "epoch": 2.097443073400331, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.570708274841309, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9049980640411377, + "num_tokens": 640090502.0, + "step": 16488 + }, + { + "epoch": 2.0975702836789214, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.620662689208984, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8837210536003113, + "num_tokens": 640129556.0, + "step": 16489 + }, + { + "epoch": 2.097697493957512, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.515527725219727, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.916434645652771, + "num_tokens": 640166064.0, + "step": 16490 + }, + { + "epoch": 2.0978247042361025, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.55010986328125, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8965996503829956, + "num_tokens": 640207794.0, + "step": 16491 + }, + { + "epoch": 2.097951914514693, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.62973690032959, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.9001213908195496, + "num_tokens": 640241339.0, + "step": 16492 + }, + { + "epoch": 2.0980791247932835, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.562804222106934, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9063609838485718, + "num_tokens": 640280873.0, + "step": 16493 + }, + { + "epoch": 2.0982063350718736, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.472825050354004, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8955234885215759, + "num_tokens": 640324304.0, + "step": 16494 + }, + { + "epoch": 2.098333545350464, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.544809341430664, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9036991000175476, + "num_tokens": 640360410.0, + "step": 16495 + }, + { + "epoch": 2.0984607556290547, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.677876472473145, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8889996409416199, + "num_tokens": 640395502.0, + "step": 16496 + }, + { + "epoch": 2.098587965907645, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.555549621582031, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8956419825553894, + "num_tokens": 640433704.0, + "step": 16497 + }, + { + "epoch": 2.0987151761862357, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.58529281616211, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9071316719055176, + "num_tokens": 640472448.0, + "step": 16498 + }, + { + "epoch": 2.0988423864648262, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.484282493591309, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.901652991771698, + "num_tokens": 640517605.0, + "step": 16499 + }, + { + "epoch": 2.0989695967434168, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.451322555541992, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.887614369392395, + "num_tokens": 640567196.0, + "step": 16500 + }, + { + "epoch": 2.0990968070220073, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.449674606323242, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9145406484603882, + "num_tokens": 640610503.0, + "step": 16501 + }, + { + "epoch": 2.099224017300598, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.536693572998047, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.9039980173110962, + "num_tokens": 640652680.0, + "step": 16502 + }, + { + "epoch": 2.0993512275791884, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.530200958251953, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9072023630142212, + "num_tokens": 640694071.0, + "step": 16503 + }, + { + "epoch": 2.099478437857779, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.610157012939453, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.9003922939300537, + "num_tokens": 640727908.0, + "step": 16504 + }, + { + "epoch": 2.0996056481363694, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.506241798400879, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.9077521562576294, + "num_tokens": 640765490.0, + "step": 16505 + }, + { + "epoch": 2.09973285841496, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.50804615020752, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8963409662246704, + "num_tokens": 640808648.0, + "step": 16506 + }, + { + "epoch": 2.0998600686935505, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.556222915649414, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.906857967376709, + "num_tokens": 640841816.0, + "step": 16507 + }, + { + "epoch": 2.099987278972141, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.529464721679688, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8824379444122314, + "num_tokens": 640882715.0, + "step": 16508 + }, + { + "epoch": 2.1001144892507315, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.542466163635254, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9098124504089355, + "num_tokens": 640915556.0, + "step": 16509 + }, + { + "epoch": 2.100241699529322, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.674975395202637, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8963847160339355, + "num_tokens": 640950005.0, + "step": 16510 + }, + { + "epoch": 2.1003689098079126, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.478951454162598, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9036362171173096, + "num_tokens": 640988060.0, + "step": 16511 + }, + { + "epoch": 2.100496120086503, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.616048812866211, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8982712626457214, + "num_tokens": 641026531.0, + "step": 16512 + }, + { + "epoch": 2.1006233303650936, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.52316951751709, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9103298187255859, + "num_tokens": 641059581.0, + "step": 16513 + }, + { + "epoch": 2.100750540643684, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.569085121154785, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.9005922675132751, + "num_tokens": 641100122.0, + "step": 16514 + }, + { + "epoch": 2.1008777509222747, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.569437980651855, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.894223690032959, + "num_tokens": 641139829.0, + "step": 16515 + }, + { + "epoch": 2.101004961200865, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.295608520507812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.519888877868652, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8912475109100342, + "num_tokens": 641183883.0, + "step": 16516 + }, + { + "epoch": 2.1011321714794557, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.582517623901367, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8994092345237732, + "num_tokens": 641220646.0, + "step": 16517 + }, + { + "epoch": 2.1012593817580463, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.45172119140625, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.9018467664718628, + "num_tokens": 641266049.0, + "step": 16518 + }, + { + "epoch": 2.1013865920366364, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.541994094848633, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8951917886734009, + "num_tokens": 641306875.0, + "step": 16519 + }, + { + "epoch": 2.101513802315227, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.593400955200195, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9036056995391846, + "num_tokens": 641343040.0, + "step": 16520 + }, + { + "epoch": 2.1016410125938174, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.55821418762207, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.90036940574646, + "num_tokens": 641381365.0, + "step": 16521 + }, + { + "epoch": 2.101768222872408, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.516803741455078, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8962002992630005, + "num_tokens": 641423921.0, + "step": 16522 + }, + { + "epoch": 2.1018954331509985, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.535014152526855, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.9048407077789307, + "num_tokens": 641461086.0, + "step": 16523 + }, + { + "epoch": 2.102022643429589, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.5048189163208, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.9018703699111938, + "num_tokens": 641499846.0, + "step": 16524 + }, + { + "epoch": 2.1021498537081795, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.720770835876465, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8907742500305176, + "num_tokens": 641535790.0, + "step": 16525 + }, + { + "epoch": 2.10227706398677, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.580304145812988, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8920247554779053, + "num_tokens": 641576954.0, + "step": 16526 + }, + { + "epoch": 2.1024042742653606, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.441572189331055, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9051666855812073, + "num_tokens": 641613192.0, + "step": 16527 + }, + { + "epoch": 2.102531484543951, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.716586112976074, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8911179304122925, + "num_tokens": 641646024.0, + "step": 16528 + }, + { + "epoch": 2.1026586948225416, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.504204750061035, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9092276096343994, + "num_tokens": 641680409.0, + "step": 16529 + }, + { + "epoch": 2.102785905101132, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.421064376831055, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9044214487075806, + "num_tokens": 641725589.0, + "step": 16530 + }, + { + "epoch": 2.1029131153797227, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.53546142578125, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9150151014328003, + "num_tokens": 641761769.0, + "step": 16531 + }, + { + "epoch": 2.1030403256583132, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.598485946655273, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8949671983718872, + "num_tokens": 641800899.0, + "step": 16532 + }, + { + "epoch": 2.1031675359369038, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.564107894897461, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8956016898155212, + "num_tokens": 641837569.0, + "step": 16533 + }, + { + "epoch": 2.1032947462154943, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.432638168334961, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.906183123588562, + "num_tokens": 641879362.0, + "step": 16534 + }, + { + "epoch": 2.103421956494085, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.609603881835938, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8983094692230225, + "num_tokens": 641917018.0, + "step": 16535 + }, + { + "epoch": 2.1035491667726753, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.539395332336426, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9048427939414978, + "num_tokens": 641957690.0, + "step": 16536 + }, + { + "epoch": 2.103676377051266, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.60470962524414, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8975082039833069, + "num_tokens": 641988532.0, + "step": 16537 + }, + { + "epoch": 2.1038035873298564, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.543201446533203, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8981291055679321, + "num_tokens": 642026022.0, + "step": 16538 + }, + { + "epoch": 2.103930797608447, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.483333587646484, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8990776538848877, + "num_tokens": 642070627.0, + "step": 16539 + }, + { + "epoch": 2.1040580078870375, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.553374290466309, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8994821310043335, + "num_tokens": 642109306.0, + "step": 16540 + }, + { + "epoch": 2.104185218165628, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.555733680725098, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.9021183252334595, + "num_tokens": 642143920.0, + "step": 16541 + }, + { + "epoch": 2.1043124284442185, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.584901809692383, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8927216529846191, + "num_tokens": 642181904.0, + "step": 16542 + }, + { + "epoch": 2.1044396387228086, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.56293773651123, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.9052842855453491, + "num_tokens": 642217973.0, + "step": 16543 + }, + { + "epoch": 2.104566849001399, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.556371688842773, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8924990892410278, + "num_tokens": 642260461.0, + "step": 16544 + }, + { + "epoch": 2.1046940592799896, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.551698684692383, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8948276042938232, + "num_tokens": 642298187.0, + "step": 16545 + }, + { + "epoch": 2.10482126955858, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.533629417419434, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8964579105377197, + "num_tokens": 642337994.0, + "step": 16546 + }, + { + "epoch": 2.1049484798371707, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.639074325561523, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.9002760648727417, + "num_tokens": 642373236.0, + "step": 16547 + }, + { + "epoch": 2.1050756901157612, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.492169380187988, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9061193466186523, + "num_tokens": 642416151.0, + "step": 16548 + }, + { + "epoch": 2.1052029003943518, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.591663360595703, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8936155438423157, + "num_tokens": 642453523.0, + "step": 16549 + }, + { + "epoch": 2.1053301106729423, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.566256523132324, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8983606100082397, + "num_tokens": 642489510.0, + "step": 16550 + }, + { + "epoch": 2.105457320951533, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.62529182434082, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8897159099578857, + "num_tokens": 642524899.0, + "step": 16551 + }, + { + "epoch": 2.1055845312301233, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.524258613586426, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.900961697101593, + "num_tokens": 642562735.0, + "step": 16552 + }, + { + "epoch": 2.105711741508714, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.69428825378418, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.9043965339660645, + "num_tokens": 642594467.0, + "step": 16553 + }, + { + "epoch": 2.1058389517873044, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.458778381347656, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8973697423934937, + "num_tokens": 642638983.0, + "step": 16554 + }, + { + "epoch": 2.105966162065895, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.543218612670898, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8986747860908508, + "num_tokens": 642677187.0, + "step": 16555 + }, + { + "epoch": 2.1060933723444855, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.50065803527832, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9015356302261353, + "num_tokens": 642720574.0, + "step": 16556 + }, + { + "epoch": 2.106220582623076, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.585956573486328, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.9047614932060242, + "num_tokens": 642756118.0, + "step": 16557 + }, + { + "epoch": 2.1063477929016665, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.522112846374512, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.9034483432769775, + "num_tokens": 642793341.0, + "step": 16558 + }, + { + "epoch": 2.106475003180257, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.502327919006348, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8973633050918579, + "num_tokens": 642836326.0, + "step": 16559 + }, + { + "epoch": 2.1066022134588476, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.613248825073242, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9141392111778259, + "num_tokens": 642871568.0, + "step": 16560 + }, + { + "epoch": 2.106729423737438, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.598649978637695, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8883646726608276, + "num_tokens": 642912050.0, + "step": 16561 + }, + { + "epoch": 2.1068566340160286, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.56855297088623, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8904749155044556, + "num_tokens": 642952600.0, + "step": 16562 + }, + { + "epoch": 2.106983844294619, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.446450233459473, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9065735340118408, + "num_tokens": 642993582.0, + "step": 16563 + }, + { + "epoch": 2.1071110545732097, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.543830871582031, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9084017276763916, + "num_tokens": 643034554.0, + "step": 16564 + }, + { + "epoch": 2.1072382648518, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.539917945861816, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.9046194553375244, + "num_tokens": 643069668.0, + "step": 16565 + }, + { + "epoch": 2.1073654751303907, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.59470272064209, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.9033422470092773, + "num_tokens": 643106632.0, + "step": 16566 + }, + { + "epoch": 2.107492685408981, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.495336532592773, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9062720537185669, + "num_tokens": 643144126.0, + "step": 16567 + }, + { + "epoch": 2.1076198956875714, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.502424240112305, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.9025146961212158, + "num_tokens": 643185101.0, + "step": 16568 + }, + { + "epoch": 2.107747105966162, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.746362686157227, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8926074504852295, + "num_tokens": 643216565.0, + "step": 16569 + }, + { + "epoch": 2.1078743162447524, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.600366592407227, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.908584475517273, + "num_tokens": 643253108.0, + "step": 16570 + }, + { + "epoch": 2.108001526523343, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.642637252807617, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9033921957015991, + "num_tokens": 643286515.0, + "step": 16571 + }, + { + "epoch": 2.1081287368019335, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.477520942687988, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9144076108932495, + "num_tokens": 643322553.0, + "step": 16572 + }, + { + "epoch": 2.108255947080524, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.564969062805176, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8978676795959473, + "num_tokens": 643364096.0, + "step": 16573 + }, + { + "epoch": 2.1083831573591145, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.49593448638916, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9081348180770874, + "num_tokens": 643401852.0, + "step": 16574 + }, + { + "epoch": 2.108510367637705, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.559242248535156, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.900757908821106, + "num_tokens": 643439363.0, + "step": 16575 + }, + { + "epoch": 2.1086375779162956, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.558045387268066, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.901178240776062, + "num_tokens": 643475839.0, + "step": 16576 + }, + { + "epoch": 2.108764788194886, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.53758430480957, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.9018807411193848, + "num_tokens": 643521905.0, + "step": 16577 + }, + { + "epoch": 2.1088919984734766, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.54169750213623, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9040076732635498, + "num_tokens": 643557276.0, + "step": 16578 + }, + { + "epoch": 2.109019208752067, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.664454460144043, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.9009112119674683, + "num_tokens": 643594399.0, + "step": 16579 + }, + { + "epoch": 2.1091464190306577, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.575371742248535, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8892908692359924, + "num_tokens": 643632902.0, + "step": 16580 + }, + { + "epoch": 2.109273629309248, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.60993766784668, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9035544991493225, + "num_tokens": 643667942.0, + "step": 16581 + }, + { + "epoch": 2.1094008395878387, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.542275428771973, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.906201183795929, + "num_tokens": 643706514.0, + "step": 16582 + }, + { + "epoch": 2.1095280498664293, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.477104187011719, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.8999779224395752, + "num_tokens": 643745903.0, + "step": 16583 + }, + { + "epoch": 2.10965526014502, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.48820972442627, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9063993096351624, + "num_tokens": 643786285.0, + "step": 16584 + }, + { + "epoch": 2.1097824704236103, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.515800476074219, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9081106185913086, + "num_tokens": 643830810.0, + "step": 16585 + }, + { + "epoch": 2.109909680702201, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.596941947937012, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9099457263946533, + "num_tokens": 643862756.0, + "step": 16586 + }, + { + "epoch": 2.1100368909807914, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.481608390808105, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8927140235900879, + "num_tokens": 643906693.0, + "step": 16587 + }, + { + "epoch": 2.110164101259382, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.582897186279297, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8891834616661072, + "num_tokens": 643947791.0, + "step": 16588 + }, + { + "epoch": 2.1102913115379724, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.58549690246582, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8913553357124329, + "num_tokens": 643983257.0, + "step": 16589 + }, + { + "epoch": 2.110418521816563, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.56412410736084, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8948191404342651, + "num_tokens": 644020046.0, + "step": 16590 + }, + { + "epoch": 2.1105457320951535, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.578620910644531, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8849107623100281, + "num_tokens": 644057099.0, + "step": 16591 + }, + { + "epoch": 2.1106729423737436, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.526044845581055, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8939664959907532, + "num_tokens": 644099007.0, + "step": 16592 + }, + { + "epoch": 2.110800152652334, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.699923515319824, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8865347504615784, + "num_tokens": 644132785.0, + "step": 16593 + }, + { + "epoch": 2.1109273629309246, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.529842376708984, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8981890082359314, + "num_tokens": 644170917.0, + "step": 16594 + }, + { + "epoch": 2.111054573209515, + "ewc_loss": 0.018798828125, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.532971382141113, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9011264443397522, + "num_tokens": 644213539.0, + "step": 16595 + }, + { + "epoch": 2.1111817834881057, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.563602447509766, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9050008058547974, + "num_tokens": 644249645.0, + "step": 16596 + }, + { + "epoch": 2.1113089937666962, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.478718757629395, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8967723846435547, + "num_tokens": 644296861.0, + "step": 16597 + }, + { + "epoch": 2.1114362040452868, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.553479194641113, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8990507125854492, + "num_tokens": 644334722.0, + "step": 16598 + }, + { + "epoch": 2.1115634143238773, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.560543060302734, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.9063670635223389, + "num_tokens": 644375860.0, + "step": 16599 + }, + { + "epoch": 2.111690624602468, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.674430847167969, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9115489721298218, + "num_tokens": 644404573.0, + "step": 16600 + }, + { + "epoch": 2.1118178348810583, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.577705383300781, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8985428214073181, + "num_tokens": 644446651.0, + "step": 16601 + }, + { + "epoch": 2.111945045159649, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.524458885192871, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.9114691615104675, + "num_tokens": 644480363.0, + "step": 16602 + }, + { + "epoch": 2.1120722554382394, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.531454086303711, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.9065150022506714, + "num_tokens": 644521625.0, + "step": 16603 + }, + { + "epoch": 2.11219946571683, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.761112213134766, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8805938959121704, + "num_tokens": 644555474.0, + "step": 16604 + }, + { + "epoch": 2.1123266759954205, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.343292236328125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.551348686218262, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8958807587623596, + "num_tokens": 644590443.0, + "step": 16605 + }, + { + "epoch": 2.112453886274011, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.57676887512207, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8994988799095154, + "num_tokens": 644633817.0, + "step": 16606 + }, + { + "epoch": 2.1125810965526015, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.583974838256836, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.893004834651947, + "num_tokens": 644671309.0, + "step": 16607 + }, + { + "epoch": 2.112708306831192, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.580886840820312, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9056969881057739, + "num_tokens": 644702866.0, + "step": 16608 + }, + { + "epoch": 2.1128355171097826, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.576523780822754, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.9017302989959717, + "num_tokens": 644739944.0, + "step": 16609 + }, + { + "epoch": 2.112962727388373, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.50596809387207, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8914512395858765, + "num_tokens": 644783191.0, + "step": 16610 + }, + { + "epoch": 2.1130899376669636, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.606612205505371, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9024819731712341, + "num_tokens": 644820028.0, + "step": 16611 + }, + { + "epoch": 2.113217147945554, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.631057739257812, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8986599445343018, + "num_tokens": 644859575.0, + "step": 16612 + }, + { + "epoch": 2.1133443582241447, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.534096717834473, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.9020895957946777, + "num_tokens": 644900591.0, + "step": 16613 + }, + { + "epoch": 2.113471568502735, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.559968948364258, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.9019111394882202, + "num_tokens": 644940245.0, + "step": 16614 + }, + { + "epoch": 2.1135987787813257, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.636449813842773, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.9055721163749695, + "num_tokens": 644974576.0, + "step": 16615 + }, + { + "epoch": 2.1137259890599163, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.481398582458496, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8944171667098999, + "num_tokens": 645017925.0, + "step": 16616 + }, + { + "epoch": 2.1138531993385064, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.632181167602539, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.891579806804657, + "num_tokens": 645057346.0, + "step": 16617 + }, + { + "epoch": 2.113980409617097, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.515220642089844, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.9008116722106934, + "num_tokens": 645102445.0, + "step": 16618 + }, + { + "epoch": 2.1141076198956874, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.519963264465332, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9087731242179871, + "num_tokens": 645137673.0, + "step": 16619 + }, + { + "epoch": 2.114234830174278, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.626827239990234, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.90460604429245, + "num_tokens": 645171499.0, + "step": 16620 + }, + { + "epoch": 2.1143620404528685, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.507259368896484, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9055348634719849, + "num_tokens": 645211964.0, + "step": 16621 + }, + { + "epoch": 2.114489250731459, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.478838920593262, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8926382660865784, + "num_tokens": 645252792.0, + "step": 16622 + }, + { + "epoch": 2.1146164610100495, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.650147438049316, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8957700729370117, + "num_tokens": 645292763.0, + "step": 16623 + }, + { + "epoch": 2.11474367128864, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.633389472961426, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8868529796600342, + "num_tokens": 645332276.0, + "step": 16624 + }, + { + "epoch": 2.1148708815672306, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.526690483093262, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8974562883377075, + "num_tokens": 645369057.0, + "step": 16625 + }, + { + "epoch": 2.114998091845821, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.61699390411377, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8956743478775024, + "num_tokens": 645405033.0, + "step": 16626 + }, + { + "epoch": 2.1151253021244116, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.626262664794922, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9021140336990356, + "num_tokens": 645436503.0, + "step": 16627 + }, + { + "epoch": 2.115252512403002, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.608514785766602, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8889186382293701, + "num_tokens": 645473354.0, + "step": 16628 + }, + { + "epoch": 2.1153797226815927, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.445055961608887, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8995771408081055, + "num_tokens": 645516614.0, + "step": 16629 + }, + { + "epoch": 2.115506932960183, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.503852844238281, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.90077143907547, + "num_tokens": 645556936.0, + "step": 16630 + }, + { + "epoch": 2.1156341432387737, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.583924293518066, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8933603763580322, + "num_tokens": 645594549.0, + "step": 16631 + }, + { + "epoch": 2.1157613535173643, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.576107025146484, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8932867050170898, + "num_tokens": 645630475.0, + "step": 16632 + }, + { + "epoch": 2.115888563795955, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.572280883789062, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8792625665664673, + "num_tokens": 645671957.0, + "step": 16633 + }, + { + "epoch": 2.1160157740745453, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.424775123596191, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.9066318273544312, + "num_tokens": 645715785.0, + "step": 16634 + }, + { + "epoch": 2.116142984353136, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.635065078735352, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.9044325351715088, + "num_tokens": 645749107.0, + "step": 16635 + }, + { + "epoch": 2.1162701946317264, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.519645690917969, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8950350880622864, + "num_tokens": 645785914.0, + "step": 16636 + }, + { + "epoch": 2.116397404910317, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.47899055480957, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.907167375087738, + "num_tokens": 645826283.0, + "step": 16637 + }, + { + "epoch": 2.1165246151889074, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.545462608337402, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8996649384498596, + "num_tokens": 645864317.0, + "step": 16638 + }, + { + "epoch": 2.116651825467498, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.661178588867188, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.9028236865997314, + "num_tokens": 645896448.0, + "step": 16639 + }, + { + "epoch": 2.116779035746088, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.593432426452637, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.899114727973938, + "num_tokens": 645929576.0, + "step": 16640 + }, + { + "epoch": 2.1169062460246786, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.491623878479004, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8917735815048218, + "num_tokens": 645973297.0, + "step": 16641 + }, + { + "epoch": 2.117033456303269, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.567797660827637, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8866958618164062, + "num_tokens": 646009942.0, + "step": 16642 + }, + { + "epoch": 2.1171606665818596, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.558611869812012, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8977351188659668, + "num_tokens": 646049397.0, + "step": 16643 + }, + { + "epoch": 2.11728787686045, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.501792907714844, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9069433212280273, + "num_tokens": 646087156.0, + "step": 16644 + }, + { + "epoch": 2.1174150871390407, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.517001152038574, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8992024660110474, + "num_tokens": 646127586.0, + "step": 16645 + }, + { + "epoch": 2.1175422974176312, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.58342456817627, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9087945222854614, + "num_tokens": 646162730.0, + "step": 16646 + }, + { + "epoch": 2.1176695076962218, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.498995780944824, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8978986740112305, + "num_tokens": 646206878.0, + "step": 16647 + }, + { + "epoch": 2.1177967179748123, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.600678443908691, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9037186503410339, + "num_tokens": 646246002.0, + "step": 16648 + }, + { + "epoch": 2.117923928253403, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.498689651489258, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9113848209381104, + "num_tokens": 646286726.0, + "step": 16649 + }, + { + "epoch": 2.1180511385319933, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.60780143737793, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9050452709197998, + "num_tokens": 646324299.0, + "step": 16650 + }, + { + "epoch": 2.118178348810584, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.518054008483887, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.9043048024177551, + "num_tokens": 646366198.0, + "step": 16651 + }, + { + "epoch": 2.1183055590891744, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.68871021270752, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8936110734939575, + "num_tokens": 646402226.0, + "step": 16652 + }, + { + "epoch": 2.118432769367765, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.550941467285156, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.9041587710380554, + "num_tokens": 646437284.0, + "step": 16653 + }, + { + "epoch": 2.1185599796463555, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.53447437286377, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.9025171995162964, + "num_tokens": 646474096.0, + "step": 16654 + }, + { + "epoch": 2.118687189924946, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.547833442687988, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8962695598602295, + "num_tokens": 646515533.0, + "step": 16655 + }, + { + "epoch": 2.1188144002035365, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.56651782989502, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9071658849716187, + "num_tokens": 646549907.0, + "step": 16656 + }, + { + "epoch": 2.118941610482127, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.623964309692383, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8957719206809998, + "num_tokens": 646583723.0, + "step": 16657 + }, + { + "epoch": 2.1190688207607176, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.662261962890625, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8865537047386169, + "num_tokens": 646619065.0, + "step": 16658 + }, + { + "epoch": 2.119196031039308, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.577624320983887, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.9005258083343506, + "num_tokens": 646659966.0, + "step": 16659 + }, + { + "epoch": 2.1193232413178986, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.547213554382324, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9077047109603882, + "num_tokens": 646700386.0, + "step": 16660 + }, + { + "epoch": 2.119450451596489, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.677289962768555, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8886414766311646, + "num_tokens": 646735335.0, + "step": 16661 + }, + { + "epoch": 2.1195776618750797, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.607318878173828, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8933556079864502, + "num_tokens": 646771291.0, + "step": 16662 + }, + { + "epoch": 2.11970487215367, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.559590339660645, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8951776623725891, + "num_tokens": 646809922.0, + "step": 16663 + }, + { + "epoch": 2.1198320824322607, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.673398971557617, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8902649879455566, + "num_tokens": 646848114.0, + "step": 16664 + }, + { + "epoch": 2.119959292710851, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.50384521484375, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8957629203796387, + "num_tokens": 646887226.0, + "step": 16665 + }, + { + "epoch": 2.1200865029894413, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.551346778869629, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8838791847229004, + "num_tokens": 646933503.0, + "step": 16666 + }, + { + "epoch": 2.120213713268032, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.580400466918945, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.9023618698120117, + "num_tokens": 646968823.0, + "step": 16667 + }, + { + "epoch": 2.1203409235466224, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.538129806518555, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.9020535945892334, + "num_tokens": 647005874.0, + "step": 16668 + }, + { + "epoch": 2.120468133825213, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.577592849731445, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.888619065284729, + "num_tokens": 647050790.0, + "step": 16669 + }, + { + "epoch": 2.1205953441038035, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.539263725280762, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8974539041519165, + "num_tokens": 647088221.0, + "step": 16670 + }, + { + "epoch": 2.120722554382394, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.510594367980957, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.9065072536468506, + "num_tokens": 647127185.0, + "step": 16671 + }, + { + "epoch": 2.1208497646609845, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.518037796020508, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8966034650802612, + "num_tokens": 647166950.0, + "step": 16672 + }, + { + "epoch": 2.120976974939575, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.523038864135742, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9055327773094177, + "num_tokens": 647209596.0, + "step": 16673 + }, + { + "epoch": 2.1211041852181656, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.651583671569824, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8968915939331055, + "num_tokens": 647242564.0, + "step": 16674 + }, + { + "epoch": 2.121231395496756, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.532044410705566, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.896124005317688, + "num_tokens": 647286801.0, + "step": 16675 + }, + { + "epoch": 2.1213586057753466, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.390975952148438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.518152236938477, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9098280668258667, + "num_tokens": 647330013.0, + "step": 16676 + }, + { + "epoch": 2.121485816053937, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.511958122253418, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.902454137802124, + "num_tokens": 647365300.0, + "step": 16677 + }, + { + "epoch": 2.1216130263325277, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.654187202453613, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8893502950668335, + "num_tokens": 647403978.0, + "step": 16678 + }, + { + "epoch": 2.121740236611118, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.494811058044434, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9089237451553345, + "num_tokens": 647440341.0, + "step": 16679 + }, + { + "epoch": 2.1218674468897087, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.50986099243164, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.9035201668739319, + "num_tokens": 647481577.0, + "step": 16680 + }, + { + "epoch": 2.1219946571682993, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.54399585723877, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8969129323959351, + "num_tokens": 647521163.0, + "step": 16681 + }, + { + "epoch": 2.12212186744689, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.517066955566406, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.9014025926589966, + "num_tokens": 647561098.0, + "step": 16682 + }, + { + "epoch": 2.1222490777254803, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.582294464111328, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8873025178909302, + "num_tokens": 647603154.0, + "step": 16683 + }, + { + "epoch": 2.122376288004071, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.475407600402832, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.9034167528152466, + "num_tokens": 647647241.0, + "step": 16684 + }, + { + "epoch": 2.1225034982826614, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.528250694274902, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8956301212310791, + "num_tokens": 647691130.0, + "step": 16685 + }, + { + "epoch": 2.122630708561252, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.48421573638916, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.9059645533561707, + "num_tokens": 647732694.0, + "step": 16686 + }, + { + "epoch": 2.1227579188398424, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.620741844177246, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8997582197189331, + "num_tokens": 647767279.0, + "step": 16687 + }, + { + "epoch": 2.122885129118433, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.508329391479492, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9153337478637695, + "num_tokens": 647806090.0, + "step": 16688 + }, + { + "epoch": 2.1230123393970235, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.526199340820312, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.9037041664123535, + "num_tokens": 647844980.0, + "step": 16689 + }, + { + "epoch": 2.1231395496756136, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.719606399536133, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8916728496551514, + "num_tokens": 647876710.0, + "step": 16690 + }, + { + "epoch": 2.123266759954204, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.512293815612793, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9015089273452759, + "num_tokens": 647915128.0, + "step": 16691 + }, + { + "epoch": 2.1233939702327946, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.491988182067871, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9052137136459351, + "num_tokens": 647959752.0, + "step": 16692 + }, + { + "epoch": 2.123521180511385, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.575525283813477, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.9049079418182373, + "num_tokens": 647997519.0, + "step": 16693 + }, + { + "epoch": 2.1236483907899757, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.570188522338867, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9090787172317505, + "num_tokens": 648028956.0, + "step": 16694 + }, + { + "epoch": 2.123775601068566, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.606422424316406, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9031904935836792, + "num_tokens": 648061793.0, + "step": 16695 + }, + { + "epoch": 2.1239028113471567, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.584874153137207, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9100023508071899, + "num_tokens": 648093739.0, + "step": 16696 + }, + { + "epoch": 2.1240300216257473, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.469910621643066, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8954867720603943, + "num_tokens": 648138625.0, + "step": 16697 + }, + { + "epoch": 2.124157231904338, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.513484954833984, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.892900824546814, + "num_tokens": 648177736.0, + "step": 16698 + }, + { + "epoch": 2.1242844421829283, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.579180717468262, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.904512882232666, + "num_tokens": 648213902.0, + "step": 16699 + }, + { + "epoch": 2.124411652461519, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.528834342956543, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8900752663612366, + "num_tokens": 648254392.0, + "step": 16700 + }, + { + "epoch": 2.1245388627401094, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.742480278015137, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.883683443069458, + "num_tokens": 648289332.0, + "step": 16701 + }, + { + "epoch": 2.1246660730187, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.518428802490234, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9144314527511597, + "num_tokens": 648326720.0, + "step": 16702 + }, + { + "epoch": 2.1247932832972904, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.582235336303711, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8978792428970337, + "num_tokens": 648366039.0, + "step": 16703 + }, + { + "epoch": 2.124920493575881, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.56551742553711, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9083912372589111, + "num_tokens": 648401194.0, + "step": 16704 + }, + { + "epoch": 2.1250477038544715, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.607094764709473, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.898052453994751, + "num_tokens": 648437825.0, + "step": 16705 + }, + { + "epoch": 2.125174914133062, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.572760581970215, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9012547731399536, + "num_tokens": 648474124.0, + "step": 16706 + }, + { + "epoch": 2.1253021244116526, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.55846118927002, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8899070620536804, + "num_tokens": 648522145.0, + "step": 16707 + }, + { + "epoch": 2.125429334690243, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.516847610473633, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.9045950770378113, + "num_tokens": 648562680.0, + "step": 16708 + }, + { + "epoch": 2.1255565449688336, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.563150405883789, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.9025124311447144, + "num_tokens": 648599716.0, + "step": 16709 + }, + { + "epoch": 2.125683755247424, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.718730926513672, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8863265514373779, + "num_tokens": 648634453.0, + "step": 16710 + }, + { + "epoch": 2.1258109655260147, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.529388427734375, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.903725802898407, + "num_tokens": 648671167.0, + "step": 16711 + }, + { + "epoch": 2.125938175804605, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.54286003112793, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.9006850719451904, + "num_tokens": 648712661.0, + "step": 16712 + }, + { + "epoch": 2.1260653860831957, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.524526596069336, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.9059779644012451, + "num_tokens": 648754151.0, + "step": 16713 + }, + { + "epoch": 2.1261925963617863, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.640576362609863, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8889873027801514, + "num_tokens": 648792216.0, + "step": 16714 + }, + { + "epoch": 2.1263198066403763, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.43865966796875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.558661460876465, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.896714985370636, + "num_tokens": 648828355.0, + "step": 16715 + }, + { + "epoch": 2.126447016918967, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.659252166748047, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.894882082939148, + "num_tokens": 648864081.0, + "step": 16716 + }, + { + "epoch": 2.1265742271975574, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.527664184570312, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.899986982345581, + "num_tokens": 648907191.0, + "step": 16717 + }, + { + "epoch": 2.126701437476148, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.563148498535156, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8973696231842041, + "num_tokens": 648945641.0, + "step": 16718 + }, + { + "epoch": 2.1268286477547385, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.748628616333008, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9024633765220642, + "num_tokens": 648973744.0, + "step": 16719 + }, + { + "epoch": 2.126955858033329, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.589677810668945, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.9006455540657043, + "num_tokens": 649008043.0, + "step": 16720 + }, + { + "epoch": 2.1270830683119195, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.570720672607422, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8967041969299316, + "num_tokens": 649043109.0, + "step": 16721 + }, + { + "epoch": 2.12721027859051, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.628689765930176, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8999616503715515, + "num_tokens": 649081308.0, + "step": 16722 + }, + { + "epoch": 2.1273374888691006, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.564043998718262, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.9044145345687866, + "num_tokens": 649121516.0, + "step": 16723 + }, + { + "epoch": 2.127464699147691, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.636975288391113, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8887202143669128, + "num_tokens": 649160490.0, + "step": 16724 + }, + { + "epoch": 2.1275919094262816, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.605361938476562, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8927513360977173, + "num_tokens": 649203534.0, + "step": 16725 + }, + { + "epoch": 2.127719119704872, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.476181030273438, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9112865328788757, + "num_tokens": 649248439.0, + "step": 16726 + }, + { + "epoch": 2.1278463299834627, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.642668724060059, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8889374136924744, + "num_tokens": 649286629.0, + "step": 16727 + }, + { + "epoch": 2.127973540262053, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.602256774902344, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8963924646377563, + "num_tokens": 649327279.0, + "step": 16728 + }, + { + "epoch": 2.1281007505406437, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.433966636657715, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.9073753356933594, + "num_tokens": 649370979.0, + "step": 16729 + }, + { + "epoch": 2.1282279608192343, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.61184024810791, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8927876949310303, + "num_tokens": 649416544.0, + "step": 16730 + }, + { + "epoch": 2.128355171097825, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.518634796142578, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9067825675010681, + "num_tokens": 649450936.0, + "step": 16731 + }, + { + "epoch": 2.1284823813764153, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.608816146850586, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8918020129203796, + "num_tokens": 649491332.0, + "step": 16732 + }, + { + "epoch": 2.128609591655006, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.514327049255371, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8978049755096436, + "num_tokens": 649533864.0, + "step": 16733 + }, + { + "epoch": 2.1287368019335964, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.583279609680176, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9024369716644287, + "num_tokens": 649570296.0, + "step": 16734 + }, + { + "epoch": 2.128864012212187, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.683045387268066, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8831481337547302, + "num_tokens": 649613469.0, + "step": 16735 + }, + { + "epoch": 2.1289912224907774, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.573518753051758, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.9007371664047241, + "num_tokens": 649649673.0, + "step": 16736 + }, + { + "epoch": 2.129118432769368, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.605737686157227, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9046667814254761, + "num_tokens": 649685847.0, + "step": 16737 + }, + { + "epoch": 2.129245643047958, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.566420555114746, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8983506560325623, + "num_tokens": 649725806.0, + "step": 16738 + }, + { + "epoch": 2.129372853326549, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.605201721191406, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.9049252271652222, + "num_tokens": 649763187.0, + "step": 16739 + }, + { + "epoch": 2.129500063605139, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.571935653686523, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9032861590385437, + "num_tokens": 649801805.0, + "step": 16740 + }, + { + "epoch": 2.1296272738837296, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.652444839477539, + "learning_rate": 1e-06, + "loss": 0.25, + "mean_token_accuracy": 0.9131191968917847, + "num_tokens": 649837573.0, + "step": 16741 + }, + { + "epoch": 2.12975448416232, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.659202575683594, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.9049944877624512, + "num_tokens": 649869782.0, + "step": 16742 + }, + { + "epoch": 2.1298816944409107, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.637718200683594, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8958470225334167, + "num_tokens": 649905969.0, + "step": 16743 + }, + { + "epoch": 2.130008904719501, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.530767440795898, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8898391127586365, + "num_tokens": 649951269.0, + "step": 16744 + }, + { + "epoch": 2.1301361149980917, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.545666694641113, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9003133773803711, + "num_tokens": 649990813.0, + "step": 16745 + }, + { + "epoch": 2.1302633252766823, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 10.127959251403809, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8977968692779541, + "num_tokens": 650031762.0, + "step": 16746 + }, + { + "epoch": 2.130390535555273, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.66981315612793, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8840889930725098, + "num_tokens": 650068705.0, + "step": 16747 + }, + { + "epoch": 2.1305177458338633, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.636981010437012, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8896320462226868, + "num_tokens": 650103239.0, + "step": 16748 + }, + { + "epoch": 2.130644956112454, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.617753982543945, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8986401557922363, + "num_tokens": 650143838.0, + "step": 16749 + }, + { + "epoch": 2.1307721663910444, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.699152946472168, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8904458284378052, + "num_tokens": 650180746.0, + "step": 16750 + }, + { + "epoch": 2.130899376669635, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.630410194396973, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8932777643203735, + "num_tokens": 650220469.0, + "step": 16751 + }, + { + "epoch": 2.1310265869482254, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.514522552490234, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9042639136314392, + "num_tokens": 650259935.0, + "step": 16752 + }, + { + "epoch": 2.131153797226816, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.605966567993164, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8981918096542358, + "num_tokens": 650298385.0, + "step": 16753 + }, + { + "epoch": 2.1312810075054065, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.5467529296875, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9088397026062012, + "num_tokens": 650330930.0, + "step": 16754 + }, + { + "epoch": 2.131408217783997, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.621389389038086, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8925116062164307, + "num_tokens": 650370266.0, + "step": 16755 + }, + { + "epoch": 2.1315354280625876, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.575017929077148, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.9013223648071289, + "num_tokens": 650411684.0, + "step": 16756 + }, + { + "epoch": 2.131662638341178, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.569401741027832, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.9015164375305176, + "num_tokens": 650453065.0, + "step": 16757 + }, + { + "epoch": 2.1317898486197686, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.651243209838867, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.891305685043335, + "num_tokens": 650494059.0, + "step": 16758 + }, + { + "epoch": 2.131917058898359, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.600288391113281, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8957210779190063, + "num_tokens": 650531871.0, + "step": 16759 + }, + { + "epoch": 2.1320442691769497, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.529643058776855, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8979673385620117, + "num_tokens": 650578962.0, + "step": 16760 + }, + { + "epoch": 2.13217147945554, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.507259368896484, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.9036300182342529, + "num_tokens": 650621799.0, + "step": 16761 + }, + { + "epoch": 2.1322986897341307, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.630148887634277, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8874648809432983, + "num_tokens": 650661451.0, + "step": 16762 + }, + { + "epoch": 2.132425900012721, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.605857849121094, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8987293243408203, + "num_tokens": 650696539.0, + "step": 16763 + }, + { + "epoch": 2.1325531102913113, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.542259216308594, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.9071170091629028, + "num_tokens": 650737721.0, + "step": 16764 + }, + { + "epoch": 2.132680320569902, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.56701946258545, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8934680223464966, + "num_tokens": 650778421.0, + "step": 16765 + }, + { + "epoch": 2.1328075308484924, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.612679481506348, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8975276350975037, + "num_tokens": 650817245.0, + "step": 16766 + }, + { + "epoch": 2.132934741127083, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.617982864379883, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.898647665977478, + "num_tokens": 650853937.0, + "step": 16767 + }, + { + "epoch": 2.1330619514056735, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.554356575012207, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9114696979522705, + "num_tokens": 650889963.0, + "step": 16768 + }, + { + "epoch": 2.133189161684264, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.548445701599121, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8967462778091431, + "num_tokens": 650928850.0, + "step": 16769 + }, + { + "epoch": 2.1333163719628545, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.603153228759766, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8965587615966797, + "num_tokens": 650962297.0, + "step": 16770 + }, + { + "epoch": 2.133443582241445, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.48518180847168, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8995004892349243, + "num_tokens": 651006262.0, + "step": 16771 + }, + { + "epoch": 2.1335707925200356, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.544393539428711, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.9052536487579346, + "num_tokens": 651048112.0, + "step": 16772 + }, + { + "epoch": 2.133698002798626, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.507834434509277, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.9109851121902466, + "num_tokens": 651087800.0, + "step": 16773 + }, + { + "epoch": 2.1338252130772166, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.525986671447754, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9100238680839539, + "num_tokens": 651128242.0, + "step": 16774 + }, + { + "epoch": 2.133952423355807, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.613490104675293, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8895232081413269, + "num_tokens": 651167271.0, + "step": 16775 + }, + { + "epoch": 2.1340796336343977, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.828997611999512, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8876364231109619, + "num_tokens": 651201970.0, + "step": 16776 + }, + { + "epoch": 2.134206843912988, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.5452880859375, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8912830948829651, + "num_tokens": 651242331.0, + "step": 16777 + }, + { + "epoch": 2.1343340541915787, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.492281913757324, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8864833116531372, + "num_tokens": 651288837.0, + "step": 16778 + }, + { + "epoch": 2.1344612644701693, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.694632530212402, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.891133189201355, + "num_tokens": 651325367.0, + "step": 16779 + }, + { + "epoch": 2.13458847474876, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.47038745880127, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8982847929000854, + "num_tokens": 651367639.0, + "step": 16780 + }, + { + "epoch": 2.1347156850273503, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.614377975463867, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9005682468414307, + "num_tokens": 651406178.0, + "step": 16781 + }, + { + "epoch": 2.134842895305941, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.656432151794434, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8822068572044373, + "num_tokens": 651444799.0, + "step": 16782 + }, + { + "epoch": 2.1349701055845314, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.50497055053711, + "learning_rate": 1e-06, + "loss": 0.2454, + "mean_token_accuracy": 0.916983962059021, + "num_tokens": 651487171.0, + "step": 16783 + }, + { + "epoch": 2.135097315863122, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.625099182128906, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8959975838661194, + "num_tokens": 651525780.0, + "step": 16784 + }, + { + "epoch": 2.1352245261417124, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.580628395080566, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8982983827590942, + "num_tokens": 651560794.0, + "step": 16785 + }, + { + "epoch": 2.135351736420303, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.520442008972168, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.910975456237793, + "num_tokens": 651596171.0, + "step": 16786 + }, + { + "epoch": 2.1354789466988935, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.453089714050293, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9029868245124817, + "num_tokens": 651639030.0, + "step": 16787 + }, + { + "epoch": 2.1356061569774836, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.58941650390625, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.9038423299789429, + "num_tokens": 651680945.0, + "step": 16788 + }, + { + "epoch": 2.135733367256074, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.599760055541992, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9083091020584106, + "num_tokens": 651716549.0, + "step": 16789 + }, + { + "epoch": 2.1358605775346646, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.599947929382324, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8933764100074768, + "num_tokens": 651753812.0, + "step": 16790 + }, + { + "epoch": 2.135987787813255, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.458318710327148, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.9026412963867188, + "num_tokens": 651800489.0, + "step": 16791 + }, + { + "epoch": 2.1361149980918457, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.580065727233887, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9097806215286255, + "num_tokens": 651838924.0, + "step": 16792 + }, + { + "epoch": 2.136242208370436, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.512920379638672, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.897397518157959, + "num_tokens": 651886165.0, + "step": 16793 + }, + { + "epoch": 2.1363694186490267, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.606109619140625, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.9016686081886292, + "num_tokens": 651925340.0, + "step": 16794 + }, + { + "epoch": 2.1364966289276173, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.583175659179688, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9061927795410156, + "num_tokens": 651963377.0, + "step": 16795 + }, + { + "epoch": 2.136623839206208, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.55898380279541, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9109076261520386, + "num_tokens": 652005042.0, + "step": 16796 + }, + { + "epoch": 2.1367510494847983, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.486343383789062e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.55676555633545, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8946646451950073, + "num_tokens": 652044123.0, + "step": 16797 + }, + { + "epoch": 2.136878259763389, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.551580429077148, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8976923823356628, + "num_tokens": 652081771.0, + "step": 16798 + }, + { + "epoch": 2.1370054700419794, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.527328491210938, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9044791460037231, + "num_tokens": 652120290.0, + "step": 16799 + }, + { + "epoch": 2.13713268032057, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.506513595581055, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8986105918884277, + "num_tokens": 652166450.0, + "step": 16800 + }, + { + "epoch": 2.1372598905991604, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.575069427490234, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8977545499801636, + "num_tokens": 652205354.0, + "step": 16801 + }, + { + "epoch": 2.137387100877751, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.602916717529297, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.901240885257721, + "num_tokens": 652242080.0, + "step": 16802 + }, + { + "epoch": 2.1375143111563415, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.676541328430176, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8914794921875, + "num_tokens": 652278263.0, + "step": 16803 + }, + { + "epoch": 2.137641521434932, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.555957794189453, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.90755695104599, + "num_tokens": 652315849.0, + "step": 16804 + }, + { + "epoch": 2.1377687317135226, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.545096397399902, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8959863185882568, + "num_tokens": 652354972.0, + "step": 16805 + }, + { + "epoch": 2.137895941992113, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.57755184173584, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8957384824752808, + "num_tokens": 652390825.0, + "step": 16806 + }, + { + "epoch": 2.1380231522707036, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.54020881652832, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.9007837176322937, + "num_tokens": 652431976.0, + "step": 16807 + }, + { + "epoch": 2.138150362549294, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.554437637329102, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.9011438488960266, + "num_tokens": 652469272.0, + "step": 16808 + }, + { + "epoch": 2.1382775728278847, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.571260452270508, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8818777799606323, + "num_tokens": 652516551.0, + "step": 16809 + }, + { + "epoch": 2.138404783106475, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.506623268127441, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9103643894195557, + "num_tokens": 652561177.0, + "step": 16810 + }, + { + "epoch": 2.1385319933850653, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.622575759887695, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9051856994628906, + "num_tokens": 652595209.0, + "step": 16811 + }, + { + "epoch": 2.1386592036636562, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.632661819458008, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8969632983207703, + "num_tokens": 652635546.0, + "step": 16812 + }, + { + "epoch": 2.1387864139422463, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.470110893249512, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.9037245512008667, + "num_tokens": 652676864.0, + "step": 16813 + }, + { + "epoch": 2.138913624220837, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.593979835510254, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.884001612663269, + "num_tokens": 652717664.0, + "step": 16814 + }, + { + "epoch": 2.1390408344994274, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.654973983764648, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8961936235427856, + "num_tokens": 652751445.0, + "step": 16815 + }, + { + "epoch": 2.139168044778018, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.490843772888184, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8987283706665039, + "num_tokens": 652797498.0, + "step": 16816 + }, + { + "epoch": 2.1392952550566084, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.509774208068848, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9048028588294983, + "num_tokens": 652834243.0, + "step": 16817 + }, + { + "epoch": 2.139422465335199, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.672466278076172, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8925374746322632, + "num_tokens": 652869284.0, + "step": 16818 + }, + { + "epoch": 2.1395496756137895, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.680108070373535, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.899162769317627, + "num_tokens": 652900409.0, + "step": 16819 + }, + { + "epoch": 2.13967688589238, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.523116111755371, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8892472982406616, + "num_tokens": 652946065.0, + "step": 16820 + }, + { + "epoch": 2.1398040961709706, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.748714447021484, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9005883932113647, + "num_tokens": 652980959.0, + "step": 16821 + }, + { + "epoch": 2.139931306449561, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.600646018981934, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8994162678718567, + "num_tokens": 653018673.0, + "step": 16822 + }, + { + "epoch": 2.1400585167281516, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.689245223999023, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8976198434829712, + "num_tokens": 653049569.0, + "step": 16823 + }, + { + "epoch": 2.140185727006742, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.51779556274414, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.9004438519477844, + "num_tokens": 653091528.0, + "step": 16824 + }, + { + "epoch": 2.1403129372853327, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.495081901550293, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8912432193756104, + "num_tokens": 653135208.0, + "step": 16825 + }, + { + "epoch": 2.140440147563923, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.624561309814453, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.9039391875267029, + "num_tokens": 653171321.0, + "step": 16826 + }, + { + "epoch": 2.1405673578425137, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.554173469543457, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.9026216268539429, + "num_tokens": 653215043.0, + "step": 16827 + }, + { + "epoch": 2.1406945681211043, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.688018798828125, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8887169361114502, + "num_tokens": 653252369.0, + "step": 16828 + }, + { + "epoch": 2.140821778399695, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.469812393188477, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8999316692352295, + "num_tokens": 653292001.0, + "step": 16829 + }, + { + "epoch": 2.1409489886782853, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.642763137817383, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8861140012741089, + "num_tokens": 653334111.0, + "step": 16830 + }, + { + "epoch": 2.141076198956876, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.541540145874023, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8963291645050049, + "num_tokens": 653374457.0, + "step": 16831 + }, + { + "epoch": 2.1412034092354664, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.528604507446289, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9073385000228882, + "num_tokens": 653407328.0, + "step": 16832 + }, + { + "epoch": 2.141330619514057, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.516670227050781, + "learning_rate": 1e-06, + "loss": 0.2442, + "mean_token_accuracy": 0.916242778301239, + "num_tokens": 653443760.0, + "step": 16833 + }, + { + "epoch": 2.1414578297926474, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.642485618591309, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8937420845031738, + "num_tokens": 653478242.0, + "step": 16834 + }, + { + "epoch": 2.141585040071238, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.558327674865723, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9047833681106567, + "num_tokens": 653519203.0, + "step": 16835 + }, + { + "epoch": 2.141712250349828, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.509725570678711, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8987687826156616, + "num_tokens": 653560968.0, + "step": 16836 + }, + { + "epoch": 2.141839460628419, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.484619140625, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.905484139919281, + "num_tokens": 653605734.0, + "step": 16837 + }, + { + "epoch": 2.141966670907009, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.508464813232422, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9102709293365479, + "num_tokens": 653645571.0, + "step": 16838 + }, + { + "epoch": 2.1420938811855996, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.63709831237793, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9008615016937256, + "num_tokens": 653680948.0, + "step": 16839 + }, + { + "epoch": 2.14222109146419, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.718666076660156, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8920804262161255, + "num_tokens": 653714323.0, + "step": 16840 + }, + { + "epoch": 2.1423483017427807, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.653674125671387, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.9073631763458252, + "num_tokens": 653749157.0, + "step": 16841 + }, + { + "epoch": 2.142475512021371, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.575018882751465, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9062387347221375, + "num_tokens": 653788450.0, + "step": 16842 + }, + { + "epoch": 2.1426027222999617, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.553450584411621, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9053076505661011, + "num_tokens": 653828862.0, + "step": 16843 + }, + { + "epoch": 2.1427299325785523, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.603899002075195, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8943501710891724, + "num_tokens": 653867336.0, + "step": 16844 + }, + { + "epoch": 2.142857142857143, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.506570816040039, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.9027354121208191, + "num_tokens": 653909965.0, + "step": 16845 + }, + { + "epoch": 2.1429843531357333, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.63418960571289, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8991236686706543, + "num_tokens": 653945165.0, + "step": 16846 + }, + { + "epoch": 2.143111563414324, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.498653411865234, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8978402614593506, + "num_tokens": 653987916.0, + "step": 16847 + }, + { + "epoch": 2.1432387736929144, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.672441482543945, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8988351821899414, + "num_tokens": 654022406.0, + "step": 16848 + }, + { + "epoch": 2.143365983971505, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.54659366607666, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9098600149154663, + "num_tokens": 654057823.0, + "step": 16849 + }, + { + "epoch": 2.1434931942500954, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.590737342834473, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8905555605888367, + "num_tokens": 654098268.0, + "step": 16850 + }, + { + "epoch": 2.143620404528686, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.665083885192871, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8955512642860413, + "num_tokens": 654137188.0, + "step": 16851 + }, + { + "epoch": 2.1437476148072765, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.552438735961914, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8833004832267761, + "num_tokens": 654177654.0, + "step": 16852 + }, + { + "epoch": 2.143874825085867, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.557558059692383, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9037982225418091, + "num_tokens": 654217710.0, + "step": 16853 + }, + { + "epoch": 2.1440020353644575, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.55764102935791, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9077156782150269, + "num_tokens": 654251896.0, + "step": 16854 + }, + { + "epoch": 2.144129245643048, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.462011337280273, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.9054690599441528, + "num_tokens": 654295951.0, + "step": 16855 + }, + { + "epoch": 2.1442564559216386, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.511834144592285, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.9009551405906677, + "num_tokens": 654338695.0, + "step": 16856 + }, + { + "epoch": 2.144383666200229, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.505939483642578, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.9033263325691223, + "num_tokens": 654384382.0, + "step": 16857 + }, + { + "epoch": 2.1445108764788197, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.641640663146973, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8846471905708313, + "num_tokens": 654423503.0, + "step": 16858 + }, + { + "epoch": 2.14463808675741, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.600593566894531, + "learning_rate": 1e-06, + "loss": 0.2624, + "mean_token_accuracy": 0.9092714786529541, + "num_tokens": 654465212.0, + "step": 16859 + }, + { + "epoch": 2.1447652970360007, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.52408218383789, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.9053713083267212, + "num_tokens": 654505214.0, + "step": 16860 + }, + { + "epoch": 2.144892507314591, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.624065399169922, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9111983776092529, + "num_tokens": 654545406.0, + "step": 16861 + }, + { + "epoch": 2.1450197175931813, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.497055053710938, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.9133777618408203, + "num_tokens": 654584387.0, + "step": 16862 + }, + { + "epoch": 2.145146927871772, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.655137062072754, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.9040549397468567, + "num_tokens": 654621577.0, + "step": 16863 + }, + { + "epoch": 2.1452741381503624, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.602235794067383, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8989120721817017, + "num_tokens": 654661308.0, + "step": 16864 + }, + { + "epoch": 2.145401348428953, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.536269187927246, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.9004631638526917, + "num_tokens": 654704771.0, + "step": 16865 + }, + { + "epoch": 2.1455285587075434, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.604255676269531, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9032876491546631, + "num_tokens": 654745047.0, + "step": 16866 + }, + { + "epoch": 2.145655768986134, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.631924629211426, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8862467408180237, + "num_tokens": 654785962.0, + "step": 16867 + }, + { + "epoch": 2.1457829792647245, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.684406280517578, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.9009572267532349, + "num_tokens": 654820950.0, + "step": 16868 + }, + { + "epoch": 2.145910189543315, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.684508323669434, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8877619504928589, + "num_tokens": 654861609.0, + "step": 16869 + }, + { + "epoch": 2.1460373998219056, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.666726112365723, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9028494358062744, + "num_tokens": 654898744.0, + "step": 16870 + }, + { + "epoch": 2.146164610100496, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.677449226379395, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8895453214645386, + "num_tokens": 654936559.0, + "step": 16871 + }, + { + "epoch": 2.1462918203790866, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.541223526000977, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8875790238380432, + "num_tokens": 654983229.0, + "step": 16872 + }, + { + "epoch": 2.146419030657677, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.542980194091797, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.9005556106567383, + "num_tokens": 655023151.0, + "step": 16873 + }, + { + "epoch": 2.1465462409362677, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.748602867126465, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8850746154785156, + "num_tokens": 655058725.0, + "step": 16874 + }, + { + "epoch": 2.146673451214858, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.585687637329102, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8832138776779175, + "num_tokens": 655103747.0, + "step": 16875 + }, + { + "epoch": 2.1468006614934487, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.526379585266113, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.893535852432251, + "num_tokens": 655149512.0, + "step": 16876 + }, + { + "epoch": 2.1469278717720393, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.645066261291504, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8807988166809082, + "num_tokens": 655188047.0, + "step": 16877 + }, + { + "epoch": 2.14705508205063, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.465095520019531, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9065420627593994, + "num_tokens": 655230287.0, + "step": 16878 + }, + { + "epoch": 2.1471822923292203, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.65605640411377, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.900337815284729, + "num_tokens": 655272526.0, + "step": 16879 + }, + { + "epoch": 2.147309502607811, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.579103469848633, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8960434794425964, + "num_tokens": 655306885.0, + "step": 16880 + }, + { + "epoch": 2.1474367128864014, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.496258735656738, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8876768350601196, + "num_tokens": 655353004.0, + "step": 16881 + }, + { + "epoch": 2.147563923164992, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.597236633300781, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.907943844795227, + "num_tokens": 655386585.0, + "step": 16882 + }, + { + "epoch": 2.1476911334435824, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.552175521850586, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9074249863624573, + "num_tokens": 655427151.0, + "step": 16883 + }, + { + "epoch": 2.147818343722173, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.568300247192383, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8918882608413696, + "num_tokens": 655468350.0, + "step": 16884 + }, + { + "epoch": 2.1479455540007635, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.536311149597168, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.9055348634719849, + "num_tokens": 655514969.0, + "step": 16885 + }, + { + "epoch": 2.1480727642793536, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.61448860168457, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9008166193962097, + "num_tokens": 655550848.0, + "step": 16886 + }, + { + "epoch": 2.148199974557944, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.51433277130127, + "learning_rate": 1e-06, + "loss": 0.2575, + "mean_token_accuracy": 0.913012683391571, + "num_tokens": 655593252.0, + "step": 16887 + }, + { + "epoch": 2.1483271848365346, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.517850875854492, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9078814387321472, + "num_tokens": 655632307.0, + "step": 16888 + }, + { + "epoch": 2.148454395115125, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.563824653625488, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8909791707992554, + "num_tokens": 655674915.0, + "step": 16889 + }, + { + "epoch": 2.1485816053937157, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.575193405151367, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.9016858339309692, + "num_tokens": 655713169.0, + "step": 16890 + }, + { + "epoch": 2.148708815672306, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.688694953918457, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8815640211105347, + "num_tokens": 655747421.0, + "step": 16891 + }, + { + "epoch": 2.1488360259508967, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.563421249389648, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.899012565612793, + "num_tokens": 655784566.0, + "step": 16892 + }, + { + "epoch": 2.1489632362294873, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.598891258239746, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8978654742240906, + "num_tokens": 655825056.0, + "step": 16893 + }, + { + "epoch": 2.149090446508078, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.7244873046875, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8970805406570435, + "num_tokens": 655858640.0, + "step": 16894 + }, + { + "epoch": 2.1492176567866683, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.559042930603027, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8997880816459656, + "num_tokens": 655896434.0, + "step": 16895 + }, + { + "epoch": 2.149344867065259, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.590524673461914, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.9027832746505737, + "num_tokens": 655935953.0, + "step": 16896 + }, + { + "epoch": 2.1494720773438494, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.576103210449219, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9018375873565674, + "num_tokens": 655974128.0, + "step": 16897 + }, + { + "epoch": 2.14959928762244, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.539233207702637, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8877546787261963, + "num_tokens": 656019200.0, + "step": 16898 + }, + { + "epoch": 2.1497264979010304, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.561222076416016, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.9010056257247925, + "num_tokens": 656057476.0, + "step": 16899 + }, + { + "epoch": 2.149853708179621, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.574135780334473, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8975039720535278, + "num_tokens": 656101925.0, + "step": 16900 + }, + { + "epoch": 2.1499809184582115, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.590455055236816, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8939727544784546, + "num_tokens": 656139207.0, + "step": 16901 + }, + { + "epoch": 2.150108128736802, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.615103721618652, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.9067317247390747, + "num_tokens": 656181185.0, + "step": 16902 + }, + { + "epoch": 2.1502353390153925, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.551652908325195, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8990287780761719, + "num_tokens": 656221181.0, + "step": 16903 + }, + { + "epoch": 2.150362549293983, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.65067195892334, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8898879289627075, + "num_tokens": 656260651.0, + "step": 16904 + }, + { + "epoch": 2.1504897595725736, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.47566032409668, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8958677053451538, + "num_tokens": 656305078.0, + "step": 16905 + }, + { + "epoch": 2.150616969851164, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.684005737304688, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8909919857978821, + "num_tokens": 656344421.0, + "step": 16906 + }, + { + "epoch": 2.1507441801297547, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.60635757446289, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8999089002609253, + "num_tokens": 656381609.0, + "step": 16907 + }, + { + "epoch": 2.150871390408345, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.6187105178833, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8996294736862183, + "num_tokens": 656418305.0, + "step": 16908 + }, + { + "epoch": 2.1509986006869353, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.612191200256348, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.9001318216323853, + "num_tokens": 656455570.0, + "step": 16909 + }, + { + "epoch": 2.1511258109655262, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.592103958129883, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8873292803764343, + "num_tokens": 656501402.0, + "step": 16910 + }, + { + "epoch": 2.1512530212441163, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.699956893920898, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8940955400466919, + "num_tokens": 656539271.0, + "step": 16911 + }, + { + "epoch": 2.151380231522707, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.520699501037598, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8975609540939331, + "num_tokens": 656577459.0, + "step": 16912 + }, + { + "epoch": 2.1515074418012974, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.642240524291992, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9069939851760864, + "num_tokens": 656613560.0, + "step": 16913 + }, + { + "epoch": 2.151634652079888, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.575645446777344, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8950313329696655, + "num_tokens": 656653543.0, + "step": 16914 + }, + { + "epoch": 2.1517618623584784, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.545220375061035, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9134020805358887, + "num_tokens": 656692630.0, + "step": 16915 + }, + { + "epoch": 2.151889072637069, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.589944839477539, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8970466256141663, + "num_tokens": 656729322.0, + "step": 16916 + }, + { + "epoch": 2.1520162829156595, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.467772483825684, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9117283225059509, + "num_tokens": 656771887.0, + "step": 16917 + }, + { + "epoch": 2.15214349319425, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.683510780334473, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8847128748893738, + "num_tokens": 656811185.0, + "step": 16918 + }, + { + "epoch": 2.1522707034728406, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.584400177001953, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8961048126220703, + "num_tokens": 656852964.0, + "step": 16919 + }, + { + "epoch": 2.152397913751431, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.6108980178833, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8910897970199585, + "num_tokens": 656896740.0, + "step": 16920 + }, + { + "epoch": 2.1525251240300216, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.67199420928955, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8886361122131348, + "num_tokens": 656938377.0, + "step": 16921 + }, + { + "epoch": 2.152652334308612, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.531652450561523, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8941477537155151, + "num_tokens": 656980305.0, + "step": 16922 + }, + { + "epoch": 2.1527795445872027, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.650079727172852, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8928932547569275, + "num_tokens": 657017798.0, + "step": 16923 + }, + { + "epoch": 2.152906754865793, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.653669357299805, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8967536091804504, + "num_tokens": 657056406.0, + "step": 16924 + }, + { + "epoch": 2.1530339651443837, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.605347633361816, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.9034961462020874, + "num_tokens": 657090260.0, + "step": 16925 + }, + { + "epoch": 2.1531611754229742, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.664857864379883, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8946527242660522, + "num_tokens": 657130396.0, + "step": 16926 + }, + { + "epoch": 2.1532883857015648, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.598567962646484, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.9060304164886475, + "num_tokens": 657164061.0, + "step": 16927 + }, + { + "epoch": 2.1534155959801553, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.540629386901855, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9115834832191467, + "num_tokens": 657203544.0, + "step": 16928 + }, + { + "epoch": 2.153542806258746, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.64378833770752, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8959871530532837, + "num_tokens": 657238347.0, + "step": 16929 + }, + { + "epoch": 2.1536700165373364, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.641677856445312, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8933059573173523, + "num_tokens": 657279369.0, + "step": 16930 + }, + { + "epoch": 2.153797226815927, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.47039794921875, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.904628574848175, + "num_tokens": 657325522.0, + "step": 16931 + }, + { + "epoch": 2.1539244370945174, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.601088523864746, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.898891270160675, + "num_tokens": 657366215.0, + "step": 16932 + }, + { + "epoch": 2.154051647373108, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.631967544555664, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8941437602043152, + "num_tokens": 657407063.0, + "step": 16933 + }, + { + "epoch": 2.154178857651698, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.611123085021973, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.9002400040626526, + "num_tokens": 657443417.0, + "step": 16934 + }, + { + "epoch": 2.1543060679302886, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.640950202941895, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.891968846321106, + "num_tokens": 657482609.0, + "step": 16935 + }, + { + "epoch": 2.154433278208879, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.611769676208496, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.884002685546875, + "num_tokens": 657524300.0, + "step": 16936 + }, + { + "epoch": 2.1545604884874696, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.532309532165527, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8963547945022583, + "num_tokens": 657569711.0, + "step": 16937 + }, + { + "epoch": 2.15468769876606, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.568622589111328, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.901671290397644, + "num_tokens": 657608351.0, + "step": 16938 + }, + { + "epoch": 2.1548149090446507, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.559282302856445, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.911630392074585, + "num_tokens": 657648195.0, + "step": 16939 + }, + { + "epoch": 2.154942119323241, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.526308059692383, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.9160659313201904, + "num_tokens": 657682598.0, + "step": 16940 + }, + { + "epoch": 2.1550693296018317, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.688042640686035, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8900268077850342, + "num_tokens": 657717626.0, + "step": 16941 + }, + { + "epoch": 2.1551965398804223, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.665271759033203, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8832905292510986, + "num_tokens": 657754647.0, + "step": 16942 + }, + { + "epoch": 2.155323750159013, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.601183891296387, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8899717926979065, + "num_tokens": 657796371.0, + "step": 16943 + }, + { + "epoch": 2.1554509604376033, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.622979164123535, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8959048390388489, + "num_tokens": 657836595.0, + "step": 16944 + }, + { + "epoch": 2.155578170716194, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.578802108764648, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9049032926559448, + "num_tokens": 657879164.0, + "step": 16945 + }, + { + "epoch": 2.1557053809947844, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.582310676574707, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9093365669250488, + "num_tokens": 657917303.0, + "step": 16946 + }, + { + "epoch": 2.155832591273375, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.558451652526855, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.9063939452171326, + "num_tokens": 657956175.0, + "step": 16947 + }, + { + "epoch": 2.1559598015519654, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.846858978271484, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.90032958984375, + "num_tokens": 657990101.0, + "step": 16948 + }, + { + "epoch": 2.156087011830556, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.550535202026367, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8926311731338501, + "num_tokens": 658032751.0, + "step": 16949 + }, + { + "epoch": 2.1562142221091465, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.60645580291748, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8873176574707031, + "num_tokens": 658072254.0, + "step": 16950 + }, + { + "epoch": 2.156341432387737, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.534478187561035, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8961012363433838, + "num_tokens": 658111383.0, + "step": 16951 + }, + { + "epoch": 2.1564686426663275, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.605857849121094, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8975927829742432, + "num_tokens": 658149611.0, + "step": 16952 + }, + { + "epoch": 2.156595852944918, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.568802833557129, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8991819620132446, + "num_tokens": 658183974.0, + "step": 16953 + }, + { + "epoch": 2.1567230632235086, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.707306861877441, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8959270715713501, + "num_tokens": 658219992.0, + "step": 16954 + }, + { + "epoch": 2.156850273502099, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.620299339294434, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.9015682935714722, + "num_tokens": 658254825.0, + "step": 16955 + }, + { + "epoch": 2.1569774837806897, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.576552391052246, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9108397960662842, + "num_tokens": 658291931.0, + "step": 16956 + }, + { + "epoch": 2.15710469405928, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.46735954284668, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9039832353591919, + "num_tokens": 658336141.0, + "step": 16957 + }, + { + "epoch": 2.1572319043378707, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.601312637329102, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8973857164382935, + "num_tokens": 658381197.0, + "step": 16958 + }, + { + "epoch": 2.157359114616461, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.549816131591797, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.906707763671875, + "num_tokens": 658423190.0, + "step": 16959 + }, + { + "epoch": 2.1574863248950513, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.71469497680664, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9055036902427673, + "num_tokens": 658456882.0, + "step": 16960 + }, + { + "epoch": 2.157613535173642, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.621841430664062, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8959643840789795, + "num_tokens": 658497699.0, + "step": 16961 + }, + { + "epoch": 2.1577407454522324, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.644723892211914, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8972417116165161, + "num_tokens": 658538765.0, + "step": 16962 + }, + { + "epoch": 2.157867955730823, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.682723045349121, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8900413513183594, + "num_tokens": 658574633.0, + "step": 16963 + }, + { + "epoch": 2.1579951660094134, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.621732711791992, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8902576565742493, + "num_tokens": 658617044.0, + "step": 16964 + }, + { + "epoch": 2.158122376288004, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.581985473632812, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9079462885856628, + "num_tokens": 658651144.0, + "step": 16965 + }, + { + "epoch": 2.1582495865665945, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.687150001525879, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.890560507774353, + "num_tokens": 658692721.0, + "step": 16966 + }, + { + "epoch": 2.158376796845185, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.602359771728516, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8900991678237915, + "num_tokens": 658730384.0, + "step": 16967 + }, + { + "epoch": 2.1585040071237755, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.593979835510254, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8928316831588745, + "num_tokens": 658774152.0, + "step": 16968 + }, + { + "epoch": 2.158631217402366, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.540066719055176, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8999227285385132, + "num_tokens": 658816184.0, + "step": 16969 + }, + { + "epoch": 2.1587584276809566, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.549962043762207, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8968225121498108, + "num_tokens": 658854850.0, + "step": 16970 + }, + { + "epoch": 2.158885637959547, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.710943222045898, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8938199877738953, + "num_tokens": 658890747.0, + "step": 16971 + }, + { + "epoch": 2.1590128482381377, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.581398963928223, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.9003012180328369, + "num_tokens": 658931854.0, + "step": 16972 + }, + { + "epoch": 2.159140058516728, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.642610549926758, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8938148021697998, + "num_tokens": 658972382.0, + "step": 16973 + }, + { + "epoch": 2.1592672687953187, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.596517562866211, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9097849130630493, + "num_tokens": 659007308.0, + "step": 16974 + }, + { + "epoch": 2.1593944790739092, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.579182624816895, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8976414203643799, + "num_tokens": 659047917.0, + "step": 16975 + }, + { + "epoch": 2.1595216893524998, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.584092140197754, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9050726294517517, + "num_tokens": 659086318.0, + "step": 16976 + }, + { + "epoch": 2.1596488996310903, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.542466163635254, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8877543807029724, + "num_tokens": 659133014.0, + "step": 16977 + }, + { + "epoch": 2.159776109909681, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.581496238708496, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8950045704841614, + "num_tokens": 659175085.0, + "step": 16978 + }, + { + "epoch": 2.1599033201882714, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.476081848144531, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9144179224967957, + "num_tokens": 659223546.0, + "step": 16979 + }, + { + "epoch": 2.160030530466862, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.593467712402344, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8835538625717163, + "num_tokens": 659268701.0, + "step": 16980 + }, + { + "epoch": 2.1601577407454524, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.651763916015625, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8915440440177917, + "num_tokens": 659307721.0, + "step": 16981 + }, + { + "epoch": 2.160284951024043, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.613038063049316, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8930090665817261, + "num_tokens": 659346699.0, + "step": 16982 + }, + { + "epoch": 2.1604121613026335, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.531743049621582, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9045377969741821, + "num_tokens": 659387576.0, + "step": 16983 + }, + { + "epoch": 2.1605393715812236, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.702300071716309, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8905913829803467, + "num_tokens": 659427170.0, + "step": 16984 + }, + { + "epoch": 2.160666581859814, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.622566223144531, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8887933492660522, + "num_tokens": 659465832.0, + "step": 16985 + }, + { + "epoch": 2.1607937921384046, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.604133605957031, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8914617300033569, + "num_tokens": 659507762.0, + "step": 16986 + }, + { + "epoch": 2.160921002416995, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.531332015991211, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9089545011520386, + "num_tokens": 659543479.0, + "step": 16987 + }, + { + "epoch": 2.1610482126955857, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.487256050109863, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9069108963012695, + "num_tokens": 659586052.0, + "step": 16988 + }, + { + "epoch": 2.161175422974176, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.628911018371582, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8977283239364624, + "num_tokens": 659622129.0, + "step": 16989 + }, + { + "epoch": 2.1613026332527667, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.58806037902832, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8965843915939331, + "num_tokens": 659658664.0, + "step": 16990 + }, + { + "epoch": 2.1614298435313573, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.526483535766602, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8926162719726562, + "num_tokens": 659703547.0, + "step": 16991 + }, + { + "epoch": 2.161557053809948, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.550111770629883, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8870227336883545, + "num_tokens": 659749368.0, + "step": 16992 + }, + { + "epoch": 2.1616842640885383, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.650636672973633, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9096678495407104, + "num_tokens": 659785088.0, + "step": 16993 + }, + { + "epoch": 2.161811474367129, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.563735008239746, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.909125566482544, + "num_tokens": 659821091.0, + "step": 16994 + }, + { + "epoch": 2.1619386846457194, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.651387214660645, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9073264598846436, + "num_tokens": 659854810.0, + "step": 16995 + }, + { + "epoch": 2.16206589492431, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.65384578704834, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8873780965805054, + "num_tokens": 659891463.0, + "step": 16996 + }, + { + "epoch": 2.1621931052029004, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.604595184326172, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.911139965057373, + "num_tokens": 659924830.0, + "step": 16997 + }, + { + "epoch": 2.162320315481491, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.56114673614502, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.9051689505577087, + "num_tokens": 659966904.0, + "step": 16998 + }, + { + "epoch": 2.1624475257600815, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.639274597167969, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9110205769538879, + "num_tokens": 660006237.0, + "step": 16999 + }, + { + "epoch": 2.162574736038672, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.47370719909668, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.913774847984314, + "num_tokens": 660048886.0, + "step": 17000 + }, + { + "epoch": 2.1627019463172625, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.545379638671875, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9032803773880005, + "num_tokens": 660089316.0, + "step": 17001 + }, + { + "epoch": 2.162829156595853, + "ewc_loss": 0.0189208984375, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.1324882507324219e-05, + "grad_norm": 9.608877182006836, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.9003019332885742, + "num_tokens": 660129846.0, + "step": 17002 + }, + { + "epoch": 2.1629563668744436, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.534027099609375e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.600275993347168, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9040697813034058, + "num_tokens": 660167037.0, + "step": 17003 + }, + { + "epoch": 2.163083577153034, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.660930633544922, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.9014525413513184, + "num_tokens": 660202043.0, + "step": 17004 + }, + { + "epoch": 2.1632107874316246, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.603643417358398, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8993911743164062, + "num_tokens": 660241590.0, + "step": 17005 + }, + { + "epoch": 2.163337997710215, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.623579025268555, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9019283056259155, + "num_tokens": 660279602.0, + "step": 17006 + }, + { + "epoch": 2.1634652079888053, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.534907341003418, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8948933482170105, + "num_tokens": 660323020.0, + "step": 17007 + }, + { + "epoch": 2.1635924182673962, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.611345291137695, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.890831708908081, + "num_tokens": 660362859.0, + "step": 17008 + }, + { + "epoch": 2.1637196285459863, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.61540699005127, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8981190323829651, + "num_tokens": 660406008.0, + "step": 17009 + }, + { + "epoch": 2.163846838824577, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.549273490905762, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8959025144577026, + "num_tokens": 660450438.0, + "step": 17010 + }, + { + "epoch": 2.1639740491031674, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.618046760559082, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8990276455879211, + "num_tokens": 660487823.0, + "step": 17011 + }, + { + "epoch": 2.164101259381758, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.567307472229004, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9090203046798706, + "num_tokens": 660522163.0, + "step": 17012 + }, + { + "epoch": 2.1642284696603484, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.65521240234375, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9078432321548462, + "num_tokens": 660559714.0, + "step": 17013 + }, + { + "epoch": 2.164355679938939, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.58157730102539, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8967641592025757, + "num_tokens": 660600706.0, + "step": 17014 + }, + { + "epoch": 2.1644828902175295, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.573272705078125, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.905307948589325, + "num_tokens": 660637299.0, + "step": 17015 + }, + { + "epoch": 2.16461010049612, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.709207534790039, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8982681035995483, + "num_tokens": 660668574.0, + "step": 17016 + }, + { + "epoch": 2.1647373107747105, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.563188552856445, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.915543794631958, + "num_tokens": 660704261.0, + "step": 17017 + }, + { + "epoch": 2.164864521053301, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.585908889770508, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9113259315490723, + "num_tokens": 660741780.0, + "step": 17018 + }, + { + "epoch": 2.1649917313318916, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.669849395751953, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8940380811691284, + "num_tokens": 660779081.0, + "step": 17019 + }, + { + "epoch": 2.165118941610482, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.733642578125, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8979099988937378, + "num_tokens": 660809827.0, + "step": 17020 + }, + { + "epoch": 2.1652461518890727, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.687429428100586, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8869025707244873, + "num_tokens": 660846790.0, + "step": 17021 + }, + { + "epoch": 2.165373362167663, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.61953067779541, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9068236947059631, + "num_tokens": 660879350.0, + "step": 17022 + }, + { + "epoch": 2.1655005724462537, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.632733345031738, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.904050350189209, + "num_tokens": 660913896.0, + "step": 17023 + }, + { + "epoch": 2.1656277827248442, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.603398323059082, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8983889818191528, + "num_tokens": 660952413.0, + "step": 17024 + }, + { + "epoch": 2.1657549930034348, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.707389831542969, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8910349607467651, + "num_tokens": 660989700.0, + "step": 17025 + }, + { + "epoch": 2.1658822032820253, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.530012130737305, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8991129398345947, + "num_tokens": 661030785.0, + "step": 17026 + }, + { + "epoch": 2.166009413560616, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.48483657836914, + "learning_rate": 1e-06, + "loss": 0.2391, + "mean_token_accuracy": 0.9184896945953369, + "num_tokens": 661070739.0, + "step": 17027 + }, + { + "epoch": 2.1661366238392064, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.581783294677734, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9091508388519287, + "num_tokens": 661107188.0, + "step": 17028 + }, + { + "epoch": 2.166263834117797, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.699873924255371, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8825289607048035, + "num_tokens": 661144825.0, + "step": 17029 + }, + { + "epoch": 2.1663910443963874, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.570648193359375, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8947247862815857, + "num_tokens": 661188791.0, + "step": 17030 + }, + { + "epoch": 2.166518254674978, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.572057723999023, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.9086679816246033, + "num_tokens": 661230801.0, + "step": 17031 + }, + { + "epoch": 2.166645464953568, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.60549545288086, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.904171347618103, + "num_tokens": 661271373.0, + "step": 17032 + }, + { + "epoch": 2.1667726752321586, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.70111083984375, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.909733235836029, + "num_tokens": 661304138.0, + "step": 17033 + }, + { + "epoch": 2.166899885510749, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.63723087310791, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8963150382041931, + "num_tokens": 661344496.0, + "step": 17034 + }, + { + "epoch": 2.1670270957893396, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.776540756225586, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8923511505126953, + "num_tokens": 661377620.0, + "step": 17035 + }, + { + "epoch": 2.16715430606793, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.580683708190918, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9073233008384705, + "num_tokens": 661417529.0, + "step": 17036 + }, + { + "epoch": 2.1672815163465207, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.552824020385742, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.9052157998085022, + "num_tokens": 661459569.0, + "step": 17037 + }, + { + "epoch": 2.167408726625111, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.612850189208984, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9127954840660095, + "num_tokens": 661499681.0, + "step": 17038 + }, + { + "epoch": 2.1675359369037017, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.610804557800293, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9064152240753174, + "num_tokens": 661533965.0, + "step": 17039 + }, + { + "epoch": 2.1676631471822922, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.686984062194824, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8893395662307739, + "num_tokens": 661570413.0, + "step": 17040 + }, + { + "epoch": 2.1677903574608828, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.519989967346191, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9107518792152405, + "num_tokens": 661612834.0, + "step": 17041 + }, + { + "epoch": 2.1679175677394733, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.713520050048828, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8851226568222046, + "num_tokens": 661652475.0, + "step": 17042 + }, + { + "epoch": 2.168044778018064, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.628000259399414, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8931194543838501, + "num_tokens": 661688872.0, + "step": 17043 + }, + { + "epoch": 2.1681719882966544, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.599956512451172, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.893256664276123, + "num_tokens": 661726346.0, + "step": 17044 + }, + { + "epoch": 2.168299198575245, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.71272087097168, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8855271935462952, + "num_tokens": 661761579.0, + "step": 17045 + }, + { + "epoch": 2.1684264088538354, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.562939643859863, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8959593772888184, + "num_tokens": 661800292.0, + "step": 17046 + }, + { + "epoch": 2.168553619132426, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.627866744995117, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8939164876937866, + "num_tokens": 661838137.0, + "step": 17047 + }, + { + "epoch": 2.1686808294110165, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.64974594116211, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.9035745859146118, + "num_tokens": 661872608.0, + "step": 17048 + }, + { + "epoch": 2.168808039689607, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.469253540039062, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.9007686972618103, + "num_tokens": 661914988.0, + "step": 17049 + }, + { + "epoch": 2.1689352499681975, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.735746383666992, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8939787149429321, + "num_tokens": 661954947.0, + "step": 17050 + }, + { + "epoch": 2.169062460246788, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.59660530090332, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8916698694229126, + "num_tokens": 661992329.0, + "step": 17051 + }, + { + "epoch": 2.1691896705253786, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.646976470947266, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8960659503936768, + "num_tokens": 662034910.0, + "step": 17052 + }, + { + "epoch": 2.169316880803969, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.581951141357422, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.9035701751708984, + "num_tokens": 662070278.0, + "step": 17053 + }, + { + "epoch": 2.1694440910825596, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.71091365814209, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8879913091659546, + "num_tokens": 662109954.0, + "step": 17054 + }, + { + "epoch": 2.16957130136115, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.618310928344727, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8959496021270752, + "num_tokens": 662144550.0, + "step": 17055 + }, + { + "epoch": 2.1696985116397407, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.583724021911621, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.913020133972168, + "num_tokens": 662181619.0, + "step": 17056 + }, + { + "epoch": 2.169825721918331, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.480454444885254, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.9026939868927002, + "num_tokens": 662222049.0, + "step": 17057 + }, + { + "epoch": 2.1699529321969213, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.592320442199707, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9011444449424744, + "num_tokens": 662265694.0, + "step": 17058 + }, + { + "epoch": 2.170080142475512, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.581477165222168, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8918824195861816, + "num_tokens": 662306391.0, + "step": 17059 + }, + { + "epoch": 2.1702073527541024, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.590165138244629, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.9048323631286621, + "num_tokens": 662344697.0, + "step": 17060 + }, + { + "epoch": 2.170334563032693, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.650323867797852, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.9053248167037964, + "num_tokens": 662384421.0, + "step": 17061 + }, + { + "epoch": 2.1704617733112834, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.479873657226562, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8964117169380188, + "num_tokens": 662431498.0, + "step": 17062 + }, + { + "epoch": 2.170588983589874, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.623116493225098, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9069212079048157, + "num_tokens": 662470349.0, + "step": 17063 + }, + { + "epoch": 2.1707161938684645, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.537650108337402, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9081112742424011, + "num_tokens": 662510154.0, + "step": 17064 + }, + { + "epoch": 2.170843404147055, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.564846992492676, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8965544700622559, + "num_tokens": 662549751.0, + "step": 17065 + }, + { + "epoch": 2.1709706144256455, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.731454849243164, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.889527440071106, + "num_tokens": 662584311.0, + "step": 17066 + }, + { + "epoch": 2.171097824704236, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.527386665344238, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8975948095321655, + "num_tokens": 662630658.0, + "step": 17067 + }, + { + "epoch": 2.1712250349828266, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.532073974609375, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.9016059637069702, + "num_tokens": 662674621.0, + "step": 17068 + }, + { + "epoch": 2.171352245261417, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.74954891204834, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.894584059715271, + "num_tokens": 662710027.0, + "step": 17069 + }, + { + "epoch": 2.1714794555400077, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.623894691467285, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.899828314781189, + "num_tokens": 662747282.0, + "step": 17070 + }, + { + "epoch": 2.171606665818598, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.549463272094727, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8911856412887573, + "num_tokens": 662793578.0, + "step": 17071 + }, + { + "epoch": 2.1717338760971887, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.580648422241211, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8999708294868469, + "num_tokens": 662831407.0, + "step": 17072 + }, + { + "epoch": 2.1718610863757792, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.658474922180176, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9057206511497498, + "num_tokens": 662865546.0, + "step": 17073 + }, + { + "epoch": 2.1719882966543698, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.755744934082031, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9038035273551941, + "num_tokens": 662894762.0, + "step": 17074 + }, + { + "epoch": 2.1721155069329603, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.659875869750977, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8899168968200684, + "num_tokens": 662933862.0, + "step": 17075 + }, + { + "epoch": 2.172242717211551, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.515077590942383, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.892369270324707, + "num_tokens": 662975067.0, + "step": 17076 + }, + { + "epoch": 2.1723699274901414, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.584391593933105, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9016720056533813, + "num_tokens": 663013695.0, + "step": 17077 + }, + { + "epoch": 2.172497137768732, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.572786331176758, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9026973843574524, + "num_tokens": 663053895.0, + "step": 17078 + }, + { + "epoch": 2.1726243480473224, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.556252479553223, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8958073258399963, + "num_tokens": 663094058.0, + "step": 17079 + }, + { + "epoch": 2.172751558325913, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.575170516967773, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.9003725647926331, + "num_tokens": 663131416.0, + "step": 17080 + }, + { + "epoch": 2.1728787686045035, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.545209884643555, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8938754796981812, + "num_tokens": 663172546.0, + "step": 17081 + }, + { + "epoch": 2.1730059788830935, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.645474433898926, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.9068145751953125, + "num_tokens": 663207271.0, + "step": 17082 + }, + { + "epoch": 2.173133189161684, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.535910606384277, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9093526601791382, + "num_tokens": 663252264.0, + "step": 17083 + }, + { + "epoch": 2.1732603994402746, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.578391075134277, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9104011654853821, + "num_tokens": 663288370.0, + "step": 17084 + }, + { + "epoch": 2.173387609718865, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.614378929138184, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8976307511329651, + "num_tokens": 663325099.0, + "step": 17085 + }, + { + "epoch": 2.1735148199974557, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.678665161132812, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.891337513923645, + "num_tokens": 663360968.0, + "step": 17086 + }, + { + "epoch": 2.173642030276046, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.676409721374512, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.901197075843811, + "num_tokens": 663392952.0, + "step": 17087 + }, + { + "epoch": 2.1737692405546367, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.505897521972656, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8990049362182617, + "num_tokens": 663439766.0, + "step": 17088 + }, + { + "epoch": 2.1738964508332272, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.648695945739746, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9051740169525146, + "num_tokens": 663475084.0, + "step": 17089 + }, + { + "epoch": 2.1740236611118178, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.607266426086426, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8918019533157349, + "num_tokens": 663513807.0, + "step": 17090 + }, + { + "epoch": 2.1741508713904083, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.591898918151855, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.890283465385437, + "num_tokens": 663548953.0, + "step": 17091 + }, + { + "epoch": 2.174278081668999, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.621984481811523, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8932492733001709, + "num_tokens": 663587644.0, + "step": 17092 + }, + { + "epoch": 2.1744052919475894, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.552872657775879, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9002934694290161, + "num_tokens": 663628140.0, + "step": 17093 + }, + { + "epoch": 2.17453250222618, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.631866455078125, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8980814814567566, + "num_tokens": 663662693.0, + "step": 17094 + }, + { + "epoch": 2.1746597125047704, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.598800659179688, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.9028100371360779, + "num_tokens": 663701539.0, + "step": 17095 + }, + { + "epoch": 2.174786922783361, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.624195098876953, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8974924683570862, + "num_tokens": 663735151.0, + "step": 17096 + }, + { + "epoch": 2.1749141330619515, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.607587814331055, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8983783721923828, + "num_tokens": 663776383.0, + "step": 17097 + }, + { + "epoch": 2.175041343340542, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.60193920135498, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8889641761779785, + "num_tokens": 663813965.0, + "step": 17098 + }, + { + "epoch": 2.1751685536191325, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.577219009399414, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9090051651000977, + "num_tokens": 663852366.0, + "step": 17099 + }, + { + "epoch": 2.175295763897723, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.519179344177246, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8969365954399109, + "num_tokens": 663901650.0, + "step": 17100 + }, + { + "epoch": 2.1754229741763136, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.758883476257324, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8820621967315674, + "num_tokens": 663937386.0, + "step": 17101 + }, + { + "epoch": 2.175550184454904, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.598143577575684, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8964242339134216, + "num_tokens": 663978161.0, + "step": 17102 + }, + { + "epoch": 2.1756773947334946, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.550821304321289, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9062407612800598, + "num_tokens": 664019505.0, + "step": 17103 + }, + { + "epoch": 2.175804605012085, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.592146873474121, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9084399938583374, + "num_tokens": 664055761.0, + "step": 17104 + }, + { + "epoch": 2.1759318152906753, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.651111602783203, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9077743291854858, + "num_tokens": 664091748.0, + "step": 17105 + }, + { + "epoch": 2.1760590255692662, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.606931686401367, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8985025882720947, + "num_tokens": 664131087.0, + "step": 17106 + }, + { + "epoch": 2.1761862358478563, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.748797416687012, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.8730480074882507, + "num_tokens": 664171014.0, + "step": 17107 + }, + { + "epoch": 2.176313446126447, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.746367454528809, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8937873840332031, + "num_tokens": 664211197.0, + "step": 17108 + }, + { + "epoch": 2.1764406564050374, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.570060729980469, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.907463788986206, + "num_tokens": 664249599.0, + "step": 17109 + }, + { + "epoch": 2.176567866683628, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.556527137756348, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8938370943069458, + "num_tokens": 664291749.0, + "step": 17110 + }, + { + "epoch": 2.1766950769622184, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.669456481933594, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8918960094451904, + "num_tokens": 664329571.0, + "step": 17111 + }, + { + "epoch": 2.176822287240809, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.563339233398438, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9100360870361328, + "num_tokens": 664364946.0, + "step": 17112 + }, + { + "epoch": 2.1769494975193995, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.618080139160156, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8999409079551697, + "num_tokens": 664401723.0, + "step": 17113 + }, + { + "epoch": 2.17707670779799, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.626690864562988, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.9038686752319336, + "num_tokens": 664442339.0, + "step": 17114 + }, + { + "epoch": 2.1772039180765805, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.603078842163086, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8983925580978394, + "num_tokens": 664483357.0, + "step": 17115 + }, + { + "epoch": 2.177331128355171, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.645184516906738, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8949782848358154, + "num_tokens": 664521101.0, + "step": 17116 + }, + { + "epoch": 2.1774583386337616, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.649601936340332, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8934529423713684, + "num_tokens": 664556085.0, + "step": 17117 + }, + { + "epoch": 2.177585548912352, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.49640941619873, + "learning_rate": 1e-06, + "loss": 0.264, + "mean_token_accuracy": 0.9095079302787781, + "num_tokens": 664599877.0, + "step": 17118 + }, + { + "epoch": 2.1777127591909426, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.604970932006836, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9055783748626709, + "num_tokens": 664632473.0, + "step": 17119 + }, + { + "epoch": 2.177839969469533, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.589521408081055, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9119650721549988, + "num_tokens": 664673386.0, + "step": 17120 + }, + { + "epoch": 2.1779671797481237, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.601812362670898, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8930118083953857, + "num_tokens": 664715940.0, + "step": 17121 + }, + { + "epoch": 2.1780943900267142, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.493876457214355, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9015981554985046, + "num_tokens": 664760617.0, + "step": 17122 + }, + { + "epoch": 2.1782216003053048, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.581710815429688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.691009521484375, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8933823704719543, + "num_tokens": 664797120.0, + "step": 17123 + }, + { + "epoch": 2.1783488105838953, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.562044143676758, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9114294648170471, + "num_tokens": 664837718.0, + "step": 17124 + }, + { + "epoch": 2.178476020862486, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.701549530029297, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.889737606048584, + "num_tokens": 664876413.0, + "step": 17125 + }, + { + "epoch": 2.1786032311410763, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.505462646484375, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9115886092185974, + "num_tokens": 664921614.0, + "step": 17126 + }, + { + "epoch": 2.178730441419667, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.602364540100098, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.9050291776657104, + "num_tokens": 664955641.0, + "step": 17127 + }, + { + "epoch": 2.1788576516982574, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.600860595703125, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.901145875453949, + "num_tokens": 664993545.0, + "step": 17128 + }, + { + "epoch": 2.178984861976848, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.58285140991211, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9030942916870117, + "num_tokens": 665032665.0, + "step": 17129 + }, + { + "epoch": 2.179112072255438, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.502069473266602, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.9044214487075806, + "num_tokens": 665075863.0, + "step": 17130 + }, + { + "epoch": 2.1792392825340285, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.612078666687012, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8968952298164368, + "num_tokens": 665118799.0, + "step": 17131 + }, + { + "epoch": 2.179366492812619, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.578622817993164, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.9034916162490845, + "num_tokens": 665159475.0, + "step": 17132 + }, + { + "epoch": 2.1794937030912096, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.675434112548828, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8903148174285889, + "num_tokens": 665199545.0, + "step": 17133 + }, + { + "epoch": 2.1796209133698, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.598780632019043, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9071948528289795, + "num_tokens": 665239105.0, + "step": 17134 + }, + { + "epoch": 2.1797481236483907, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.545995712280273, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9107073545455933, + "num_tokens": 665277467.0, + "step": 17135 + }, + { + "epoch": 2.179875333926981, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.560293197631836, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9087768793106079, + "num_tokens": 665314532.0, + "step": 17136 + }, + { + "epoch": 2.1800025442055717, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.726821899414062, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9056524634361267, + "num_tokens": 665345387.0, + "step": 17137 + }, + { + "epoch": 2.1801297544841622, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.67966365814209, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8908568620681763, + "num_tokens": 665382722.0, + "step": 17138 + }, + { + "epoch": 2.1802569647627528, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.570831298828125, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9116551876068115, + "num_tokens": 665417248.0, + "step": 17139 + }, + { + "epoch": 2.1803841750413433, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.57360553741455, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8993668556213379, + "num_tokens": 665456395.0, + "step": 17140 + }, + { + "epoch": 2.180511385319934, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.65575885772705, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.88629150390625, + "num_tokens": 665493581.0, + "step": 17141 + }, + { + "epoch": 2.1806385955985244, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.629881858825684, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8958941102027893, + "num_tokens": 665536071.0, + "step": 17142 + }, + { + "epoch": 2.180765805877115, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.597155570983887, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9027149677276611, + "num_tokens": 665573309.0, + "step": 17143 + }, + { + "epoch": 2.1808930161557054, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.564559936523438, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.906193196773529, + "num_tokens": 665614430.0, + "step": 17144 + }, + { + "epoch": 2.181020226434296, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.6314058303833, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.895290732383728, + "num_tokens": 665650560.0, + "step": 17145 + }, + { + "epoch": 2.1811474367128865, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.571106910705566, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.9055178165435791, + "num_tokens": 665691136.0, + "step": 17146 + }, + { + "epoch": 2.181274646991477, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.648124694824219, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8964908123016357, + "num_tokens": 665727602.0, + "step": 17147 + }, + { + "epoch": 2.1814018572700675, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.58344554901123, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8870629072189331, + "num_tokens": 665768881.0, + "step": 17148 + }, + { + "epoch": 2.181529067548658, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.665878295898438, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8953148126602173, + "num_tokens": 665805748.0, + "step": 17149 + }, + { + "epoch": 2.1816562778272486, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.556501388549805, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9051526784896851, + "num_tokens": 665847112.0, + "step": 17150 + }, + { + "epoch": 2.181783488105839, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.613676071166992, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.90662682056427, + "num_tokens": 665884769.0, + "step": 17151 + }, + { + "epoch": 2.1819106983844296, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.503905296325684, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9093858003616333, + "num_tokens": 665930719.0, + "step": 17152 + }, + { + "epoch": 2.18203790866302, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.686725616455078, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9085303544998169, + "num_tokens": 665967933.0, + "step": 17153 + }, + { + "epoch": 2.1821651189416107, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.664772987365723, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8927831649780273, + "num_tokens": 666007442.0, + "step": 17154 + }, + { + "epoch": 2.1822923292202008, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.56162166595459, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8918476104736328, + "num_tokens": 666047256.0, + "step": 17155 + }, + { + "epoch": 2.1824195394987913, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.569558143615723, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.9042233228683472, + "num_tokens": 666087281.0, + "step": 17156 + }, + { + "epoch": 2.182546749777382, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.604231834411621, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9050145745277405, + "num_tokens": 666127564.0, + "step": 17157 + }, + { + "epoch": 2.1826739600559724, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.61282730102539, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8938565850257874, + "num_tokens": 666171311.0, + "step": 17158 + }, + { + "epoch": 2.182801170334563, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.709792137145996, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8883106112480164, + "num_tokens": 666208349.0, + "step": 17159 + }, + { + "epoch": 2.1829283806131534, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.720934867858887, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8979293704032898, + "num_tokens": 666240147.0, + "step": 17160 + }, + { + "epoch": 2.183055590891744, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.58304214477539, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8814497590065002, + "num_tokens": 666284639.0, + "step": 17161 + }, + { + "epoch": 2.1831828011703345, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.658332824707031, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8966460227966309, + "num_tokens": 666322571.0, + "step": 17162 + }, + { + "epoch": 2.183310011448925, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.514920234680176, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9080682992935181, + "num_tokens": 666365231.0, + "step": 17163 + }, + { + "epoch": 2.1834372217275155, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.54761791229248, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9084223508834839, + "num_tokens": 666405131.0, + "step": 17164 + }, + { + "epoch": 2.183564432006106, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.6376314163208, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.9022616744041443, + "num_tokens": 666446244.0, + "step": 17165 + }, + { + "epoch": 2.1836916422846966, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.524353981018066, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.9023163318634033, + "num_tokens": 666490742.0, + "step": 17166 + }, + { + "epoch": 2.183818852563287, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.78065013885498, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8970774412155151, + "num_tokens": 666521897.0, + "step": 17167 + }, + { + "epoch": 2.1839460628418776, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.720728874206543, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.888919472694397, + "num_tokens": 666560494.0, + "step": 17168 + }, + { + "epoch": 2.184073273120468, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.641253471374512, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9027553796768188, + "num_tokens": 666593537.0, + "step": 17169 + }, + { + "epoch": 2.1842004833990587, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.648589134216309, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8958147168159485, + "num_tokens": 666629597.0, + "step": 17170 + }, + { + "epoch": 2.1843276936776492, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.566959381103516, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.905035138130188, + "num_tokens": 666667169.0, + "step": 17171 + }, + { + "epoch": 2.1844549039562398, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.61640739440918, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8958579301834106, + "num_tokens": 666703831.0, + "step": 17172 + }, + { + "epoch": 2.1845821142348303, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.688401222229004, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9064394235610962, + "num_tokens": 666740495.0, + "step": 17173 + }, + { + "epoch": 2.184709324513421, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.586413383483887, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.9066466689109802, + "num_tokens": 666781854.0, + "step": 17174 + }, + { + "epoch": 2.1848365347920113, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.604249000549316, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9117227792739868, + "num_tokens": 666830996.0, + "step": 17175 + }, + { + "epoch": 2.184963745070602, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.672845840454102, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8935680389404297, + "num_tokens": 666871029.0, + "step": 17176 + }, + { + "epoch": 2.1850909553491924, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.669926643371582, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9120199680328369, + "num_tokens": 666907651.0, + "step": 17177 + }, + { + "epoch": 2.185218165627783, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.73682975769043, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8983381986618042, + "num_tokens": 666944068.0, + "step": 17178 + }, + { + "epoch": 2.1853453759063735, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.558372497558594, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.9030917286872864, + "num_tokens": 666987069.0, + "step": 17179 + }, + { + "epoch": 2.1854725861849635, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.637205123901367, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.9003424048423767, + "num_tokens": 667024061.0, + "step": 17180 + }, + { + "epoch": 2.185599796463554, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.670077323913574, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8947006464004517, + "num_tokens": 667066876.0, + "step": 17181 + }, + { + "epoch": 2.1857270067421446, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.602776527404785, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8953119516372681, + "num_tokens": 667107856.0, + "step": 17182 + }, + { + "epoch": 2.185854217020735, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.834136962890625, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8852896690368652, + "num_tokens": 667142231.0, + "step": 17183 + }, + { + "epoch": 2.1859814272993257, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.561767578125, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.9064596891403198, + "num_tokens": 667180610.0, + "step": 17184 + }, + { + "epoch": 2.186108637577916, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.693760871887207, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.9056674838066101, + "num_tokens": 667210630.0, + "step": 17185 + }, + { + "epoch": 2.1862358478565067, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.645243644714355, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8974097371101379, + "num_tokens": 667250084.0, + "step": 17186 + }, + { + "epoch": 2.1863630581350972, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.58250904083252, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8872936964035034, + "num_tokens": 667293590.0, + "step": 17187 + }, + { + "epoch": 2.1864902684136878, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.5239839553833, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8930149674415588, + "num_tokens": 667337545.0, + "step": 17188 + }, + { + "epoch": 2.1866174786922783, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.593596458435059, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8909170627593994, + "num_tokens": 667380987.0, + "step": 17189 + }, + { + "epoch": 2.186744688970869, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.619175910949707, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9073413014411926, + "num_tokens": 667416765.0, + "step": 17190 + }, + { + "epoch": 2.1868718992494594, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.718844413757324, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8983449339866638, + "num_tokens": 667453977.0, + "step": 17191 + }, + { + "epoch": 2.18699910952805, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.546490669250488, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9039016962051392, + "num_tokens": 667491197.0, + "step": 17192 + }, + { + "epoch": 2.1871263198066404, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.642521858215332, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.9009619355201721, + "num_tokens": 667533578.0, + "step": 17193 + }, + { + "epoch": 2.187253530085231, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.636387825012207, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9083527326583862, + "num_tokens": 667568842.0, + "step": 17194 + }, + { + "epoch": 2.1873807403638215, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.53608226776123, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8959550857543945, + "num_tokens": 667611506.0, + "step": 17195 + }, + { + "epoch": 2.187507950642412, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.618291854858398, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9025471210479736, + "num_tokens": 667652492.0, + "step": 17196 + }, + { + "epoch": 2.1876351609210025, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.65419864654541, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8946311473846436, + "num_tokens": 667695786.0, + "step": 17197 + }, + { + "epoch": 2.187762371199593, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.699443817138672, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8913353681564331, + "num_tokens": 667739845.0, + "step": 17198 + }, + { + "epoch": 2.1878895814781836, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.67078685760498, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8896533846855164, + "num_tokens": 667777620.0, + "step": 17199 + }, + { + "epoch": 2.188016791756774, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.544650077819824, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8970632553100586, + "num_tokens": 667822248.0, + "step": 17200 + }, + { + "epoch": 2.1881440020353646, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.684839248657227, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9085971117019653, + "num_tokens": 667857124.0, + "step": 17201 + }, + { + "epoch": 2.188271212313955, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.523483276367188, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9118292331695557, + "num_tokens": 667904211.0, + "step": 17202 + }, + { + "epoch": 2.1883984225925452, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.62939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.570591926574707, + "learning_rate": 1e-06, + "loss": 0.2492, + "mean_token_accuracy": 0.9166656732559204, + "num_tokens": 667940094.0, + "step": 17203 + }, + { + "epoch": 2.188525632871136, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.658012390136719, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8967409133911133, + "num_tokens": 667977323.0, + "step": 17204 + }, + { + "epoch": 2.1886528431497263, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.701170921325684, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8929385542869568, + "num_tokens": 668013484.0, + "step": 17205 + }, + { + "epoch": 2.188780053428317, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.719014167785645, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8900091052055359, + "num_tokens": 668051165.0, + "step": 17206 + }, + { + "epoch": 2.1889072637069074, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.59776782989502, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9111717939376831, + "num_tokens": 668089600.0, + "step": 17207 + }, + { + "epoch": 2.189034473985498, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.682169914245605, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.9038155674934387, + "num_tokens": 668126292.0, + "step": 17208 + }, + { + "epoch": 2.1891616842640884, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.711444854736328, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.903375506401062, + "num_tokens": 668160457.0, + "step": 17209 + }, + { + "epoch": 2.189288894542679, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.66073989868164, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8952528238296509, + "num_tokens": 668192174.0, + "step": 17210 + }, + { + "epoch": 2.1894161048212695, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.603776931762695, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.9052311778068542, + "num_tokens": 668235045.0, + "step": 17211 + }, + { + "epoch": 2.18954331509986, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.594324111938477, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9092379808425903, + "num_tokens": 668267755.0, + "step": 17212 + }, + { + "epoch": 2.1896705253784505, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.655640602111816, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8901058435440063, + "num_tokens": 668310558.0, + "step": 17213 + }, + { + "epoch": 2.189797735657041, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.566058158874512, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8936461806297302, + "num_tokens": 668352176.0, + "step": 17214 + }, + { + "epoch": 2.1899249459356316, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.63595199584961, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8972451090812683, + "num_tokens": 668393237.0, + "step": 17215 + }, + { + "epoch": 2.190052156214222, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.758099555969238, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.9067809581756592, + "num_tokens": 668425711.0, + "step": 17216 + }, + { + "epoch": 2.1901793664928126, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.607025146484375, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.905921459197998, + "num_tokens": 668465397.0, + "step": 17217 + }, + { + "epoch": 2.190306576771403, + "ewc_loss": 0.01904296875, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.571975708007812, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.9042502045631409, + "num_tokens": 668504369.0, + "step": 17218 + }, + { + "epoch": 2.1904337870499937, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.579851150512695, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.908478319644928, + "num_tokens": 668540703.0, + "step": 17219 + }, + { + "epoch": 2.1905609973285842, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.647415161132812, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8936818242073059, + "num_tokens": 668578101.0, + "step": 17220 + }, + { + "epoch": 2.1906882076071748, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.678421974182129, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9087450504302979, + "num_tokens": 668612902.0, + "step": 17221 + }, + { + "epoch": 2.1908154178857653, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.487884521484375, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.9013403058052063, + "num_tokens": 668655892.0, + "step": 17222 + }, + { + "epoch": 2.190942628164356, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.622087478637695, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.9004397392272949, + "num_tokens": 668693756.0, + "step": 17223 + }, + { + "epoch": 2.1910698384429463, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.686427116394043, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9031226634979248, + "num_tokens": 668731288.0, + "step": 17224 + }, + { + "epoch": 2.191197048721537, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.684415817260742, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8921612501144409, + "num_tokens": 668766416.0, + "step": 17225 + }, + { + "epoch": 2.1913242590001274, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.583873748779297, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8984928131103516, + "num_tokens": 668808757.0, + "step": 17226 + }, + { + "epoch": 2.191451469278718, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.662809371948242, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8989593982696533, + "num_tokens": 668846821.0, + "step": 17227 + }, + { + "epoch": 2.191578679557308, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.60676097869873, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9100933074951172, + "num_tokens": 668883140.0, + "step": 17228 + }, + { + "epoch": 2.1917058898358985, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.588473320007324, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8913425803184509, + "num_tokens": 668929159.0, + "step": 17229 + }, + { + "epoch": 2.191833100114489, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.676900863647461, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8861096501350403, + "num_tokens": 668970992.0, + "step": 17230 + }, + { + "epoch": 2.1919603103930796, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.663740158081055, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8850756883621216, + "num_tokens": 669011460.0, + "step": 17231 + }, + { + "epoch": 2.19208752067167, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.586404800415039, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8972952961921692, + "num_tokens": 669048673.0, + "step": 17232 + }, + { + "epoch": 2.1922147309502606, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.568470001220703, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9166500568389893, + "num_tokens": 669091727.0, + "step": 17233 + }, + { + "epoch": 2.192341941228851, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.543066024780273, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.9020892381668091, + "num_tokens": 669133013.0, + "step": 17234 + }, + { + "epoch": 2.1924691515074417, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.588976860046387, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.9001103043556213, + "num_tokens": 669170352.0, + "step": 17235 + }, + { + "epoch": 2.1925963617860322, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.656542778015137, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.9034140110015869, + "num_tokens": 669203779.0, + "step": 17236 + }, + { + "epoch": 2.1927235720646228, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.627763748168945, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8983346223831177, + "num_tokens": 669237974.0, + "step": 17237 + }, + { + "epoch": 2.1928507823432133, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.63158893585205, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8948390483856201, + "num_tokens": 669278532.0, + "step": 17238 + }, + { + "epoch": 2.192977992621804, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.556892395019531, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8972362875938416, + "num_tokens": 669322469.0, + "step": 17239 + }, + { + "epoch": 2.1931052029003943, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.649820327758789, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8957600593566895, + "num_tokens": 669361999.0, + "step": 17240 + }, + { + "epoch": 2.193232413178985, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.69668197631836, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.899475634098053, + "num_tokens": 669395764.0, + "step": 17241 + }, + { + "epoch": 2.1933596234575754, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.544464111328125, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8971740007400513, + "num_tokens": 669438693.0, + "step": 17242 + }, + { + "epoch": 2.193486833736166, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.540492057800293, + "learning_rate": 1e-06, + "loss": 0.2474, + "mean_token_accuracy": 0.9171540141105652, + "num_tokens": 669477160.0, + "step": 17243 + }, + { + "epoch": 2.1936140440147565, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.703149795532227, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8976608514785767, + "num_tokens": 669514717.0, + "step": 17244 + }, + { + "epoch": 2.193741254293347, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.678131103515625, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8918482065200806, + "num_tokens": 669549328.0, + "step": 17245 + }, + { + "epoch": 2.1938684645719375, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.602940559387207, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.901085615158081, + "num_tokens": 669588718.0, + "step": 17246 + }, + { + "epoch": 2.193995674850528, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.752680778503418, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8890947103500366, + "num_tokens": 669622696.0, + "step": 17247 + }, + { + "epoch": 2.1941228851291186, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.671882629394531, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.900510311126709, + "num_tokens": 669660820.0, + "step": 17248 + }, + { + "epoch": 2.194250095407709, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.707701683044434, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8975158333778381, + "num_tokens": 669692175.0, + "step": 17249 + }, + { + "epoch": 2.1943773056862996, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.614099502563477, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8948045969009399, + "num_tokens": 669732634.0, + "step": 17250 + }, + { + "epoch": 2.19450451596489, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.527064323425293, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.9067156314849854, + "num_tokens": 669771784.0, + "step": 17251 + }, + { + "epoch": 2.1946317262434807, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.603011131286621, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9032208323478699, + "num_tokens": 669813244.0, + "step": 17252 + }, + { + "epoch": 2.1947589365220708, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.677078247070312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.586101531982422, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.9017263650894165, + "num_tokens": 669851314.0, + "step": 17253 + }, + { + "epoch": 2.1948861468006613, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.611002922058105, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.890262246131897, + "num_tokens": 669893596.0, + "step": 17254 + }, + { + "epoch": 2.195013357079252, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.631265640258789, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9025929570198059, + "num_tokens": 669933272.0, + "step": 17255 + }, + { + "epoch": 2.1951405673578424, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.629070281982422, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.894295334815979, + "num_tokens": 669973021.0, + "step": 17256 + }, + { + "epoch": 2.195267777636433, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.600902557373047, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9104249477386475, + "num_tokens": 670009278.0, + "step": 17257 + }, + { + "epoch": 2.1953949879150234, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.731014251708984, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9053909778594971, + "num_tokens": 670046650.0, + "step": 17258 + }, + { + "epoch": 2.195522198193614, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.658686637878418, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8982923030853271, + "num_tokens": 670080668.0, + "step": 17259 + }, + { + "epoch": 2.1956494084722045, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.569857597351074, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8974403142929077, + "num_tokens": 670123811.0, + "step": 17260 + }, + { + "epoch": 2.195776618750795, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.549654006958008, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8954543471336365, + "num_tokens": 670167260.0, + "step": 17261 + }, + { + "epoch": 2.1959038290293855, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.623331069946289, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.897136926651001, + "num_tokens": 670211602.0, + "step": 17262 + }, + { + "epoch": 2.196031039307976, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.56867504119873, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9066766500473022, + "num_tokens": 670250223.0, + "step": 17263 + }, + { + "epoch": 2.1961582495865666, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.658934593200684, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9002402424812317, + "num_tokens": 670283563.0, + "step": 17264 + }, + { + "epoch": 2.196285459865157, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.600748062133789, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8979507088661194, + "num_tokens": 670324075.0, + "step": 17265 + }, + { + "epoch": 2.1964126701437476, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.564109802246094, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8962417840957642, + "num_tokens": 670365067.0, + "step": 17266 + }, + { + "epoch": 2.196539880422338, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.786576271057129, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9033552408218384, + "num_tokens": 670400306.0, + "step": 17267 + }, + { + "epoch": 2.1966670907009287, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.706964492797852, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8936660289764404, + "num_tokens": 670433249.0, + "step": 17268 + }, + { + "epoch": 2.196794300979519, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.519614219665527, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.9017696380615234, + "num_tokens": 670474930.0, + "step": 17269 + }, + { + "epoch": 2.1969215112581097, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.574787139892578, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9079117774963379, + "num_tokens": 670513543.0, + "step": 17270 + }, + { + "epoch": 2.1970487215367003, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.584153175354004, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.893953263759613, + "num_tokens": 670560083.0, + "step": 17271 + }, + { + "epoch": 2.197175931815291, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.538819313049316, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9067003130912781, + "num_tokens": 670600624.0, + "step": 17272 + }, + { + "epoch": 2.1973031420938813, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.50639820098877, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9012413620948792, + "num_tokens": 670643963.0, + "step": 17273 + }, + { + "epoch": 2.197430352372472, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.625370979309082, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9075868129730225, + "num_tokens": 670677630.0, + "step": 17274 + }, + { + "epoch": 2.1975575626510624, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.583295822143555, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8941989541053772, + "num_tokens": 670717514.0, + "step": 17275 + }, + { + "epoch": 2.197684772929653, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.720683097839355, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8891503214836121, + "num_tokens": 670753339.0, + "step": 17276 + }, + { + "epoch": 2.1978119832082434, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.692181587219238, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.9023870229721069, + "num_tokens": 670791735.0, + "step": 17277 + }, + { + "epoch": 2.1979391934868335, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.62270450592041, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8822241425514221, + "num_tokens": 670834447.0, + "step": 17278 + }, + { + "epoch": 2.198066403765424, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.566934585571289, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.9049948453903198, + "num_tokens": 670877363.0, + "step": 17279 + }, + { + "epoch": 2.1981936140440146, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.729710578918457, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.9005297422409058, + "num_tokens": 670909728.0, + "step": 17280 + }, + { + "epoch": 2.198320824322605, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.563592910766602, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9016872644424438, + "num_tokens": 670950166.0, + "step": 17281 + }, + { + "epoch": 2.1984480346011956, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.7025728225708, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9057511687278748, + "num_tokens": 670981485.0, + "step": 17282 + }, + { + "epoch": 2.198575244879786, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.61327075958252, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8969273567199707, + "num_tokens": 671021636.0, + "step": 17283 + }, + { + "epoch": 2.1987024551583767, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.602869987487793, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8908247351646423, + "num_tokens": 671056877.0, + "step": 17284 + }, + { + "epoch": 2.1988296654369672, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.6686372756958, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8935549259185791, + "num_tokens": 671100373.0, + "step": 17285 + }, + { + "epoch": 2.1989568757155578, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.672441482543945, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.9043972492218018, + "num_tokens": 671135643.0, + "step": 17286 + }, + { + "epoch": 2.1990840859941483, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.571434020996094, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8970247507095337, + "num_tokens": 671176617.0, + "step": 17287 + }, + { + "epoch": 2.199211296272739, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.648415565490723, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9055346250534058, + "num_tokens": 671213583.0, + "step": 17288 + }, + { + "epoch": 2.1993385065513293, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.620311737060547, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8980203866958618, + "num_tokens": 671252915.0, + "step": 17289 + }, + { + "epoch": 2.19946571682992, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.667097091674805, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8981171250343323, + "num_tokens": 671287331.0, + "step": 17290 + }, + { + "epoch": 2.1995929271085104, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.638822555541992, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8952291011810303, + "num_tokens": 671328458.0, + "step": 17291 + }, + { + "epoch": 2.199720137387101, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.52470874786377, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9071463346481323, + "num_tokens": 671370539.0, + "step": 17292 + }, + { + "epoch": 2.1998473476656915, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.641752243041992, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8995934724807739, + "num_tokens": 671413122.0, + "step": 17293 + }, + { + "epoch": 2.199974557944282, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.566632270812988, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9009422659873962, + "num_tokens": 671458965.0, + "step": 17294 + }, + { + "epoch": 2.2001017682228725, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.64721393585205, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8970850110054016, + "num_tokens": 671497160.0, + "step": 17295 + }, + { + "epoch": 2.200228978501463, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.670205116271973, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.892357349395752, + "num_tokens": 671536514.0, + "step": 17296 + }, + { + "epoch": 2.2003561887800536, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.606289863586426, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9052454233169556, + "num_tokens": 671574822.0, + "step": 17297 + }, + { + "epoch": 2.200483399058644, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.776345252990723, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8982336521148682, + "num_tokens": 671608017.0, + "step": 17298 + }, + { + "epoch": 2.2006106093372346, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.635822296142578, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9072861075401306, + "num_tokens": 671642950.0, + "step": 17299 + }, + { + "epoch": 2.200737819615825, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.6544771194458, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8949906229972839, + "num_tokens": 671678855.0, + "step": 17300 + }, + { + "epoch": 2.2008650298944152, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.64796257019043, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.9001268744468689, + "num_tokens": 671717118.0, + "step": 17301 + }, + { + "epoch": 2.200992240173006, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.700879096984863, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8930056691169739, + "num_tokens": 671754317.0, + "step": 17302 + }, + { + "epoch": 2.2011194504515963, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.55232048034668, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.89949631690979, + "num_tokens": 671793357.0, + "step": 17303 + }, + { + "epoch": 2.201246660730187, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.65927791595459, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8948764801025391, + "num_tokens": 671834203.0, + "step": 17304 + }, + { + "epoch": 2.2013738710087773, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.577136993408203, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8953709602355957, + "num_tokens": 671880421.0, + "step": 17305 + }, + { + "epoch": 2.201501081287368, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.699021339416504, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8986202478408813, + "num_tokens": 671915429.0, + "step": 17306 + }, + { + "epoch": 2.2016282915659584, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.66497802734375, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8998690843582153, + "num_tokens": 671952788.0, + "step": 17307 + }, + { + "epoch": 2.201755501844549, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.754197120666504, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8946481943130493, + "num_tokens": 671986068.0, + "step": 17308 + }, + { + "epoch": 2.2018827121231395, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.821520805358887, + "learning_rate": 1e-06, + "loss": 0.3475, + "mean_token_accuracy": 0.882892906665802, + "num_tokens": 672019925.0, + "step": 17309 + }, + { + "epoch": 2.20200992240173, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.552968978881836, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8989623785018921, + "num_tokens": 672062403.0, + "step": 17310 + }, + { + "epoch": 2.2021371326803205, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.7234468460083, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.901459813117981, + "num_tokens": 672095598.0, + "step": 17311 + }, + { + "epoch": 2.202264342958911, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.640742301940918, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8994979858398438, + "num_tokens": 672136945.0, + "step": 17312 + }, + { + "epoch": 2.2023915532375016, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.641862869262695, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.890832781791687, + "num_tokens": 672175628.0, + "step": 17313 + }, + { + "epoch": 2.202518763516092, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.69265079498291, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8992806673049927, + "num_tokens": 672215406.0, + "step": 17314 + }, + { + "epoch": 2.2026459737946826, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.585637092590332, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9077965617179871, + "num_tokens": 672251913.0, + "step": 17315 + }, + { + "epoch": 2.202773184073273, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.703046798706055, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9038654565811157, + "num_tokens": 672287040.0, + "step": 17316 + }, + { + "epoch": 2.2029003943518637, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.685038566589355, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8912041187286377, + "num_tokens": 672323492.0, + "step": 17317 + }, + { + "epoch": 2.203027604630454, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.659334182739258, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8897753357887268, + "num_tokens": 672359579.0, + "step": 17318 + }, + { + "epoch": 2.2031548149090447, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.711820602416992, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9044845700263977, + "num_tokens": 672396509.0, + "step": 17319 + }, + { + "epoch": 2.2032820251876353, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.591824531555176, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.8996409177780151, + "num_tokens": 672432311.0, + "step": 17320 + }, + { + "epoch": 2.203409235466226, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.593019485473633, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.9062574505805969, + "num_tokens": 672471964.0, + "step": 17321 + }, + { + "epoch": 2.2035364457448163, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.5943603515625, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9086480736732483, + "num_tokens": 672507878.0, + "step": 17322 + }, + { + "epoch": 2.203663656023407, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.648053169250488, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8875352144241333, + "num_tokens": 672545562.0, + "step": 17323 + }, + { + "epoch": 2.2037908663019974, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.5678129196167, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9107040166854858, + "num_tokens": 672587620.0, + "step": 17324 + }, + { + "epoch": 2.203918076580588, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.623261451721191, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.9041080474853516, + "num_tokens": 672625670.0, + "step": 17325 + }, + { + "epoch": 2.204045286859178, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.664569854736328, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8959875106811523, + "num_tokens": 672663680.0, + "step": 17326 + }, + { + "epoch": 2.2041724971377685, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.648504257202148, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9071543216705322, + "num_tokens": 672699653.0, + "step": 17327 + }, + { + "epoch": 2.204299707416359, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.545416831970215, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9159369468688965, + "num_tokens": 672737556.0, + "step": 17328 + }, + { + "epoch": 2.2044269176949496, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.671564102172852, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8929054737091064, + "num_tokens": 672778619.0, + "step": 17329 + }, + { + "epoch": 2.20455412797354, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.618303298950195, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.9015421867370605, + "num_tokens": 672818052.0, + "step": 17330 + }, + { + "epoch": 2.2046813382521306, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.621781349182129, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8930591940879822, + "num_tokens": 672856479.0, + "step": 17331 + }, + { + "epoch": 2.204808548530721, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.573612213134766, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.91329026222229, + "num_tokens": 672890413.0, + "step": 17332 + }, + { + "epoch": 2.2049357588093117, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.516386985778809, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9100397229194641, + "num_tokens": 672933593.0, + "step": 17333 + }, + { + "epoch": 2.2050629690879022, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.614230155944824, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9046425223350525, + "num_tokens": 672974475.0, + "step": 17334 + }, + { + "epoch": 2.2051901793664928, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.567076683044434, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.9055877327919006, + "num_tokens": 673016421.0, + "step": 17335 + }, + { + "epoch": 2.2053173896450833, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.73520278930664, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9042476415634155, + "num_tokens": 673053092.0, + "step": 17336 + }, + { + "epoch": 2.205444599923674, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.61709976196289, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9038721323013306, + "num_tokens": 673089812.0, + "step": 17337 + }, + { + "epoch": 2.2055718102022643, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.67009162902832, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.906756579875946, + "num_tokens": 673126270.0, + "step": 17338 + }, + { + "epoch": 2.205699020480855, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.53708553314209, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8948785662651062, + "num_tokens": 673169937.0, + "step": 17339 + }, + { + "epoch": 2.2058262307594454, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.603835105895996, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9122304916381836, + "num_tokens": 673207349.0, + "step": 17340 + }, + { + "epoch": 2.205953441038036, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.627230644226074, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8948900103569031, + "num_tokens": 673248336.0, + "step": 17341 + }, + { + "epoch": 2.2060806513166265, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.76710033416748, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8877924680709839, + "num_tokens": 673289465.0, + "step": 17342 + }, + { + "epoch": 2.206207861595217, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.517878532409668, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8908948302268982, + "num_tokens": 673334748.0, + "step": 17343 + }, + { + "epoch": 2.2063350718738075, + "ewc_loss": 0.0191650390625, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.681859970092773, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.9009053111076355, + "num_tokens": 673371965.0, + "step": 17344 + }, + { + "epoch": 2.206462282152398, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.620843887329102, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8849345445632935, + "num_tokens": 673413205.0, + "step": 17345 + }, + { + "epoch": 2.2065894924309886, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.570093154907227, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9074483513832092, + "num_tokens": 673458400.0, + "step": 17346 + }, + { + "epoch": 2.206716702709579, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.686799049377441, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8965772390365601, + "num_tokens": 673493508.0, + "step": 17347 + }, + { + "epoch": 2.2068439129881696, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.632224082946777, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8909525871276855, + "num_tokens": 673533952.0, + "step": 17348 + }, + { + "epoch": 2.20697112326676, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.7759428024292, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.892771303653717, + "num_tokens": 673564874.0, + "step": 17349 + }, + { + "epoch": 2.2070983335453507, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.611542701721191, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.9080929756164551, + "num_tokens": 673600604.0, + "step": 17350 + }, + { + "epoch": 2.2072255438239408, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.529749870300293, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9057010412216187, + "num_tokens": 673642421.0, + "step": 17351 + }, + { + "epoch": 2.2073527541025313, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.703027725219727, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8982105255126953, + "num_tokens": 673677019.0, + "step": 17352 + }, + { + "epoch": 2.207479964381122, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.590338706970215, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9075755476951599, + "num_tokens": 673717246.0, + "step": 17353 + }, + { + "epoch": 2.2076071746597123, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.663802146911621, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8952011466026306, + "num_tokens": 673754563.0, + "step": 17354 + }, + { + "epoch": 2.207734384938303, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.547380447387695, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.9040797352790833, + "num_tokens": 673798519.0, + "step": 17355 + }, + { + "epoch": 2.2078615952168934, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.663434982299805, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9006716012954712, + "num_tokens": 673837332.0, + "step": 17356 + }, + { + "epoch": 2.207988805495484, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.755282402038574, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.894336462020874, + "num_tokens": 673870171.0, + "step": 17357 + }, + { + "epoch": 2.2081160157740745, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.657792091369629, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8849437236785889, + "num_tokens": 673913148.0, + "step": 17358 + }, + { + "epoch": 2.208243226052665, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.628730773925781, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.9040902853012085, + "num_tokens": 673944819.0, + "step": 17359 + }, + { + "epoch": 2.2083704363312555, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.585227012634277, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8960937261581421, + "num_tokens": 673989439.0, + "step": 17360 + }, + { + "epoch": 2.208497646609846, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.575505256652832, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.9013352394104004, + "num_tokens": 674029523.0, + "step": 17361 + }, + { + "epoch": 2.2086248568884366, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.649384498596191, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8951364755630493, + "num_tokens": 674067645.0, + "step": 17362 + }, + { + "epoch": 2.208752067167027, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.543907165527344, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9031147956848145, + "num_tokens": 674105981.0, + "step": 17363 + }, + { + "epoch": 2.2088792774456176, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.607746124267578, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8836439251899719, + "num_tokens": 674148930.0, + "step": 17364 + }, + { + "epoch": 2.209006487724208, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.68841552734375, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8934421539306641, + "num_tokens": 674187509.0, + "step": 17365 + }, + { + "epoch": 2.2091336980027987, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.55514144897461, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8963733911514282, + "num_tokens": 674228690.0, + "step": 17366 + }, + { + "epoch": 2.209260908281389, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.731873512268066, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8974392414093018, + "num_tokens": 674264223.0, + "step": 17367 + }, + { + "epoch": 2.2093881185599797, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.712953567504883, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.9027197957038879, + "num_tokens": 674299602.0, + "step": 17368 + }, + { + "epoch": 2.2095153288385703, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.503066062927246, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.9006450176239014, + "num_tokens": 674343666.0, + "step": 17369 + }, + { + "epoch": 2.209642539117161, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.530623435974121, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.9072834253311157, + "num_tokens": 674388904.0, + "step": 17370 + }, + { + "epoch": 2.2097697493957513, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.633416175842285, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8991498947143555, + "num_tokens": 674430569.0, + "step": 17371 + }, + { + "epoch": 2.209896959674342, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.648860931396484, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8967732191085815, + "num_tokens": 674467584.0, + "step": 17372 + }, + { + "epoch": 2.2100241699529324, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.627514839172363, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8982924818992615, + "num_tokens": 674506585.0, + "step": 17373 + }, + { + "epoch": 2.210151380231523, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.662578582763672, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8892616033554077, + "num_tokens": 674540585.0, + "step": 17374 + }, + { + "epoch": 2.2102785905101134, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.652871131896973, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8923369646072388, + "num_tokens": 674581216.0, + "step": 17375 + }, + { + "epoch": 2.2104058007887035, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.606245994567871, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8912546634674072, + "num_tokens": 674620985.0, + "step": 17376 + }, + { + "epoch": 2.210533011067294, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.889491081237793, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8991879820823669, + "num_tokens": 674660178.0, + "step": 17377 + }, + { + "epoch": 2.2106602213458846, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.635778427124023, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8821576833724976, + "num_tokens": 674700914.0, + "step": 17378 + }, + { + "epoch": 2.210787431624475, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.640548706054688, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9053609371185303, + "num_tokens": 674736023.0, + "step": 17379 + }, + { + "epoch": 2.2109146419030656, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.708966255187988, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8919358849525452, + "num_tokens": 674770230.0, + "step": 17380 + }, + { + "epoch": 2.211041852181656, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.635605812072754, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8945974111557007, + "num_tokens": 674807684.0, + "step": 17381 + }, + { + "epoch": 2.2111690624602467, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.645472526550293, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.9020506739616394, + "num_tokens": 674843454.0, + "step": 17382 + }, + { + "epoch": 2.211296272738837, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.712929725646973, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8989889621734619, + "num_tokens": 674879473.0, + "step": 17383 + }, + { + "epoch": 2.2114234830174277, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.581574440002441, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8988338708877563, + "num_tokens": 674917122.0, + "step": 17384 + }, + { + "epoch": 2.2115506932960183, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.50351333618164, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.9142065048217773, + "num_tokens": 674957675.0, + "step": 17385 + }, + { + "epoch": 2.211677903574609, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.64660358428955, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8940748572349548, + "num_tokens": 674994230.0, + "step": 17386 + }, + { + "epoch": 2.2118051138531993, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.576416015625, + "learning_rate": 1e-06, + "loss": 0.2461, + "mean_token_accuracy": 0.9134182333946228, + "num_tokens": 675030602.0, + "step": 17387 + }, + { + "epoch": 2.21193232413179, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.546526908874512, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9022433161735535, + "num_tokens": 675064891.0, + "step": 17388 + }, + { + "epoch": 2.2120595344103804, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.8067045211792, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8919459581375122, + "num_tokens": 675094667.0, + "step": 17389 + }, + { + "epoch": 2.212186744688971, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.487491607666016, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.9019542336463928, + "num_tokens": 675144505.0, + "step": 17390 + }, + { + "epoch": 2.2123139549675614, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.588456153869629, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8920904397964478, + "num_tokens": 675184306.0, + "step": 17391 + }, + { + "epoch": 2.212441165246152, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.719461441040039, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.9041788578033447, + "num_tokens": 675221181.0, + "step": 17392 + }, + { + "epoch": 2.2125683755247425, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.620001792907715, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.9017294645309448, + "num_tokens": 675258210.0, + "step": 17393 + }, + { + "epoch": 2.212695585803333, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.615260124206543, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8943794369697571, + "num_tokens": 675294516.0, + "step": 17394 + }, + { + "epoch": 2.2128227960819236, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.631158828735352, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.9020622968673706, + "num_tokens": 675334062.0, + "step": 17395 + }, + { + "epoch": 2.212950006360514, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.583707809448242, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8838580846786499, + "num_tokens": 675377770.0, + "step": 17396 + }, + { + "epoch": 2.2130772166391046, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.620572090148926, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8998560905456543, + "num_tokens": 675418191.0, + "step": 17397 + }, + { + "epoch": 2.213204426917695, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.673639297485352, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8932739496231079, + "num_tokens": 675452785.0, + "step": 17398 + }, + { + "epoch": 2.2133316371962852, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.55386734008789, + "learning_rate": 1e-06, + "loss": 0.2499, + "mean_token_accuracy": 0.914326548576355, + "num_tokens": 675492556.0, + "step": 17399 + }, + { + "epoch": 2.213458847474876, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.585880279541016, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8963549137115479, + "num_tokens": 675535214.0, + "step": 17400 + }, + { + "epoch": 2.2135860577534663, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.67683219909668, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8868294954299927, + "num_tokens": 675575447.0, + "step": 17401 + }, + { + "epoch": 2.213713268032057, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.747441291809082, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.9013445377349854, + "num_tokens": 675607502.0, + "step": 17402 + }, + { + "epoch": 2.2138404783106473, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.579085350036621, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9047385454177856, + "num_tokens": 675651838.0, + "step": 17403 + }, + { + "epoch": 2.213967688589238, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.677786827087402, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8860382437705994, + "num_tokens": 675692594.0, + "step": 17404 + }, + { + "epoch": 2.2140948988678284, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.623775482177734, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8916015028953552, + "num_tokens": 675733609.0, + "step": 17405 + }, + { + "epoch": 2.214222109146419, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.530044555664062, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.907532274723053, + "num_tokens": 675773305.0, + "step": 17406 + }, + { + "epoch": 2.2143493194250095, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.547511100769043, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9056617021560669, + "num_tokens": 675816981.0, + "step": 17407 + }, + { + "epoch": 2.2144765297036, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.632838249206543, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9043633937835693, + "num_tokens": 675859495.0, + "step": 17408 + }, + { + "epoch": 2.2146037399821905, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.652288436889648, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8939729332923889, + "num_tokens": 675894409.0, + "step": 17409 + }, + { + "epoch": 2.214730950260781, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.676559448242188, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.9033685326576233, + "num_tokens": 675932596.0, + "step": 17410 + }, + { + "epoch": 2.2148581605393716, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.571621894836426, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8985769152641296, + "num_tokens": 675977495.0, + "step": 17411 + }, + { + "epoch": 2.214985370817962, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.571846961975098, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.893614649772644, + "num_tokens": 676019713.0, + "step": 17412 + }, + { + "epoch": 2.2151125810965526, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.710210800170898, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8916769027709961, + "num_tokens": 676061477.0, + "step": 17413 + }, + { + "epoch": 2.215239791375143, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.53349781036377, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8929327726364136, + "num_tokens": 676108174.0, + "step": 17414 + }, + { + "epoch": 2.2153670016537337, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.60633373260498, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.9006778001785278, + "num_tokens": 676150625.0, + "step": 17415 + }, + { + "epoch": 2.215494211932324, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.678068161010742, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9033210277557373, + "num_tokens": 676187898.0, + "step": 17416 + }, + { + "epoch": 2.2156214222109147, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.63063907623291, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8981927633285522, + "num_tokens": 676227531.0, + "step": 17417 + }, + { + "epoch": 2.2157486324895053, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.676926612854004, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8954671025276184, + "num_tokens": 676269774.0, + "step": 17418 + }, + { + "epoch": 2.215875842768096, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.659152030944824, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8940666913986206, + "num_tokens": 676313203.0, + "step": 17419 + }, + { + "epoch": 2.2160030530466863, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.603522300720215, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9014043807983398, + "num_tokens": 676352743.0, + "step": 17420 + }, + { + "epoch": 2.216130263325277, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.673417091369629, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.886786937713623, + "num_tokens": 676398620.0, + "step": 17421 + }, + { + "epoch": 2.2162574736038674, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.639606475830078, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.9025943875312805, + "num_tokens": 676439137.0, + "step": 17422 + }, + { + "epoch": 2.216384683882458, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.635210037231445, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8911498785018921, + "num_tokens": 676479337.0, + "step": 17423 + }, + { + "epoch": 2.216511894161048, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.675812721252441, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8918590545654297, + "num_tokens": 676516304.0, + "step": 17424 + }, + { + "epoch": 2.2166391044396385, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.65347671508789, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8903499841690063, + "num_tokens": 676557851.0, + "step": 17425 + }, + { + "epoch": 2.216766314718229, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.831905364990234, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.887823224067688, + "num_tokens": 676593029.0, + "step": 17426 + }, + { + "epoch": 2.2168935249968196, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.54598617553711, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.897525429725647, + "num_tokens": 676637685.0, + "step": 17427 + }, + { + "epoch": 2.21702073527541, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.939218521118164, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8776770234107971, + "num_tokens": 676668721.0, + "step": 17428 + }, + { + "epoch": 2.2171479455540006, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.683302879333496, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8964354991912842, + "num_tokens": 676700776.0, + "step": 17429 + }, + { + "epoch": 2.217275155832591, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.59011173248291, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.9051898717880249, + "num_tokens": 676736383.0, + "step": 17430 + }, + { + "epoch": 2.2174023661111817, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.628885269165039, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8898813128471375, + "num_tokens": 676776682.0, + "step": 17431 + }, + { + "epoch": 2.217529576389772, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.54860782623291, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8946583867073059, + "num_tokens": 676826560.0, + "step": 17432 + }, + { + "epoch": 2.2176567866683627, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.712804794311523, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8764770030975342, + "num_tokens": 676862627.0, + "step": 17433 + }, + { + "epoch": 2.2177839969469533, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.713869094848633, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.9013305902481079, + "num_tokens": 676897601.0, + "step": 17434 + }, + { + "epoch": 2.217911207225544, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.453662872314453, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.9004829525947571, + "num_tokens": 676943677.0, + "step": 17435 + }, + { + "epoch": 2.2180384175041343, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.74254035949707, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8911421298980713, + "num_tokens": 676982662.0, + "step": 17436 + }, + { + "epoch": 2.218165627782725, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.591583251953125, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9139196872711182, + "num_tokens": 677017581.0, + "step": 17437 + }, + { + "epoch": 2.2182928380613154, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.882450103759766, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8944117426872253, + "num_tokens": 677045475.0, + "step": 17438 + }, + { + "epoch": 2.218420048339906, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.62123966217041, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.9006286263465881, + "num_tokens": 677085937.0, + "step": 17439 + }, + { + "epoch": 2.2185472586184964, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.53254222869873, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8952345848083496, + "num_tokens": 677130448.0, + "step": 17440 + }, + { + "epoch": 2.218674468897087, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.5216064453125, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.9009872674942017, + "num_tokens": 677170030.0, + "step": 17441 + }, + { + "epoch": 2.2188016791756775, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.659392356872559, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9019155502319336, + "num_tokens": 677211098.0, + "step": 17442 + }, + { + "epoch": 2.218928889454268, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.626293182373047, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8963025212287903, + "num_tokens": 677249906.0, + "step": 17443 + }, + { + "epoch": 2.2190560997328586, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.666216850280762, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8961519598960876, + "num_tokens": 677286990.0, + "step": 17444 + }, + { + "epoch": 2.219183310011449, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.735369682312012, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8896210193634033, + "num_tokens": 677323912.0, + "step": 17445 + }, + { + "epoch": 2.2193105202900396, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.66042709350586, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9070677757263184, + "num_tokens": 677358235.0, + "step": 17446 + }, + { + "epoch": 2.21943773056863, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.598718643188477, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8998500108718872, + "num_tokens": 677399034.0, + "step": 17447 + }, + { + "epoch": 2.2195649408472207, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.5737886428833, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.9020936489105225, + "num_tokens": 677444039.0, + "step": 17448 + }, + { + "epoch": 2.2196921511258108, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.71720027923584, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.9038782119750977, + "num_tokens": 677479374.0, + "step": 17449 + }, + { + "epoch": 2.2198193614044013, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.57507610321045, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8966370820999146, + "num_tokens": 677518172.0, + "step": 17450 + }, + { + "epoch": 2.219946571682992, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.582978248596191, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8990241289138794, + "num_tokens": 677561087.0, + "step": 17451 + }, + { + "epoch": 2.2200737819615823, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.70900821685791, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8989138603210449, + "num_tokens": 677600441.0, + "step": 17452 + }, + { + "epoch": 2.220200992240173, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.581500053405762, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8954155445098877, + "num_tokens": 677647234.0, + "step": 17453 + }, + { + "epoch": 2.2203282025187634, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.612824440002441, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.907453715801239, + "num_tokens": 677686826.0, + "step": 17454 + }, + { + "epoch": 2.220455412797354, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.66330623626709, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9038872718811035, + "num_tokens": 677721941.0, + "step": 17455 + }, + { + "epoch": 2.2205826230759445, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.61855697631836, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9026609659194946, + "num_tokens": 677760339.0, + "step": 17456 + }, + { + "epoch": 2.220709833354535, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.61294937133789, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8821200132369995, + "num_tokens": 677800819.0, + "step": 17457 + }, + { + "epoch": 2.2208370436331255, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.746050834655762, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.889782190322876, + "num_tokens": 677843638.0, + "step": 17458 + }, + { + "epoch": 2.220964253911716, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.60025405883789, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8889692425727844, + "num_tokens": 677884221.0, + "step": 17459 + }, + { + "epoch": 2.2210914641903066, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.898850440979004, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8919062614440918, + "num_tokens": 677913367.0, + "step": 17460 + }, + { + "epoch": 2.221218674468897, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.584410667419434, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.9015543460845947, + "num_tokens": 677954336.0, + "step": 17461 + }, + { + "epoch": 2.2213458847474876, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.62584114074707, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8903486728668213, + "num_tokens": 677995774.0, + "step": 17462 + }, + { + "epoch": 2.221473095026078, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.493889808654785, + "learning_rate": 1e-06, + "loss": 0.2631, + "mean_token_accuracy": 0.9101635217666626, + "num_tokens": 678041245.0, + "step": 17463 + }, + { + "epoch": 2.2216003053046687, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 10.8084716796875, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9021927118301392, + "num_tokens": 678081000.0, + "step": 17464 + }, + { + "epoch": 2.221727515583259, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.587777137756348, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9083808660507202, + "num_tokens": 678124575.0, + "step": 17465 + }, + { + "epoch": 2.2218547258618497, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.653793334960938, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.897287130355835, + "num_tokens": 678158145.0, + "step": 17466 + }, + { + "epoch": 2.2219819361404403, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.676984786987305, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8956712484359741, + "num_tokens": 678197019.0, + "step": 17467 + }, + { + "epoch": 2.222109146419031, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.575567245483398, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.9109994173049927, + "num_tokens": 678231716.0, + "step": 17468 + }, + { + "epoch": 2.2222363566976213, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.63969612121582, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9062017202377319, + "num_tokens": 678269633.0, + "step": 17469 + }, + { + "epoch": 2.222363566976212, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.646407127380371, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.9006342887878418, + "num_tokens": 678309005.0, + "step": 17470 + }, + { + "epoch": 2.2224907772548024, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.604701042175293, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9084092378616333, + "num_tokens": 678348872.0, + "step": 17471 + }, + { + "epoch": 2.222617987533393, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.698351860046387, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.9012658596038818, + "num_tokens": 678383761.0, + "step": 17472 + }, + { + "epoch": 2.2227451978119834, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.749424934387207, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8908628821372986, + "num_tokens": 678421767.0, + "step": 17473 + }, + { + "epoch": 2.2228724080905735, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.69186019897461, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8975924253463745, + "num_tokens": 678460070.0, + "step": 17474 + }, + { + "epoch": 2.222999618369164, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.692174911499023, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8988649845123291, + "num_tokens": 678495229.0, + "step": 17475 + }, + { + "epoch": 2.2231268286477546, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.632521629333496, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9061585068702698, + "num_tokens": 678534735.0, + "step": 17476 + }, + { + "epoch": 2.223254038926345, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.617659568786621, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.9042693376541138, + "num_tokens": 678577480.0, + "step": 17477 + }, + { + "epoch": 2.2233812492049356, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.536054611206055, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9093358516693115, + "num_tokens": 678621262.0, + "step": 17478 + }, + { + "epoch": 2.223508459483526, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.62078857421875, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.8999379873275757, + "num_tokens": 678662881.0, + "step": 17479 + }, + { + "epoch": 2.2236356697621167, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.597594261169434, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9003788232803345, + "num_tokens": 678706475.0, + "step": 17480 + }, + { + "epoch": 2.223762880040707, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.59694766998291, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.9079568982124329, + "num_tokens": 678749685.0, + "step": 17481 + }, + { + "epoch": 2.2238900903192977, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.675714492797852, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8933149576187134, + "num_tokens": 678786723.0, + "step": 17482 + }, + { + "epoch": 2.2240173005978883, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.837422370910645, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8869131207466125, + "num_tokens": 678824527.0, + "step": 17483 + }, + { + "epoch": 2.224144510876479, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.586434364318848, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.9062541723251343, + "num_tokens": 678861408.0, + "step": 17484 + }, + { + "epoch": 2.2242717211550693, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.8298921585083, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8936030864715576, + "num_tokens": 678893157.0, + "step": 17485 + }, + { + "epoch": 2.22439893143366, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.692310333251953, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8999828100204468, + "num_tokens": 678928730.0, + "step": 17486 + }, + { + "epoch": 2.2245261417122504, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.592816352844238, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9039297103881836, + "num_tokens": 678969933.0, + "step": 17487 + }, + { + "epoch": 2.224653351990841, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.668106079101562, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8943031430244446, + "num_tokens": 679011354.0, + "step": 17488 + }, + { + "epoch": 2.2247805622694314, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.666656494140625, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.900622546672821, + "num_tokens": 679048652.0, + "step": 17489 + }, + { + "epoch": 2.224907772548022, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.679383277893066, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8914458751678467, + "num_tokens": 679089258.0, + "step": 17490 + }, + { + "epoch": 2.2250349828266125, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.680980682373047, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8990858793258667, + "num_tokens": 679123635.0, + "step": 17491 + }, + { + "epoch": 2.225162193105203, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.685384750366211, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8910503387451172, + "num_tokens": 679163805.0, + "step": 17492 + }, + { + "epoch": 2.2252894033837936, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.69233512878418, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.9011675119400024, + "num_tokens": 679201209.0, + "step": 17493 + }, + { + "epoch": 2.225416613662384, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.668914794921875, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8961383104324341, + "num_tokens": 679237345.0, + "step": 17494 + }, + { + "epoch": 2.2255438239409746, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.578815460205078, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8973243832588196, + "num_tokens": 679280861.0, + "step": 17495 + }, + { + "epoch": 2.225671034219565, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.592179298400879, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.9060880541801453, + "num_tokens": 679324149.0, + "step": 17496 + }, + { + "epoch": 2.225798244498155, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.642790794372559, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8968150019645691, + "num_tokens": 679360792.0, + "step": 17497 + }, + { + "epoch": 2.225925454776746, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.618375778198242, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9021357297897339, + "num_tokens": 679400601.0, + "step": 17498 + }, + { + "epoch": 2.2260526650553363, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.659610748291016, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8983859419822693, + "num_tokens": 679437289.0, + "step": 17499 + }, + { + "epoch": 2.226179875333927, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.576835632324219, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8910496234893799, + "num_tokens": 679479575.0, + "step": 17500 + }, + { + "epoch": 2.2263070856125173, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.750168800354004, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8820639848709106, + "num_tokens": 679516156.0, + "step": 17501 + }, + { + "epoch": 2.226434295891108, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.726200103759766, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8970510959625244, + "num_tokens": 679548380.0, + "step": 17502 + }, + { + "epoch": 2.2265615061696984, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.719939231872559, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8860659599304199, + "num_tokens": 679581446.0, + "step": 17503 + }, + { + "epoch": 2.226688716448289, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.500575065612793, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.910792350769043, + "num_tokens": 679621338.0, + "step": 17504 + }, + { + "epoch": 2.2268159267268794, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.652037620544434, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8928806781768799, + "num_tokens": 679665994.0, + "step": 17505 + }, + { + "epoch": 2.22694313700547, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.70526123046875, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.893810510635376, + "num_tokens": 679705603.0, + "step": 17506 + }, + { + "epoch": 2.2270703472840605, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.540971755981445, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9071413278579712, + "num_tokens": 679746306.0, + "step": 17507 + }, + { + "epoch": 2.227197557562651, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.631163597106934, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8934285044670105, + "num_tokens": 679785773.0, + "step": 17508 + }, + { + "epoch": 2.2273247678412416, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.695710182189941, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8919912576675415, + "num_tokens": 679821691.0, + "step": 17509 + }, + { + "epoch": 2.227451978119832, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.601643562316895, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9023985862731934, + "num_tokens": 679862784.0, + "step": 17510 + }, + { + "epoch": 2.2275791883984226, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.583409309387207, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9116140604019165, + "num_tokens": 679899589.0, + "step": 17511 + }, + { + "epoch": 2.227706398677013, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.590310096740723, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8997848033905029, + "num_tokens": 679943877.0, + "step": 17512 + }, + { + "epoch": 2.2278336089556037, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.686141967773438, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.886750340461731, + "num_tokens": 679985777.0, + "step": 17513 + }, + { + "epoch": 2.227960819234194, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.681341171264648, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.9012435078620911, + "num_tokens": 680020764.0, + "step": 17514 + }, + { + "epoch": 2.2280880295127847, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.677339553833008, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8924728631973267, + "num_tokens": 680061266.0, + "step": 17515 + }, + { + "epoch": 2.2282152397913753, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.667805671691895, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.9013102054595947, + "num_tokens": 680095376.0, + "step": 17516 + }, + { + "epoch": 2.228342450069966, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.755539894104004, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8985255360603333, + "num_tokens": 680134886.0, + "step": 17517 + }, + { + "epoch": 2.2284696603485563, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.560832023620605, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.9081041216850281, + "num_tokens": 680177715.0, + "step": 17518 + }, + { + "epoch": 2.228596870627147, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.659066200256348, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.9009276032447815, + "num_tokens": 680219922.0, + "step": 17519 + }, + { + "epoch": 2.2287240809057374, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.677928924560547, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8820906281471252, + "num_tokens": 680262763.0, + "step": 17520 + }, + { + "epoch": 2.228851291184328, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.660039901733398, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9031870365142822, + "num_tokens": 680298324.0, + "step": 17521 + }, + { + "epoch": 2.228978501462918, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.740372657775879, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9027062654495239, + "num_tokens": 680333618.0, + "step": 17522 + }, + { + "epoch": 2.2291057117415085, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.772445678710938e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.627236366271973, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.9011533260345459, + "num_tokens": 680372936.0, + "step": 17523 + }, + { + "epoch": 2.229232922020099, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.57676887512207, + "learning_rate": 1e-06, + "loss": 0.2439, + "mean_token_accuracy": 0.9157100915908813, + "num_tokens": 680409668.0, + "step": 17524 + }, + { + "epoch": 2.2293601322986896, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.730621337890625, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8945882320404053, + "num_tokens": 680448581.0, + "step": 17525 + }, + { + "epoch": 2.22948734257728, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.723692893981934, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8983662128448486, + "num_tokens": 680479417.0, + "step": 17526 + }, + { + "epoch": 2.2296145528558706, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.761306762695312, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8919606804847717, + "num_tokens": 680516970.0, + "step": 17527 + }, + { + "epoch": 2.229741763134461, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.63042163848877, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9055745601654053, + "num_tokens": 680552779.0, + "step": 17528 + }, + { + "epoch": 2.2298689734130517, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.66338062286377, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9033499956130981, + "num_tokens": 680589512.0, + "step": 17529 + }, + { + "epoch": 2.229996183691642, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.72689437866211, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8871585130691528, + "num_tokens": 680629163.0, + "step": 17530 + }, + { + "epoch": 2.2301233939702327, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.71518611907959, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8928239345550537, + "num_tokens": 680668315.0, + "step": 17531 + }, + { + "epoch": 2.2302506042488233, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.624977111816406, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9036521911621094, + "num_tokens": 680707331.0, + "step": 17532 + }, + { + "epoch": 2.230377814527414, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.80106258392334, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.9060426354408264, + "num_tokens": 680737929.0, + "step": 17533 + }, + { + "epoch": 2.2305050248060043, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.658385276794434, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.9021691083908081, + "num_tokens": 680775951.0, + "step": 17534 + }, + { + "epoch": 2.230632235084595, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.601441383361816, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.906290590763092, + "num_tokens": 680816821.0, + "step": 17535 + }, + { + "epoch": 2.2307594453631854, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.642355918884277, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9085278511047363, + "num_tokens": 680851284.0, + "step": 17536 + }, + { + "epoch": 2.230886655641776, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.617536544799805, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.905829906463623, + "num_tokens": 680890054.0, + "step": 17537 + }, + { + "epoch": 2.2310138659203664, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.741222381591797, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8897740840911865, + "num_tokens": 680926893.0, + "step": 17538 + }, + { + "epoch": 2.231141076198957, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.642849922180176, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9060800671577454, + "num_tokens": 680961628.0, + "step": 17539 + }, + { + "epoch": 2.2312682864775475, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.526416778564453, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8995761275291443, + "num_tokens": 681008158.0, + "step": 17540 + }, + { + "epoch": 2.231395496756138, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.723883628845215, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8982170820236206, + "num_tokens": 681047197.0, + "step": 17541 + }, + { + "epoch": 2.2315227070347285, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.6559476852417, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9033972024917603, + "num_tokens": 681086360.0, + "step": 17542 + }, + { + "epoch": 2.231649917313319, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.557716369628906, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8982094526290894, + "num_tokens": 681127872.0, + "step": 17543 + }, + { + "epoch": 2.2317771275919096, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.735516548156738, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.887633204460144, + "num_tokens": 681169742.0, + "step": 17544 + }, + { + "epoch": 2.2319043378705, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.640759468078613, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8898447155952454, + "num_tokens": 681211121.0, + "step": 17545 + }, + { + "epoch": 2.2320315481490907, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.74675178527832, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8934409618377686, + "num_tokens": 681245337.0, + "step": 17546 + }, + { + "epoch": 2.2321587584276807, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.526679992675781, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9129602313041687, + "num_tokens": 681286278.0, + "step": 17547 + }, + { + "epoch": 2.2322859687062713, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.648218154907227, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.902427077293396, + "num_tokens": 681324847.0, + "step": 17548 + }, + { + "epoch": 2.232413178984862, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.701951026916504, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8903506398200989, + "num_tokens": 681363439.0, + "step": 17549 + }, + { + "epoch": 2.2325403892634523, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.763605117797852, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8936805725097656, + "num_tokens": 681401389.0, + "step": 17550 + }, + { + "epoch": 2.232667599542043, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.583172798156738, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.9007585644721985, + "num_tokens": 681446421.0, + "step": 17551 + }, + { + "epoch": 2.2327948098206334, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.639347076416016, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8941507339477539, + "num_tokens": 681484978.0, + "step": 17552 + }, + { + "epoch": 2.232922020099224, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.782126426696777, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8933078050613403, + "num_tokens": 681515581.0, + "step": 17553 + }, + { + "epoch": 2.2330492303778144, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.655526161193848, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.885959267616272, + "num_tokens": 681561709.0, + "step": 17554 + }, + { + "epoch": 2.233176440656405, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.573822021484375, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.907996654510498, + "num_tokens": 681604149.0, + "step": 17555 + }, + { + "epoch": 2.2333036509349955, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.601805686950684, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8972498178482056, + "num_tokens": 681642958.0, + "step": 17556 + }, + { + "epoch": 2.233430861213586, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.588540077209473, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9138021469116211, + "num_tokens": 681677419.0, + "step": 17557 + }, + { + "epoch": 2.2335580714921766, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.668285369873047, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8926728963851929, + "num_tokens": 681715184.0, + "step": 17558 + }, + { + "epoch": 2.233685281770767, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.655915260314941, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8946744203567505, + "num_tokens": 681760042.0, + "step": 17559 + }, + { + "epoch": 2.2338124920493576, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.678361892700195, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9008826017379761, + "num_tokens": 681795182.0, + "step": 17560 + }, + { + "epoch": 2.233939702327948, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.640562057495117, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.8991374969482422, + "num_tokens": 681836740.0, + "step": 17561 + }, + { + "epoch": 2.2340669126065387, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.669737815856934, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8994506001472473, + "num_tokens": 681872348.0, + "step": 17562 + }, + { + "epoch": 2.234194122885129, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.740965843200684, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9021213054656982, + "num_tokens": 681902557.0, + "step": 17563 + }, + { + "epoch": 2.2343213331637197, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.636303901672363, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8919677138328552, + "num_tokens": 681947401.0, + "step": 17564 + }, + { + "epoch": 2.2344485434423103, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.524657249450684, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9044972658157349, + "num_tokens": 681987547.0, + "step": 17565 + }, + { + "epoch": 2.234575753720901, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.66494083404541, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9004020690917969, + "num_tokens": 682023239.0, + "step": 17566 + }, + { + "epoch": 2.2347029639994913, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.594900131225586, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.895684003829956, + "num_tokens": 682065441.0, + "step": 17567 + }, + { + "epoch": 2.234830174278082, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.620284080505371, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8891274929046631, + "num_tokens": 682107680.0, + "step": 17568 + }, + { + "epoch": 2.2349573845566724, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.524714469909668, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.9008695483207703, + "num_tokens": 682145003.0, + "step": 17569 + }, + { + "epoch": 2.235084594835263, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.598013877868652, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.9003678560256958, + "num_tokens": 682189596.0, + "step": 17570 + }, + { + "epoch": 2.2352118051138534, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.695480346679688, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8912897109985352, + "num_tokens": 682228013.0, + "step": 17571 + }, + { + "epoch": 2.2353390153924435, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.608088493347168, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9106082916259766, + "num_tokens": 682265986.0, + "step": 17572 + }, + { + "epoch": 2.235466225671034, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.714146614074707, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.889890193939209, + "num_tokens": 682301958.0, + "step": 17573 + }, + { + "epoch": 2.2355934359496246, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.63206672668457, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9085031747817993, + "num_tokens": 682338971.0, + "step": 17574 + }, + { + "epoch": 2.235720646228215, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.64421558380127, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8971024751663208, + "num_tokens": 682380028.0, + "step": 17575 + }, + { + "epoch": 2.2358478565068056, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.730709075927734, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8979102969169617, + "num_tokens": 682414056.0, + "step": 17576 + }, + { + "epoch": 2.235975066785396, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.614710807800293, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8997490406036377, + "num_tokens": 682456141.0, + "step": 17577 + }, + { + "epoch": 2.2361022770639867, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.716140747070312, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.903207004070282, + "num_tokens": 682495930.0, + "step": 17578 + }, + { + "epoch": 2.236229487342577, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.622903823852539, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8899899125099182, + "num_tokens": 682537849.0, + "step": 17579 + }, + { + "epoch": 2.2363566976211677, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.591489791870117, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.907383143901825, + "num_tokens": 682575156.0, + "step": 17580 + }, + { + "epoch": 2.2364839078997583, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.58137321472168, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9027417898178101, + "num_tokens": 682614985.0, + "step": 17581 + }, + { + "epoch": 2.236611118178349, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.645024299621582, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.901814877986908, + "num_tokens": 682656976.0, + "step": 17582 + }, + { + "epoch": 2.2367383284569393, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.62133502960205, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8929600119590759, + "num_tokens": 682698568.0, + "step": 17583 + }, + { + "epoch": 2.23686553873553, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.58293342590332, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.9001497626304626, + "num_tokens": 682745624.0, + "step": 17584 + }, + { + "epoch": 2.2369927490141204, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.651665687561035, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.9004240036010742, + "num_tokens": 682786984.0, + "step": 17585 + }, + { + "epoch": 2.237119959292711, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.581343650817871, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.900406002998352, + "num_tokens": 682829993.0, + "step": 17586 + }, + { + "epoch": 2.2372471695713014, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.683329582214355, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.9048510193824768, + "num_tokens": 682866920.0, + "step": 17587 + }, + { + "epoch": 2.237374379849892, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.738260269165039, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8899948000907898, + "num_tokens": 682906900.0, + "step": 17588 + }, + { + "epoch": 2.2375015901284825, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.712913513183594, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.892922043800354, + "num_tokens": 682945612.0, + "step": 17589 + }, + { + "epoch": 2.237628800407073, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.67134952545166, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9125717878341675, + "num_tokens": 682985168.0, + "step": 17590 + }, + { + "epoch": 2.2377560106856635, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.670738220214844, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.9068762063980103, + "num_tokens": 683020104.0, + "step": 17591 + }, + { + "epoch": 2.237883220964254, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.616519927978516, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8975034952163696, + "num_tokens": 683057533.0, + "step": 17592 + }, + { + "epoch": 2.2380104312428446, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.611499786376953, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.883023202419281, + "num_tokens": 683100576.0, + "step": 17593 + }, + { + "epoch": 2.238137641521435, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.713958740234375, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8881936073303223, + "num_tokens": 683137628.0, + "step": 17594 + }, + { + "epoch": 2.238264851800025, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.659706115722656, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9013585448265076, + "num_tokens": 683176291.0, + "step": 17595 + }, + { + "epoch": 2.238392062078616, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.62329387664795, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9015049338340759, + "num_tokens": 683214826.0, + "step": 17596 + }, + { + "epoch": 2.2385192723572063, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.760266304016113, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8834525942802429, + "num_tokens": 683256513.0, + "step": 17597 + }, + { + "epoch": 2.238646482635797, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.598895072937012, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9063833355903625, + "num_tokens": 683297513.0, + "step": 17598 + }, + { + "epoch": 2.2387736929143873, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.669774055480957, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8954651355743408, + "num_tokens": 683337463.0, + "step": 17599 + }, + { + "epoch": 2.238900903192978, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.680167198181152, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8999789953231812, + "num_tokens": 683375042.0, + "step": 17600 + }, + { + "epoch": 2.2390281134715684, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.611793518066406, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9082396030426025, + "num_tokens": 683411167.0, + "step": 17601 + }, + { + "epoch": 2.239155323750159, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.632820129394531, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9085391759872437, + "num_tokens": 683454423.0, + "step": 17602 + }, + { + "epoch": 2.2392825340287494, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.610495567321777, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9091362357139587, + "num_tokens": 683496269.0, + "step": 17603 + }, + { + "epoch": 2.23940974430734, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.690345764160156, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.891789436340332, + "num_tokens": 683537368.0, + "step": 17604 + }, + { + "epoch": 2.2395369545859305, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.655462265014648, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8992443680763245, + "num_tokens": 683576881.0, + "step": 17605 + }, + { + "epoch": 2.239664164864521, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.770708084106445, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8976877927780151, + "num_tokens": 683616484.0, + "step": 17606 + }, + { + "epoch": 2.2397913751431116, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.706024169921875, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8784163594245911, + "num_tokens": 683655392.0, + "step": 17607 + }, + { + "epoch": 2.239918585421702, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.854911804199219, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8936289548873901, + "num_tokens": 683686467.0, + "step": 17608 + }, + { + "epoch": 2.2400457957002926, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.689165115356445, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.890355110168457, + "num_tokens": 683725229.0, + "step": 17609 + }, + { + "epoch": 2.240173005978883, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.555185317993164, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.907325029373169, + "num_tokens": 683768431.0, + "step": 17610 + }, + { + "epoch": 2.2403002162574737, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.74338436126709, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.9015728831291199, + "num_tokens": 683804637.0, + "step": 17611 + }, + { + "epoch": 2.240427426536064, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.586321830749512, + "learning_rate": 1e-06, + "loss": 0.2426, + "mean_token_accuracy": 0.9162983298301697, + "num_tokens": 683843722.0, + "step": 17612 + }, + { + "epoch": 2.2405546368146547, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.6889009475708, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8966000080108643, + "num_tokens": 683882528.0, + "step": 17613 + }, + { + "epoch": 2.2406818470932452, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.732549667358398, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.9034363031387329, + "num_tokens": 683919029.0, + "step": 17614 + }, + { + "epoch": 2.2408090573718358, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.657819747924805, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9024865627288818, + "num_tokens": 683958965.0, + "step": 17615 + }, + { + "epoch": 2.2409362676504263, + "ewc_loss": 0.019287109375, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.71318244934082, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9006437063217163, + "num_tokens": 683994367.0, + "step": 17616 + }, + { + "epoch": 2.241063477929017, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.693811416625977, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8994918465614319, + "num_tokens": 684034752.0, + "step": 17617 + }, + { + "epoch": 2.2411906882076074, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.717151641845703, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.879280686378479, + "num_tokens": 684075109.0, + "step": 17618 + }, + { + "epoch": 2.241317898486198, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.766134262084961, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8862694501876831, + "num_tokens": 684106945.0, + "step": 17619 + }, + { + "epoch": 2.241445108764788, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.623391151428223, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9068278670310974, + "num_tokens": 684143840.0, + "step": 17620 + }, + { + "epoch": 2.2415723190433785, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.542341232299805, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.907133162021637, + "num_tokens": 684187218.0, + "step": 17621 + }, + { + "epoch": 2.241699529321969, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.63674545288086, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9019892811775208, + "num_tokens": 684227847.0, + "step": 17622 + }, + { + "epoch": 2.2418267396005596, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.82012939453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.586187362670898, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8892470002174377, + "num_tokens": 684273351.0, + "step": 17623 + }, + { + "epoch": 2.24195394987915, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.867813110351562e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.641590118408203, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8948887586593628, + "num_tokens": 684316512.0, + "step": 17624 + }, + { + "epoch": 2.2420811601577406, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.709165573120117, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.9021992683410645, + "num_tokens": 684353061.0, + "step": 17625 + }, + { + "epoch": 2.242208370436331, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.666457176208496, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8982411026954651, + "num_tokens": 684394002.0, + "step": 17626 + }, + { + "epoch": 2.2423355807149217, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.59921646118164, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8982436656951904, + "num_tokens": 684436981.0, + "step": 17627 + }, + { + "epoch": 2.242462790993512, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.661169052124023, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8957805633544922, + "num_tokens": 684479708.0, + "step": 17628 + }, + { + "epoch": 2.2425900012721027, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.695188522338867, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8939023017883301, + "num_tokens": 684514623.0, + "step": 17629 + }, + { + "epoch": 2.2427172115506933, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.607916831970215, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8976014256477356, + "num_tokens": 684554606.0, + "step": 17630 + }, + { + "epoch": 2.242844421829284, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.66517448425293, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8901273012161255, + "num_tokens": 684594501.0, + "step": 17631 + }, + { + "epoch": 2.2429716321078743, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.745474815368652, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8757967948913574, + "num_tokens": 684635076.0, + "step": 17632 + }, + { + "epoch": 2.243098842386465, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.638188362121582, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8954483866691589, + "num_tokens": 684671173.0, + "step": 17633 + }, + { + "epoch": 2.2432260526650554, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.549914360046387, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.909814178943634, + "num_tokens": 684712032.0, + "step": 17634 + }, + { + "epoch": 2.243353262943646, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.625418663024902, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.9020673036575317, + "num_tokens": 684754214.0, + "step": 17635 + }, + { + "epoch": 2.2434804732222364, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.638688087463379, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8954923152923584, + "num_tokens": 684790618.0, + "step": 17636 + }, + { + "epoch": 2.243607683500827, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.658748626708984, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8818707466125488, + "num_tokens": 684834251.0, + "step": 17637 + }, + { + "epoch": 2.2437348937794175, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.578472137451172, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.9068810939788818, + "num_tokens": 684873975.0, + "step": 17638 + }, + { + "epoch": 2.243862104058008, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.777876853942871, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8818010091781616, + "num_tokens": 684913650.0, + "step": 17639 + }, + { + "epoch": 2.2439893143365985, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.536293029785156, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9066173434257507, + "num_tokens": 684954489.0, + "step": 17640 + }, + { + "epoch": 2.244116524615189, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.682476043701172, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8988410234451294, + "num_tokens": 684998427.0, + "step": 17641 + }, + { + "epoch": 2.2442437348937796, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.621121406555176, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.905754566192627, + "num_tokens": 685041444.0, + "step": 17642 + }, + { + "epoch": 2.24437094517237, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.56166934967041, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9030900001525879, + "num_tokens": 685082015.0, + "step": 17643 + }, + { + "epoch": 2.2444981554509607, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.754345893859863, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8877764940261841, + "num_tokens": 685121365.0, + "step": 17644 + }, + { + "epoch": 2.2446253657295507, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.647428512573242, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.9020392298698425, + "num_tokens": 685160436.0, + "step": 17645 + }, + { + "epoch": 2.2447525760081413, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.71606159210205, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8926051259040833, + "num_tokens": 685198939.0, + "step": 17646 + }, + { + "epoch": 2.244879786286732, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.623064994812012, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.894555389881134, + "num_tokens": 685241687.0, + "step": 17647 + }, + { + "epoch": 2.2450069965653223, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.545572280883789, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.9058806896209717, + "num_tokens": 685283687.0, + "step": 17648 + }, + { + "epoch": 2.245134206843913, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.697880744934082, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9095313549041748, + "num_tokens": 685320642.0, + "step": 17649 + }, + { + "epoch": 2.2452614171225034, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.803201675415039, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8995899558067322, + "num_tokens": 685356915.0, + "step": 17650 + }, + { + "epoch": 2.245388627401094, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.584850311279297, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8897262811660767, + "num_tokens": 685403928.0, + "step": 17651 + }, + { + "epoch": 2.2455158376796844, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.674192428588867, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9105011224746704, + "num_tokens": 685438215.0, + "step": 17652 + }, + { + "epoch": 2.245643047958275, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.653942108154297, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.89792799949646, + "num_tokens": 685477675.0, + "step": 17653 + }, + { + "epoch": 2.2457702582368655, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.588316917419434, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8963721990585327, + "num_tokens": 685520090.0, + "step": 17654 + }, + { + "epoch": 2.245897468515456, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.584985733032227, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9072239398956299, + "num_tokens": 685562814.0, + "step": 17655 + }, + { + "epoch": 2.2460246787940465, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.645577430725098, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9045072793960571, + "num_tokens": 685600256.0, + "step": 17656 + }, + { + "epoch": 2.246151889072637, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.672013282775879, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8907402753829956, + "num_tokens": 685639235.0, + "step": 17657 + }, + { + "epoch": 2.2462790993512276, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.687487602233887, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8940996527671814, + "num_tokens": 685674769.0, + "step": 17658 + }, + { + "epoch": 2.246406309629818, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.645864486694336, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8946089148521423, + "num_tokens": 685716885.0, + "step": 17659 + }, + { + "epoch": 2.2465335199084087, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.800692558288574, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8914641737937927, + "num_tokens": 685758648.0, + "step": 17660 + }, + { + "epoch": 2.246660730186999, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.557012557983398, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.9030386805534363, + "num_tokens": 685799118.0, + "step": 17661 + }, + { + "epoch": 2.2467879404655897, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.580780029296875, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.9049587249755859, + "num_tokens": 685843830.0, + "step": 17662 + }, + { + "epoch": 2.2469151507441802, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.692811965942383, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8998687863349915, + "num_tokens": 685878942.0, + "step": 17663 + }, + { + "epoch": 2.2470423610227708, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.661286354064941, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8918204307556152, + "num_tokens": 685920878.0, + "step": 17664 + }, + { + "epoch": 2.2471695713013613, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.669105529785156, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.893949031829834, + "num_tokens": 685961637.0, + "step": 17665 + }, + { + "epoch": 2.247296781579952, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.608955383300781, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9074234962463379, + "num_tokens": 685999922.0, + "step": 17666 + }, + { + "epoch": 2.2474239918585424, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.578386306762695, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.9106708765029907, + "num_tokens": 686041259.0, + "step": 17667 + }, + { + "epoch": 2.247551202137133, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.583255767822266, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8928947448730469, + "num_tokens": 686083617.0, + "step": 17668 + }, + { + "epoch": 2.2476784124157234, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.642780303955078, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9128729104995728, + "num_tokens": 686117268.0, + "step": 17669 + }, + { + "epoch": 2.2478056226943135, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.718391418457031, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8888943195343018, + "num_tokens": 686151596.0, + "step": 17670 + }, + { + "epoch": 2.247932832972904, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.676544189453125, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8915870189666748, + "num_tokens": 686189507.0, + "step": 17671 + }, + { + "epoch": 2.2480600432514946, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.586282730102539, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.9047664403915405, + "num_tokens": 686232669.0, + "step": 17672 + }, + { + "epoch": 2.248187253530085, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.719517707824707, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9029149413108826, + "num_tokens": 686264534.0, + "step": 17673 + }, + { + "epoch": 2.2483144638086756, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.651969909667969, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8939026594161987, + "num_tokens": 686303050.0, + "step": 17674 + }, + { + "epoch": 2.248441674087266, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.567543029785156, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.9013454914093018, + "num_tokens": 686348941.0, + "step": 17675 + }, + { + "epoch": 2.2485688843658567, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.660433769226074, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.899703860282898, + "num_tokens": 686390219.0, + "step": 17676 + }, + { + "epoch": 2.248696094644447, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.675148010253906, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8936444520950317, + "num_tokens": 686429448.0, + "step": 17677 + }, + { + "epoch": 2.2488233049230377, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.613794326782227, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8960897922515869, + "num_tokens": 686475001.0, + "step": 17678 + }, + { + "epoch": 2.2489505152016283, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.59959602355957, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.9033896327018738, + "num_tokens": 686516733.0, + "step": 17679 + }, + { + "epoch": 2.249077725480219, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.709341049194336, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8956024646759033, + "num_tokens": 686553327.0, + "step": 17680 + }, + { + "epoch": 2.2492049357588093, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.677908897399902, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8943428993225098, + "num_tokens": 686592710.0, + "step": 17681 + }, + { + "epoch": 2.2493321460374, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.654091835021973, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9008009433746338, + "num_tokens": 686632013.0, + "step": 17682 + }, + { + "epoch": 2.2494593563159904, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.690240859985352, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8958181142807007, + "num_tokens": 686671396.0, + "step": 17683 + }, + { + "epoch": 2.249586566594581, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.697912216186523, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8938512802124023, + "num_tokens": 686712847.0, + "step": 17684 + }, + { + "epoch": 2.2497137768731714, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.628755569458008, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9003154039382935, + "num_tokens": 686746683.0, + "step": 17685 + }, + { + "epoch": 2.249840987151762, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.56141471862793, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8913284540176392, + "num_tokens": 686792221.0, + "step": 17686 + }, + { + "epoch": 2.2499681974303525, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.579384803771973, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.9018232226371765, + "num_tokens": 686835261.0, + "step": 17687 + }, + { + "epoch": 2.250095407708943, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.670860290527344, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8918511867523193, + "num_tokens": 686875873.0, + "step": 17688 + }, + { + "epoch": 2.2502226179875335, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.682671546936035, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.900597333908081, + "num_tokens": 686915061.0, + "step": 17689 + }, + { + "epoch": 2.250349828266124, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.647566795349121, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8948314189910889, + "num_tokens": 686953612.0, + "step": 17690 + }, + { + "epoch": 2.2504770385447146, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.80817985534668, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8987160921096802, + "num_tokens": 686985996.0, + "step": 17691 + }, + { + "epoch": 2.250604248823305, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.667430877685547, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8957810401916504, + "num_tokens": 687024979.0, + "step": 17692 + }, + { + "epoch": 2.250731459101895, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.542710304260254, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9030196666717529, + "num_tokens": 687067140.0, + "step": 17693 + }, + { + "epoch": 2.250858669380486, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.748872756958008, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8952996134757996, + "num_tokens": 687101193.0, + "step": 17694 + }, + { + "epoch": 2.2509858796590763, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.669943809509277, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.896306037902832, + "num_tokens": 687139389.0, + "step": 17695 + }, + { + "epoch": 2.251113089937667, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.856823921203613, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8936609625816345, + "num_tokens": 687170265.0, + "step": 17696 + }, + { + "epoch": 2.2512403002162573, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.678496360778809, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8948527574539185, + "num_tokens": 687211456.0, + "step": 17697 + }, + { + "epoch": 2.251367510494848, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.56649398803711, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.9023456573486328, + "num_tokens": 687254186.0, + "step": 17698 + }, + { + "epoch": 2.2514947207734384, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.661140441894531, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.9016181230545044, + "num_tokens": 687294155.0, + "step": 17699 + }, + { + "epoch": 2.251621931052029, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.564957618713379, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9018961191177368, + "num_tokens": 687335144.0, + "step": 17700 + }, + { + "epoch": 2.2517491413306194, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.611109733581543, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.900958776473999, + "num_tokens": 687376033.0, + "step": 17701 + }, + { + "epoch": 2.25187635160921, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.672079086303711, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8931563496589661, + "num_tokens": 687416948.0, + "step": 17702 + }, + { + "epoch": 2.2520035618878005, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.636045455932617, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.887580931186676, + "num_tokens": 687456114.0, + "step": 17703 + }, + { + "epoch": 2.252130772166391, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.691322326660156, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.9007359743118286, + "num_tokens": 687495892.0, + "step": 17704 + }, + { + "epoch": 2.2522579824449815, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.624319076538086, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.9057767391204834, + "num_tokens": 687531400.0, + "step": 17705 + }, + { + "epoch": 2.252385192723572, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.651994705200195, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9084963202476501, + "num_tokens": 687569092.0, + "step": 17706 + }, + { + "epoch": 2.2525124030021626, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.808215141296387, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8954010009765625, + "num_tokens": 687611884.0, + "step": 17707 + }, + { + "epoch": 2.252639613280753, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.507148742675781, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.904678463935852, + "num_tokens": 687659858.0, + "step": 17708 + }, + { + "epoch": 2.2527668235593437, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.647337913513184, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.9008258581161499, + "num_tokens": 687699947.0, + "step": 17709 + }, + { + "epoch": 2.252894033837934, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.664046287536621, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9005280137062073, + "num_tokens": 687738719.0, + "step": 17710 + }, + { + "epoch": 2.2530212441165247, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.65442943572998, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8985730409622192, + "num_tokens": 687773013.0, + "step": 17711 + }, + { + "epoch": 2.2531484543951152, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.698174476623535, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8909825682640076, + "num_tokens": 687813162.0, + "step": 17712 + }, + { + "epoch": 2.2532756646737058, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.597648620605469, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8931828141212463, + "num_tokens": 687853411.0, + "step": 17713 + }, + { + "epoch": 2.2534028749522963, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.619751930236816, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.9006422162055969, + "num_tokens": 687893970.0, + "step": 17714 + }, + { + "epoch": 2.253530085230887, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.915496826171875e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.636991500854492, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.9022547602653503, + "num_tokens": 687933588.0, + "step": 17715 + }, + { + "epoch": 2.2536572955094774, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.963180541992188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.689480781555176, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8951497077941895, + "num_tokens": 687972844.0, + "step": 17716 + }, + { + "epoch": 2.253784505788068, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 7.963180541992188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.599662780761719, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.9002887606620789, + "num_tokens": 688013798.0, + "step": 17717 + }, + { + "epoch": 2.253911716066658, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.963180541992188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.615677833557129, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9121292233467102, + "num_tokens": 688051327.0, + "step": 17718 + }, + { + "epoch": 2.254038926345249, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.963180541992188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.592167854309082, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.903420090675354, + "num_tokens": 688092548.0, + "step": 17719 + }, + { + "epoch": 2.254166136623839, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.963180541992188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.670787811279297, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.903326153755188, + "num_tokens": 688129715.0, + "step": 17720 + }, + { + "epoch": 2.2542933469024296, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.963180541992188e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.613083839416504, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.9031962156295776, + "num_tokens": 688169024.0, + "step": 17721 + }, + { + "epoch": 2.25442055718102, + "ewc_loss": 0.0194091796875, + "ewc_loss_diag": 7.963180541992188e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.724515914916992, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8905356526374817, + "num_tokens": 688208475.0, + "step": 17722 + }, + { + "epoch": 2.2545477674596106, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.963180541992188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.653419494628906, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.9052833318710327, + "num_tokens": 688242905.0, + "step": 17723 + }, + { + "epoch": 2.254674977738201, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.963180541992188e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.630425453186035, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8911921977996826, + "num_tokens": 688281906.0, + "step": 17724 + }, + { + "epoch": 2.2548021880167917, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.963180541992188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.676870346069336, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9065373539924622, + "num_tokens": 688318172.0, + "step": 17725 + }, + { + "epoch": 2.254929398295382, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 7.963180541992188e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.555261611938477, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.9001543521881104, + "num_tokens": 688362396.0, + "step": 17726 + }, + { + "epoch": 2.2550566085739727, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.737464904785156, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8932526111602783, + "num_tokens": 688401218.0, + "step": 17727 + }, + { + "epoch": 2.2551838188525632, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.651832580566406, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8965073823928833, + "num_tokens": 688438510.0, + "step": 17728 + }, + { + "epoch": 2.2553110291311538, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.651888847351074, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9032323360443115, + "num_tokens": 688480402.0, + "step": 17729 + }, + { + "epoch": 2.2554382394097443, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.608321189880371, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.906231701374054, + "num_tokens": 688523590.0, + "step": 17730 + }, + { + "epoch": 2.255565449688335, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.652397155761719, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8967068195343018, + "num_tokens": 688563279.0, + "step": 17731 + }, + { + "epoch": 2.2556926599669254, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.646719932556152, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8931198120117188, + "num_tokens": 688602665.0, + "step": 17732 + }, + { + "epoch": 2.255819870245516, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.5623197555542, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8996946215629578, + "num_tokens": 688647566.0, + "step": 17733 + }, + { + "epoch": 2.2559470805241064, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.69731330871582, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8901690244674683, + "num_tokens": 688691309.0, + "step": 17734 + }, + { + "epoch": 2.256074290802697, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.729350090026855, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.897596538066864, + "num_tokens": 688728722.0, + "step": 17735 + }, + { + "epoch": 2.2562015010812875, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.612977027893066, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8937154412269592, + "num_tokens": 688769794.0, + "step": 17736 + }, + { + "epoch": 2.256328711359878, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.6914644241333, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.905096173286438, + "num_tokens": 688805483.0, + "step": 17737 + }, + { + "epoch": 2.2564559216384685, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.470048904418945, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9155123233795166, + "num_tokens": 688851187.0, + "step": 17738 + }, + { + "epoch": 2.256583131917059, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.620306015014648, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.9031575322151184, + "num_tokens": 688891570.0, + "step": 17739 + }, + { + "epoch": 2.2567103421956496, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.750067710876465, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.892154335975647, + "num_tokens": 688927144.0, + "step": 17740 + }, + { + "epoch": 2.2568375524742397, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.576006889343262, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.898550271987915, + "num_tokens": 688965163.0, + "step": 17741 + }, + { + "epoch": 2.2569647627528306, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.692224502563477, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8993417024612427, + "num_tokens": 689003437.0, + "step": 17742 + }, + { + "epoch": 2.2570919730314207, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.680593490600586, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.901698112487793, + "num_tokens": 689036648.0, + "step": 17743 + }, + { + "epoch": 2.2572191833100113, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.569575309753418, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9093527793884277, + "num_tokens": 689076073.0, + "step": 17744 + }, + { + "epoch": 2.257346393588602, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.617536544799805, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8911435604095459, + "num_tokens": 689122435.0, + "step": 17745 + }, + { + "epoch": 2.2574736038671923, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.619227409362793, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8930964469909668, + "num_tokens": 689166493.0, + "step": 17746 + }, + { + "epoch": 2.257600814145783, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.605886459350586, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9070216417312622, + "num_tokens": 689205517.0, + "step": 17747 + }, + { + "epoch": 2.2577280244243734, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.63493537902832, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.908176839351654, + "num_tokens": 689242492.0, + "step": 17748 + }, + { + "epoch": 2.257855234702964, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.679317474365234, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8959831595420837, + "num_tokens": 689283876.0, + "step": 17749 + }, + { + "epoch": 2.2579824449815544, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.77600383758545, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.889366090297699, + "num_tokens": 689320816.0, + "step": 17750 + }, + { + "epoch": 2.258109655260145, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.71044921875, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.9005147218704224, + "num_tokens": 689358596.0, + "step": 17751 + }, + { + "epoch": 2.2582368655387355, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.640478134155273, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8876309394836426, + "num_tokens": 689403414.0, + "step": 17752 + }, + { + "epoch": 2.258364075817326, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.737242698669434, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8912837505340576, + "num_tokens": 689439110.0, + "step": 17753 + }, + { + "epoch": 2.2584912860959165, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.604297637939453, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8952275514602661, + "num_tokens": 689482732.0, + "step": 17754 + }, + { + "epoch": 2.258618496374507, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.679153442382812, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9077202677726746, + "num_tokens": 689522640.0, + "step": 17755 + }, + { + "epoch": 2.2587457066530976, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.603643417358398, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.899959921836853, + "num_tokens": 689562234.0, + "step": 17756 + }, + { + "epoch": 2.258872916931688, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.815363883972168, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8925743103027344, + "num_tokens": 689594692.0, + "step": 17757 + }, + { + "epoch": 2.2590001272102787, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.635444641113281, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9065046310424805, + "num_tokens": 689636195.0, + "step": 17758 + }, + { + "epoch": 2.259127337488869, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.723179817199707, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.9008750319480896, + "num_tokens": 689673989.0, + "step": 17759 + }, + { + "epoch": 2.2592545477674597, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.661161422729492, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8964978456497192, + "num_tokens": 689716155.0, + "step": 17760 + }, + { + "epoch": 2.2593817580460502, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.669808387756348, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.9033563733100891, + "num_tokens": 689755768.0, + "step": 17761 + }, + { + "epoch": 2.2595089683246408, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.652450561523438, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9035122990608215, + "num_tokens": 689792247.0, + "step": 17762 + }, + { + "epoch": 2.2596361786032313, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.735581398010254, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8890210390090942, + "num_tokens": 689833936.0, + "step": 17763 + }, + { + "epoch": 2.259763388881822, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.676009178161621, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.905135452747345, + "num_tokens": 689875585.0, + "step": 17764 + }, + { + "epoch": 2.2598905991604123, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.708222389221191, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9036544561386108, + "num_tokens": 689910927.0, + "step": 17765 + }, + { + "epoch": 2.2600178094390024, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.834240913391113, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8973602056503296, + "num_tokens": 689943095.0, + "step": 17766 + }, + { + "epoch": 2.2601450197175934, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.69935417175293, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.9029958248138428, + "num_tokens": 689983394.0, + "step": 17767 + }, + { + "epoch": 2.2602722299961835, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.6496000289917, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.9004567861557007, + "num_tokens": 690023970.0, + "step": 17768 + }, + { + "epoch": 2.260399440274774, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.661608695983887, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.8987339735031128, + "num_tokens": 690062448.0, + "step": 17769 + }, + { + "epoch": 2.2605266505533645, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.594291687011719, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.906402587890625, + "num_tokens": 690099763.0, + "step": 17770 + }, + { + "epoch": 2.260653860831955, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.702065467834473, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8992600440979004, + "num_tokens": 690138877.0, + "step": 17771 + }, + { + "epoch": 2.2607810711105456, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.613993644714355, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.9010692238807678, + "num_tokens": 690179436.0, + "step": 17772 + }, + { + "epoch": 2.260908281389136, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.701372146606445, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.9036633968353271, + "num_tokens": 690213154.0, + "step": 17773 + }, + { + "epoch": 2.2610354916677267, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.666357040405273, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.9016863107681274, + "num_tokens": 690250166.0, + "step": 17774 + }, + { + "epoch": 2.261162701946317, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.64867877960205, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9062560200691223, + "num_tokens": 690286578.0, + "step": 17775 + }, + { + "epoch": 2.2612899122249077, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.6367769241333, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9081559777259827, + "num_tokens": 690324917.0, + "step": 17776 + }, + { + "epoch": 2.2614171225034982, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.774328231811523, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8929343223571777, + "num_tokens": 690359950.0, + "step": 17777 + }, + { + "epoch": 2.2615443327820888, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.698811531066895, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9070178866386414, + "num_tokens": 690396099.0, + "step": 17778 + }, + { + "epoch": 2.2616715430606793, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.759743690490723, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.904949426651001, + "num_tokens": 690428576.0, + "step": 17779 + }, + { + "epoch": 2.26179875333927, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.621485710144043, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.9098491668701172, + "num_tokens": 690469093.0, + "step": 17780 + }, + { + "epoch": 2.2619259636178604, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.616355895996094, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9013311266899109, + "num_tokens": 690506379.0, + "step": 17781 + }, + { + "epoch": 2.262053173896451, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.777831077575684, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8943923115730286, + "num_tokens": 690548505.0, + "step": 17782 + }, + { + "epoch": 2.2621803841750414, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.7244291305542, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9097377061843872, + "num_tokens": 690578995.0, + "step": 17783 + }, + { + "epoch": 2.262307594453632, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.720425605773926, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.899895429611206, + "num_tokens": 690614122.0, + "step": 17784 + }, + { + "epoch": 2.2624348047322225, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.502963066101074, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9162681102752686, + "num_tokens": 690660159.0, + "step": 17785 + }, + { + "epoch": 2.262562015010813, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.670963287353516, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9058746695518494, + "num_tokens": 690697737.0, + "step": 17786 + }, + { + "epoch": 2.2626892252894035, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.793973922729492, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9069687128067017, + "num_tokens": 690734229.0, + "step": 17787 + }, + { + "epoch": 2.262816435567994, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.646265029907227, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8886178135871887, + "num_tokens": 690777905.0, + "step": 17788 + }, + { + "epoch": 2.2629436458465846, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.66938304901123, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.898328423500061, + "num_tokens": 690820004.0, + "step": 17789 + }, + { + "epoch": 2.263070856125175, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.676133155822754, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8977923393249512, + "num_tokens": 690860834.0, + "step": 17790 + }, + { + "epoch": 2.263198066403765, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.693371772766113, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.907081663608551, + "num_tokens": 690893211.0, + "step": 17791 + }, + { + "epoch": 2.263325276682356, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.719661712646484, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8955676555633545, + "num_tokens": 690926809.0, + "step": 17792 + }, + { + "epoch": 2.2634524869609463, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.664956092834473, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9033684730529785, + "num_tokens": 690962279.0, + "step": 17793 + }, + { + "epoch": 2.263579697239537, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.631101608276367, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9062052965164185, + "num_tokens": 691000282.0, + "step": 17794 + }, + { + "epoch": 2.2637069075181273, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.657146453857422, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8881657123565674, + "num_tokens": 691045838.0, + "step": 17795 + }, + { + "epoch": 2.263834117796718, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.659367561340332, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8984180688858032, + "num_tokens": 691082141.0, + "step": 17796 + }, + { + "epoch": 2.2639613280753084, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.731535911560059, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.9006350040435791, + "num_tokens": 691114648.0, + "step": 17797 + }, + { + "epoch": 2.264088538353899, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.660775184631348, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8910707235336304, + "num_tokens": 691159701.0, + "step": 17798 + }, + { + "epoch": 2.2642157486324894, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.65518569946289, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.892135739326477, + "num_tokens": 691197643.0, + "step": 17799 + }, + { + "epoch": 2.26434295891108, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.707613945007324, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8847557306289673, + "num_tokens": 691238981.0, + "step": 17800 + }, + { + "epoch": 2.2644701691896705, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.610533714294434, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8843045234680176, + "num_tokens": 691284235.0, + "step": 17801 + }, + { + "epoch": 2.264597379468261, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.639899253845215, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.905953049659729, + "num_tokens": 691321722.0, + "step": 17802 + }, + { + "epoch": 2.2647245897468515, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.641453742980957, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8977035880088806, + "num_tokens": 691361334.0, + "step": 17803 + }, + { + "epoch": 2.264851800025442, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.607916831970215, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8998119831085205, + "num_tokens": 691405488.0, + "step": 17804 + }, + { + "epoch": 2.2649790103040326, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.620452880859375, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8906422853469849, + "num_tokens": 691448267.0, + "step": 17805 + }, + { + "epoch": 2.265106220582623, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.61941909790039, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9076154232025146, + "num_tokens": 691491032.0, + "step": 17806 + }, + { + "epoch": 2.2652334308612136, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.666863441467285, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8965072631835938, + "num_tokens": 691527246.0, + "step": 17807 + }, + { + "epoch": 2.265360641139804, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.660360336303711, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.9006780385971069, + "num_tokens": 691564634.0, + "step": 17808 + }, + { + "epoch": 2.2654878514183947, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.71157169342041, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8954275250434875, + "num_tokens": 691602059.0, + "step": 17809 + }, + { + "epoch": 2.2656150616969852, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.690089225769043, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8986951112747192, + "num_tokens": 691638338.0, + "step": 17810 + }, + { + "epoch": 2.2657422719755758, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.659150123596191, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9102700352668762, + "num_tokens": 691674860.0, + "step": 17811 + }, + { + "epoch": 2.2658694822541663, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.642387390136719, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.9007232189178467, + "num_tokens": 691713153.0, + "step": 17812 + }, + { + "epoch": 2.265996692532757, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.706644058227539, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8970549702644348, + "num_tokens": 691757726.0, + "step": 17813 + }, + { + "epoch": 2.2661239028113473, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.613268852233887, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9094196557998657, + "num_tokens": 691798134.0, + "step": 17814 + }, + { + "epoch": 2.266251113089938, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.73515510559082, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8999845385551453, + "num_tokens": 691835313.0, + "step": 17815 + }, + { + "epoch": 2.266378323368528, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.710952758789062, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9054515361785889, + "num_tokens": 691875670.0, + "step": 17816 + }, + { + "epoch": 2.266505533647119, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.699630737304688, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.9066824316978455, + "num_tokens": 691912061.0, + "step": 17817 + }, + { + "epoch": 2.266632743925709, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.754189491271973, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8962355852127075, + "num_tokens": 691946187.0, + "step": 17818 + }, + { + "epoch": 2.2667599542042995, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.138448715209961e-05, + "grad_norm": 9.662687301635742, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.899276852607727, + "num_tokens": 691988229.0, + "step": 17819 + }, + { + "epoch": 2.26688716448289, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.695281028747559, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.89640873670578, + "num_tokens": 692030731.0, + "step": 17820 + }, + { + "epoch": 2.2670143747614806, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.73447036743164, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.9022805690765381, + "num_tokens": 692068784.0, + "step": 17821 + }, + { + "epoch": 2.267141585040071, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.674775123596191, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.9044256210327148, + "num_tokens": 692101987.0, + "step": 17822 + }, + { + "epoch": 2.2672687953186617, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.751466751098633, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8945450782775879, + "num_tokens": 692138365.0, + "step": 17823 + }, + { + "epoch": 2.267396005597252, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.51875114440918, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9141994118690491, + "num_tokens": 692180609.0, + "step": 17824 + }, + { + "epoch": 2.2675232158758427, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.642815589904785, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.910004198551178, + "num_tokens": 692219690.0, + "step": 17825 + }, + { + "epoch": 2.2676504261544332, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.646903038024902, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9074075222015381, + "num_tokens": 692255821.0, + "step": 17826 + }, + { + "epoch": 2.2677776364330238, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.820197105407715, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.892108142375946, + "num_tokens": 692290549.0, + "step": 17827 + }, + { + "epoch": 2.2679048467116143, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.621238708496094, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9082977175712585, + "num_tokens": 692332659.0, + "step": 17828 + }, + { + "epoch": 2.268032056990205, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.639398574829102, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8950470685958862, + "num_tokens": 692371811.0, + "step": 17829 + }, + { + "epoch": 2.2681592672687954, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.614351272583008, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9092932939529419, + "num_tokens": 692412049.0, + "step": 17830 + }, + { + "epoch": 2.268286477547386, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.675864219665527, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8913803696632385, + "num_tokens": 692457090.0, + "step": 17831 + }, + { + "epoch": 2.2684136878259764, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.696208000183105, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8832036256790161, + "num_tokens": 692503428.0, + "step": 17832 + }, + { + "epoch": 2.268540898104567, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.59481143951416, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9131349921226501, + "num_tokens": 692545950.0, + "step": 17833 + }, + { + "epoch": 2.2686681083831575, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.601472854614258, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.9051198363304138, + "num_tokens": 692586486.0, + "step": 17834 + }, + { + "epoch": 2.268795318661748, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.724380493164062, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8993418216705322, + "num_tokens": 692623789.0, + "step": 17835 + }, + { + "epoch": 2.2689225289403385, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.663680076599121, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8964835405349731, + "num_tokens": 692667142.0, + "step": 17836 + }, + { + "epoch": 2.269049739218929, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.721819877624512, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.906485378742218, + "num_tokens": 692705462.0, + "step": 17837 + }, + { + "epoch": 2.2691769494975196, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.79443359375, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9113560318946838, + "num_tokens": 692738936.0, + "step": 17838 + }, + { + "epoch": 2.2693041597761097, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1444091796875e-05, + "grad_norm": 9.612394332885742, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.9019080400466919, + "num_tokens": 692783629.0, + "step": 17839 + }, + { + "epoch": 2.2694313700547006, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.776671409606934, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8983175158500671, + "num_tokens": 692823189.0, + "step": 17840 + }, + { + "epoch": 2.2695585803332907, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.648856163024902, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8975263237953186, + "num_tokens": 692865860.0, + "step": 17841 + }, + { + "epoch": 2.2696857906118812, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.759357452392578, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8940514922142029, + "num_tokens": 692907661.0, + "step": 17842 + }, + { + "epoch": 2.2698130008904718, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.603740692138672, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9112527370452881, + "num_tokens": 692947571.0, + "step": 17843 + }, + { + "epoch": 2.2699402111690623, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.73289966583252, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.900958776473999, + "num_tokens": 692982708.0, + "step": 17844 + }, + { + "epoch": 2.270067421447653, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.849594116210938, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8848444223403931, + "num_tokens": 693019844.0, + "step": 17845 + }, + { + "epoch": 2.2701946317262434, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.672082901000977, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8885093331336975, + "num_tokens": 693060464.0, + "step": 17846 + }, + { + "epoch": 2.270321842004834, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.773632049560547, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.9009439945220947, + "num_tokens": 693098094.0, + "step": 17847 + }, + { + "epoch": 2.2704490522834244, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.659764289855957, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9061839580535889, + "num_tokens": 693135640.0, + "step": 17848 + }, + { + "epoch": 2.270576262562015, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.662732124328613, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8896352648735046, + "num_tokens": 693178936.0, + "step": 17849 + }, + { + "epoch": 2.2707034728406055, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.696660995483398, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9082309007644653, + "num_tokens": 693214000.0, + "step": 17850 + }, + { + "epoch": 2.270830683119196, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.740096092224121, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.895874559879303, + "num_tokens": 693251371.0, + "step": 17851 + }, + { + "epoch": 2.2709578933977865, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.597926139831543, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9088108539581299, + "num_tokens": 693291874.0, + "step": 17852 + }, + { + "epoch": 2.271085103676377, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.721786499023438, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.900163471698761, + "num_tokens": 693327323.0, + "step": 17853 + }, + { + "epoch": 2.2712123139549676, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.796269416809082, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.9014891386032104, + "num_tokens": 693360061.0, + "step": 17854 + }, + { + "epoch": 2.271339524233558, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.762654304504395, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8970683813095093, + "num_tokens": 693392610.0, + "step": 17855 + }, + { + "epoch": 2.2714667345121486, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.834580421447754, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8996134400367737, + "num_tokens": 693433795.0, + "step": 17856 + }, + { + "epoch": 2.271593944790739, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.711969375610352, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.910843014717102, + "num_tokens": 693464978.0, + "step": 17857 + }, + { + "epoch": 2.2717211550693297, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.629854202270508, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8880652189254761, + "num_tokens": 693508069.0, + "step": 17858 + }, + { + "epoch": 2.2718483653479202, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.688042640686035, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8944360017776489, + "num_tokens": 693549818.0, + "step": 17859 + }, + { + "epoch": 2.2719755756265108, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.658909797668457, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8935797810554504, + "num_tokens": 693592039.0, + "step": 17860 + }, + { + "epoch": 2.2721027859051013, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.66675090789795, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8955003619194031, + "num_tokens": 693635755.0, + "step": 17861 + }, + { + "epoch": 2.272229996183692, + "ewc_loss": 0.01953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.70887565612793, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8999412059783936, + "num_tokens": 693669619.0, + "step": 17862 + }, + { + "epoch": 2.2723572064622823, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.660534858703613, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8917879462242126, + "num_tokens": 693710679.0, + "step": 17863 + }, + { + "epoch": 2.2724844167408724, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.617769241333008, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.9001806378364563, + "num_tokens": 693749207.0, + "step": 17864 + }, + { + "epoch": 2.2726116270194634, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.651927947998047, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8878505229949951, + "num_tokens": 693794671.0, + "step": 17865 + }, + { + "epoch": 2.2727388372980535, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.611442565917969, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.900662899017334, + "num_tokens": 693837784.0, + "step": 17866 + }, + { + "epoch": 2.272866047576644, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.683113098144531, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8984571099281311, + "num_tokens": 693876110.0, + "step": 17867 + }, + { + "epoch": 2.2729932578552345, + "ewc_loss": 0.0196533203125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.684903144836426, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8853835463523865, + "num_tokens": 693913511.0, + "step": 17868 + }, + { + "epoch": 2.273120468133825, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.669037818908691, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9075419902801514, + "num_tokens": 693949303.0, + "step": 17869 + }, + { + "epoch": 2.2732476784124156, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.58838939666748, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.898707926273346, + "num_tokens": 693992013.0, + "step": 17870 + }, + { + "epoch": 2.273374888691006, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.789441108703613, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8837940692901611, + "num_tokens": 694028570.0, + "step": 17871 + }, + { + "epoch": 2.2735020989695967, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.66072940826416, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8969109058380127, + "num_tokens": 694067428.0, + "step": 17872 + }, + { + "epoch": 2.273629309248187, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.745122909545898, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8958539962768555, + "num_tokens": 694104026.0, + "step": 17873 + }, + { + "epoch": 2.2737565195267777, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.677263259887695, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9150856733322144, + "num_tokens": 694139853.0, + "step": 17874 + }, + { + "epoch": 2.2738837298053682, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.584458351135254, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.9014914035797119, + "num_tokens": 694180959.0, + "step": 17875 + }, + { + "epoch": 2.2740109400839588, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.64570140838623, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.9028287529945374, + "num_tokens": 694218747.0, + "step": 17876 + }, + { + "epoch": 2.2741381503625493, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.72969913482666, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9013315439224243, + "num_tokens": 694253851.0, + "step": 17877 + }, + { + "epoch": 2.27426536064114, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.636116981506348, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8970056176185608, + "num_tokens": 694292249.0, + "step": 17878 + }, + { + "epoch": 2.2743925709197303, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.685111045837402, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9076330065727234, + "num_tokens": 694326878.0, + "step": 17879 + }, + { + "epoch": 2.274519781198321, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.711400985717773, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8990972638130188, + "num_tokens": 694364421.0, + "step": 17880 + }, + { + "epoch": 2.2746469914769114, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.713007926940918, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9083155989646912, + "num_tokens": 694397845.0, + "step": 17881 + }, + { + "epoch": 2.274774201755502, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.605232238769531, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.9005383253097534, + "num_tokens": 694436599.0, + "step": 17882 + }, + { + "epoch": 2.2749014120340925, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.677022933959961, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9073652029037476, + "num_tokens": 694471301.0, + "step": 17883 + }, + { + "epoch": 2.275028622312683, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.646524429321289, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.9000188112258911, + "num_tokens": 694510600.0, + "step": 17884 + }, + { + "epoch": 2.2751558325912735, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.728704452514648, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8980703353881836, + "num_tokens": 694542728.0, + "step": 17885 + }, + { + "epoch": 2.275283042869864, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.632417678833008, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.902144193649292, + "num_tokens": 694583806.0, + "step": 17886 + }, + { + "epoch": 2.2754102531484546, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.654748916625977, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.904032826423645, + "num_tokens": 694627101.0, + "step": 17887 + }, + { + "epoch": 2.275537463427045, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.698667526245117, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8847118616104126, + "num_tokens": 694671097.0, + "step": 17888 + }, + { + "epoch": 2.275664673705635, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.713125228881836, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8958355188369751, + "num_tokens": 694711567.0, + "step": 17889 + }, + { + "epoch": 2.275791883984226, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.69190502166748, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8965996503829956, + "num_tokens": 694749720.0, + "step": 17890 + }, + { + "epoch": 2.2759190942628162, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.68699836730957, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8948498964309692, + "num_tokens": 694786924.0, + "step": 17891 + }, + { + "epoch": 2.2760463045414068, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.672257423400879, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.9056602716445923, + "num_tokens": 694830354.0, + "step": 17892 + }, + { + "epoch": 2.2761735148199973, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.696907997131348, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9017547965049744, + "num_tokens": 694863150.0, + "step": 17893 + }, + { + "epoch": 2.276300725098588, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.058547973632812e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.634220123291016, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.9047250747680664, + "num_tokens": 694904346.0, + "step": 17894 + }, + { + "epoch": 2.2764279353771784, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.616804122924805, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8959782123565674, + "num_tokens": 694950074.0, + "step": 17895 + }, + { + "epoch": 2.276555145655769, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.674549102783203, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.9046849012374878, + "num_tokens": 694989142.0, + "step": 17896 + }, + { + "epoch": 2.2766823559343594, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.732943534851074, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.9032831192016602, + "num_tokens": 695023919.0, + "step": 17897 + }, + { + "epoch": 2.27680956621295, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.675198554992676, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9098963141441345, + "num_tokens": 695062427.0, + "step": 17898 + }, + { + "epoch": 2.2769367764915405, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.787039756774902, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8981624841690063, + "num_tokens": 695098605.0, + "step": 17899 + }, + { + "epoch": 2.277063986770131, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.694082260131836, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.901118278503418, + "num_tokens": 695139757.0, + "step": 17900 + }, + { + "epoch": 2.2771911970487215, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.686280250549316, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8954163789749146, + "num_tokens": 695178289.0, + "step": 17901 + }, + { + "epoch": 2.277318407327312, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.654048919677734, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.906146764755249, + "num_tokens": 695217333.0, + "step": 17902 + }, + { + "epoch": 2.2774456176059026, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.708894729614258, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8899981379508972, + "num_tokens": 695255149.0, + "step": 17903 + }, + { + "epoch": 2.277572827884493, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.657011032104492, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8859959244728088, + "num_tokens": 695298487.0, + "step": 17904 + }, + { + "epoch": 2.2777000381630836, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.725221633911133, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.885850727558136, + "num_tokens": 695338861.0, + "step": 17905 + }, + { + "epoch": 2.277827248441674, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.700864791870117, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8998039364814758, + "num_tokens": 695375702.0, + "step": 17906 + }, + { + "epoch": 2.2779544587202647, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.732572555541992, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.9057757258415222, + "num_tokens": 695407681.0, + "step": 17907 + }, + { + "epoch": 2.2780816689988552, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.68647289276123, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8946675062179565, + "num_tokens": 695448030.0, + "step": 17908 + }, + { + "epoch": 2.2782088792774458, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.684439659118652, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8912364840507507, + "num_tokens": 695487574.0, + "step": 17909 + }, + { + "epoch": 2.2783360895560363, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.649216651916504, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8958604335784912, + "num_tokens": 695529912.0, + "step": 17910 + }, + { + "epoch": 2.278463299834627, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.62631607055664, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8997898101806641, + "num_tokens": 695569446.0, + "step": 17911 + }, + { + "epoch": 2.2785905101132173, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.665606498718262, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8992331027984619, + "num_tokens": 695608878.0, + "step": 17912 + }, + { + "epoch": 2.278717720391808, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.776102066040039, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8925652503967285, + "num_tokens": 695643605.0, + "step": 17913 + }, + { + "epoch": 2.278844930670398, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.637171745300293, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.895404577255249, + "num_tokens": 695686733.0, + "step": 17914 + }, + { + "epoch": 2.278972140948989, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.66331672668457, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8936651945114136, + "num_tokens": 695728123.0, + "step": 17915 + }, + { + "epoch": 2.279099351227579, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.668418884277344, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8943189382553101, + "num_tokens": 695768688.0, + "step": 17916 + }, + { + "epoch": 2.2792265615061695, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.625438690185547, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.9073010087013245, + "num_tokens": 695807274.0, + "step": 17917 + }, + { + "epoch": 2.27935377178476, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.797735214233398, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8894413709640503, + "num_tokens": 695840775.0, + "step": 17918 + }, + { + "epoch": 2.2794809820633506, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.821121215820312, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.885494589805603, + "num_tokens": 695879462.0, + "step": 17919 + }, + { + "epoch": 2.279608192341941, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.699563980102539, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8942852020263672, + "num_tokens": 695919807.0, + "step": 17920 + }, + { + "epoch": 2.2797354026205316, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.600722312927246, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.9023107886314392, + "num_tokens": 695963953.0, + "step": 17921 + }, + { + "epoch": 2.279862612899122, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.839935302734375, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8972381949424744, + "num_tokens": 695995065.0, + "step": 17922 + }, + { + "epoch": 2.2799898231777127, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.740606307983398, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8952310085296631, + "num_tokens": 696029113.0, + "step": 17923 + }, + { + "epoch": 2.2801170334563032, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.640766143798828, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8873545527458191, + "num_tokens": 696075272.0, + "step": 17924 + }, + { + "epoch": 2.2802442437348938, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.845002174377441, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8968863487243652, + "num_tokens": 696107302.0, + "step": 17925 + }, + { + "epoch": 2.2803714540134843, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.642082214355469, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8998544812202454, + "num_tokens": 696145117.0, + "step": 17926 + }, + { + "epoch": 2.280498664292075, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.627679824829102, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.9064954519271851, + "num_tokens": 696184149.0, + "step": 17927 + }, + { + "epoch": 2.2806258745706653, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.786942481994629, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.9033111333847046, + "num_tokens": 696216222.0, + "step": 17928 + }, + { + "epoch": 2.280753084849256, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.697816848754883, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.9054368734359741, + "num_tokens": 696254898.0, + "step": 17929 + }, + { + "epoch": 2.2808802951278464, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.758131980895996, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8775734901428223, + "num_tokens": 696293591.0, + "step": 17930 + }, + { + "epoch": 2.281007505406437, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.605615615844727, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8967361450195312, + "num_tokens": 696337325.0, + "step": 17931 + }, + { + "epoch": 2.2811347156850275, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.80850887298584, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8864421844482422, + "num_tokens": 696371650.0, + "step": 17932 + }, + { + "epoch": 2.281261925963618, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.60116958618164, + "learning_rate": 1e-06, + "loss": 0.2116, + "mean_token_accuracy": 0.9281009435653687, + "num_tokens": 696409063.0, + "step": 17933 + }, + { + "epoch": 2.2813891362422085, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.622612953186035, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.900151789188385, + "num_tokens": 696449342.0, + "step": 17934 + }, + { + "epoch": 2.281516346520799, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.761493682861328, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9052649736404419, + "num_tokens": 696486413.0, + "step": 17935 + }, + { + "epoch": 2.2816435567993896, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.732250213623047, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8887298703193665, + "num_tokens": 696525517.0, + "step": 17936 + }, + { + "epoch": 2.2817707670779797, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.798828125, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8918805122375488, + "num_tokens": 696558552.0, + "step": 17937 + }, + { + "epoch": 2.2818979773565706, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.654468536376953, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8880236148834229, + "num_tokens": 696599994.0, + "step": 17938 + }, + { + "epoch": 2.2820251876351607, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.66965389251709, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8998327255249023, + "num_tokens": 696638673.0, + "step": 17939 + }, + { + "epoch": 2.2821523979137512, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.715309143066406, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8827844858169556, + "num_tokens": 696678919.0, + "step": 17940 + }, + { + "epoch": 2.2822796081923418, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.728686332702637, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8855487108230591, + "num_tokens": 696715995.0, + "step": 17941 + }, + { + "epoch": 2.2824068184709323, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.753958702087402, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.897787868976593, + "num_tokens": 696756168.0, + "step": 17942 + }, + { + "epoch": 2.282534028749523, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.609176635742188, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8999828100204468, + "num_tokens": 696795701.0, + "step": 17943 + }, + { + "epoch": 2.2826612390281134, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.772738456726074, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9078513979911804, + "num_tokens": 696826819.0, + "step": 17944 + }, + { + "epoch": 2.282788449306704, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.620865821838379, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.9024295210838318, + "num_tokens": 696869618.0, + "step": 17945 + }, + { + "epoch": 2.2829156595852944, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.695011138916016, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8962397575378418, + "num_tokens": 696907545.0, + "step": 17946 + }, + { + "epoch": 2.283042869863885, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.684452056884766, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9071782827377319, + "num_tokens": 696946646.0, + "step": 17947 + }, + { + "epoch": 2.2831700801424755, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.671363830566406, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9109423160552979, + "num_tokens": 696982828.0, + "step": 17948 + }, + { + "epoch": 2.283297290421066, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.793679237365723, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9041961431503296, + "num_tokens": 697019650.0, + "step": 17949 + }, + { + "epoch": 2.2834245006996565, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.72415542602539, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8993192911148071, + "num_tokens": 697061526.0, + "step": 17950 + }, + { + "epoch": 2.283551710978247, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.762375831604004, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8891232013702393, + "num_tokens": 697100287.0, + "step": 17951 + }, + { + "epoch": 2.2836789212568376, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.731297492980957, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.894101083278656, + "num_tokens": 697134748.0, + "step": 17952 + }, + { + "epoch": 2.283806131535428, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.718199729919434, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.89234858751297, + "num_tokens": 697175374.0, + "step": 17953 + }, + { + "epoch": 2.2839333418140186, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.742643356323242, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8927747011184692, + "num_tokens": 697214888.0, + "step": 17954 + }, + { + "epoch": 2.284060552092609, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.615676879882812, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9019373655319214, + "num_tokens": 697259260.0, + "step": 17955 + }, + { + "epoch": 2.2841877623711997, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.724886894226074, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9019100666046143, + "num_tokens": 697298139.0, + "step": 17956 + }, + { + "epoch": 2.28431497264979, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.70905590057373, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.9019154906272888, + "num_tokens": 697337474.0, + "step": 17957 + }, + { + "epoch": 2.2844421829283807, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.669576644897461, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.9001039266586304, + "num_tokens": 697372444.0, + "step": 17958 + }, + { + "epoch": 2.2845693932069713, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.759879112243652, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9029899835586548, + "num_tokens": 697410634.0, + "step": 17959 + }, + { + "epoch": 2.284696603485562, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.676093101501465, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8930085897445679, + "num_tokens": 697448039.0, + "step": 17960 + }, + { + "epoch": 2.2848238137641523, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.723921775817871, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9019483327865601, + "num_tokens": 697481714.0, + "step": 17961 + }, + { + "epoch": 2.2849510240427424, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.680230140686035, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8956261873245239, + "num_tokens": 697518432.0, + "step": 17962 + }, + { + "epoch": 2.2850782343213334, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.709310531616211, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.893303632736206, + "num_tokens": 697558956.0, + "step": 17963 + }, + { + "epoch": 2.2852054445999235, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.598329544067383, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9045388102531433, + "num_tokens": 697598496.0, + "step": 17964 + }, + { + "epoch": 2.285332654878514, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.659567832946777, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.9037696719169617, + "num_tokens": 697636176.0, + "step": 17965 + }, + { + "epoch": 2.2854598651571045, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.711494445800781, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8957924842834473, + "num_tokens": 697674773.0, + "step": 17966 + }, + { + "epoch": 2.285587075435695, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.678266525268555, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.901617705821991, + "num_tokens": 697713854.0, + "step": 17967 + }, + { + "epoch": 2.2857142857142856, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.68702220916748, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8993033170700073, + "num_tokens": 697750112.0, + "step": 17968 + }, + { + "epoch": 2.285841495992876, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.715998649597168, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8876525163650513, + "num_tokens": 697788153.0, + "step": 17969 + }, + { + "epoch": 2.2859687062714666, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.736550331115723, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8818304538726807, + "num_tokens": 697827359.0, + "step": 17970 + }, + { + "epoch": 2.286095916550057, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.539824485778809, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9139045476913452, + "num_tokens": 697865252.0, + "step": 17971 + }, + { + "epoch": 2.2862231268286477, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.754775047302246, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8984725475311279, + "num_tokens": 697898990.0, + "step": 17972 + }, + { + "epoch": 2.2863503371072382, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.635149002075195, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8963940739631653, + "num_tokens": 697940578.0, + "step": 17973 + }, + { + "epoch": 2.2864775473858288, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.650517463684082, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.891901969909668, + "num_tokens": 697982913.0, + "step": 17974 + }, + { + "epoch": 2.2866047576644193, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.639015197753906, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8999320268630981, + "num_tokens": 698021684.0, + "step": 17975 + }, + { + "epoch": 2.28673196794301, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.674456596374512, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.9019851684570312, + "num_tokens": 698058689.0, + "step": 17976 + }, + { + "epoch": 2.2868591782216003, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.790274620056152, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8998540043830872, + "num_tokens": 698094447.0, + "step": 17977 + }, + { + "epoch": 2.286986388500191, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.698646545410156, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9091955423355103, + "num_tokens": 698126823.0, + "step": 17978 + }, + { + "epoch": 2.2871135987787814, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.717074394226074, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.9023494720458984, + "num_tokens": 698163968.0, + "step": 17979 + }, + { + "epoch": 2.287240809057372, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.801959037780762, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9012154936790466, + "num_tokens": 698203987.0, + "step": 17980 + }, + { + "epoch": 2.2873680193359625, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.78201961517334, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8904714584350586, + "num_tokens": 698239464.0, + "step": 17981 + }, + { + "epoch": 2.287495229614553, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.698517799377441, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8974153399467468, + "num_tokens": 698283604.0, + "step": 17982 + }, + { + "epoch": 2.2876224398931435, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.706801414489746, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8966420888900757, + "num_tokens": 698322403.0, + "step": 17983 + }, + { + "epoch": 2.287749650171734, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.78947925567627, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.9021291732788086, + "num_tokens": 698355695.0, + "step": 17984 + }, + { + "epoch": 2.2878768604503246, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.654151916503906, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.9025176763534546, + "num_tokens": 698397269.0, + "step": 17985 + }, + { + "epoch": 2.288004070728915, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.721977233886719, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8998613953590393, + "num_tokens": 698433839.0, + "step": 17986 + }, + { + "epoch": 2.288131281007505, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.695110321044922, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8928433656692505, + "num_tokens": 698471652.0, + "step": 17987 + }, + { + "epoch": 2.288258491286096, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.799668312072754, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8883939385414124, + "num_tokens": 698508520.0, + "step": 17988 + }, + { + "epoch": 2.2883857015646862, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.599458694458008, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.9011411070823669, + "num_tokens": 698549818.0, + "step": 17989 + }, + { + "epoch": 2.2885129118432768, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.689596176147461, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8995420336723328, + "num_tokens": 698586151.0, + "step": 17990 + }, + { + "epoch": 2.2886401221218673, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.74498176574707, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9011745452880859, + "num_tokens": 698620643.0, + "step": 17991 + }, + { + "epoch": 2.288767332400458, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.699677467346191, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9070166945457458, + "num_tokens": 698655933.0, + "step": 17992 + }, + { + "epoch": 2.2888945426790483, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.712749481201172, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9078816175460815, + "num_tokens": 698694094.0, + "step": 17993 + }, + { + "epoch": 2.289021752957639, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.769948959350586, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8987581729888916, + "num_tokens": 698726445.0, + "step": 17994 + }, + { + "epoch": 2.2891489632362294, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.770404815673828, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.9033036231994629, + "num_tokens": 698756861.0, + "step": 17995 + }, + { + "epoch": 2.28927617351482, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.672741889953613, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.911573588848114, + "num_tokens": 698793060.0, + "step": 17996 + }, + { + "epoch": 2.2894033837934105, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.693174362182617, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.9001789093017578, + "num_tokens": 698832435.0, + "step": 17997 + }, + { + "epoch": 2.289530594072001, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.77760124206543, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8946940898895264, + "num_tokens": 698869334.0, + "step": 17998 + }, + { + "epoch": 2.2896578043505915, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.73685359954834, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8959842920303345, + "num_tokens": 698902793.0, + "step": 17999 + }, + { + "epoch": 2.289785014629182, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.67523193359375, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.9010224938392639, + "num_tokens": 698941521.0, + "step": 18000 + }, + { + "epoch": 2.2899122249077726, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.676887512207031, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.9014135599136353, + "num_tokens": 698975927.0, + "step": 18001 + }, + { + "epoch": 2.290039435186363, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.686129570007324, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8940324783325195, + "num_tokens": 699017573.0, + "step": 18002 + }, + { + "epoch": 2.2901666454649536, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.645962715148926, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8948197364807129, + "num_tokens": 699059825.0, + "step": 18003 + }, + { + "epoch": 2.290293855743544, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.6480712890625, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9048730134963989, + "num_tokens": 699097781.0, + "step": 18004 + }, + { + "epoch": 2.2904210660221347, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.699081420898438, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.9006407260894775, + "num_tokens": 699133444.0, + "step": 18005 + }, + { + "epoch": 2.290548276300725, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.83534049987793, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8868298530578613, + "num_tokens": 699171354.0, + "step": 18006 + }, + { + "epoch": 2.2906754865793157, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.631132125854492, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8910195827484131, + "num_tokens": 699215508.0, + "step": 18007 + }, + { + "epoch": 2.2908026968579063, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.736337661743164, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8946314454078674, + "num_tokens": 699255840.0, + "step": 18008 + }, + { + "epoch": 2.290929907136497, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.611044883728027, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.9030078053474426, + "num_tokens": 699299897.0, + "step": 18009 + }, + { + "epoch": 2.2910571174150873, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.73332691192627, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8927197456359863, + "num_tokens": 699338408.0, + "step": 18010 + }, + { + "epoch": 2.291184327693678, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.579652786254883, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.9025563597679138, + "num_tokens": 699382339.0, + "step": 18011 + }, + { + "epoch": 2.291311537972268, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.711004257202148, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8974124193191528, + "num_tokens": 699421832.0, + "step": 18012 + }, + { + "epoch": 2.291438748250859, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.63632583618164, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9034430384635925, + "num_tokens": 699463567.0, + "step": 18013 + }, + { + "epoch": 2.291565958529449, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.657344818115234, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8946033120155334, + "num_tokens": 699509077.0, + "step": 18014 + }, + { + "epoch": 2.2916931688080395, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.712498664855957, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8879529237747192, + "num_tokens": 699548728.0, + "step": 18015 + }, + { + "epoch": 2.29182037908663, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.759658813476562, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.902747392654419, + "num_tokens": 699582026.0, + "step": 18016 + }, + { + "epoch": 2.2919475893652206, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.659013748168945, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8994931578636169, + "num_tokens": 699619985.0, + "step": 18017 + }, + { + "epoch": 2.292074799643811, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.637470245361328, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9078063368797302, + "num_tokens": 699657492.0, + "step": 18018 + }, + { + "epoch": 2.2922020099224016, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.709501266479492, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8998486399650574, + "num_tokens": 699691451.0, + "step": 18019 + }, + { + "epoch": 2.292329220200992, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.730020523071289, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8944337368011475, + "num_tokens": 699731443.0, + "step": 18020 + }, + { + "epoch": 2.2924564304795827, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.671531677246094, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8987709283828735, + "num_tokens": 699771327.0, + "step": 18021 + }, + { + "epoch": 2.2925836407581732, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.809873580932617, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8905604481697083, + "num_tokens": 699806562.0, + "step": 18022 + }, + { + "epoch": 2.2927108510367638, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.84322738647461, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.883232831954956, + "num_tokens": 699839500.0, + "step": 18023 + }, + { + "epoch": 2.2928380613153543, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.669013977050781, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8996890187263489, + "num_tokens": 699880651.0, + "step": 18024 + }, + { + "epoch": 2.292965271593945, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.66135025024414, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8952246904373169, + "num_tokens": 699927437.0, + "step": 18025 + }, + { + "epoch": 2.2930924818725353, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.681949615478516, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8983694314956665, + "num_tokens": 699970422.0, + "step": 18026 + }, + { + "epoch": 2.293219692151126, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.106231689453125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.78099536895752, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8883256316184998, + "num_tokens": 700009318.0, + "step": 18027 + }, + { + "epoch": 2.2933469024297164, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.73851490020752, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.9047099351882935, + "num_tokens": 700048339.0, + "step": 18028 + }, + { + "epoch": 2.293474112708307, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.760900497436523, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8894453048706055, + "num_tokens": 700085336.0, + "step": 18029 + }, + { + "epoch": 2.2936013229868975, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.671142578125, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.903019368648529, + "num_tokens": 700125218.0, + "step": 18030 + }, + { + "epoch": 2.293728533265488, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.756902694702148, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8998245596885681, + "num_tokens": 700164390.0, + "step": 18031 + }, + { + "epoch": 2.2938557435440785, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.776514053344727, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8940385580062866, + "num_tokens": 700203563.0, + "step": 18032 + }, + { + "epoch": 2.293982953822669, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.67943286895752, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9052015542984009, + "num_tokens": 700243416.0, + "step": 18033 + }, + { + "epoch": 2.2941101641012596, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.69405460357666, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8975976705551147, + "num_tokens": 700283994.0, + "step": 18034 + }, + { + "epoch": 2.2942373743798496, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.675888061523438, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.895409882068634, + "num_tokens": 700324851.0, + "step": 18035 + }, + { + "epoch": 2.2943645846584406, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.67790412902832, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.9019775390625, + "num_tokens": 700365465.0, + "step": 18036 + }, + { + "epoch": 2.2944917949370307, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.765727996826172, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.89760422706604, + "num_tokens": 700404277.0, + "step": 18037 + }, + { + "epoch": 2.2946190052156212, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.785550117492676, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8785523772239685, + "num_tokens": 700445868.0, + "step": 18038 + }, + { + "epoch": 2.2947462154942118, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.73392105102539, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.899103045463562, + "num_tokens": 700482291.0, + "step": 18039 + }, + { + "epoch": 2.2948734257728023, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.676828384399414, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.902298092842102, + "num_tokens": 700519120.0, + "step": 18040 + }, + { + "epoch": 2.295000636051393, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.723825454711914, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.9005736112594604, + "num_tokens": 700560299.0, + "step": 18041 + }, + { + "epoch": 2.2951278463299833, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.857559204101562, + "learning_rate": 1e-06, + "loss": 0.3734, + "mean_token_accuracy": 0.8770638704299927, + "num_tokens": 700596973.0, + "step": 18042 + }, + { + "epoch": 2.295255056608574, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.659004211425781, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8975796699523926, + "num_tokens": 700638328.0, + "step": 18043 + }, + { + "epoch": 2.2953822668871644, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.685665130615234, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8876262903213501, + "num_tokens": 700680103.0, + "step": 18044 + }, + { + "epoch": 2.295509477165755, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.733932495117188, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8930144309997559, + "num_tokens": 700720083.0, + "step": 18045 + }, + { + "epoch": 2.2956366874443455, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.602823257446289, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8902140855789185, + "num_tokens": 700765613.0, + "step": 18046 + }, + { + "epoch": 2.295763897722936, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.653575897216797, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9060095548629761, + "num_tokens": 700803035.0, + "step": 18047 + }, + { + "epoch": 2.2958911080015265, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.663137435913086, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9007911682128906, + "num_tokens": 700842764.0, + "step": 18048 + }, + { + "epoch": 2.296018318280117, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.688691139221191, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9012155532836914, + "num_tokens": 700882498.0, + "step": 18049 + }, + { + "epoch": 2.2961455285587076, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.684379577636719, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8944753408432007, + "num_tokens": 700925410.0, + "step": 18050 + }, + { + "epoch": 2.296272738837298, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.699646949768066, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.9011636972427368, + "num_tokens": 700964132.0, + "step": 18051 + }, + { + "epoch": 2.2963999491158886, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.727598190307617, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9038643836975098, + "num_tokens": 700997762.0, + "step": 18052 + }, + { + "epoch": 2.296527159394479, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.737141609191895, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8920865058898926, + "num_tokens": 701034843.0, + "step": 18053 + }, + { + "epoch": 2.2966543696730697, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.819608688354492, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8990180492401123, + "num_tokens": 701074293.0, + "step": 18054 + }, + { + "epoch": 2.29678157995166, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.725201606750488, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.9055781364440918, + "num_tokens": 701109519.0, + "step": 18055 + }, + { + "epoch": 2.2969087902302507, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.751322746276855, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8965902924537659, + "num_tokens": 701143276.0, + "step": 18056 + }, + { + "epoch": 2.2970360005088413, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.760246276855469, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8922954797744751, + "num_tokens": 701177877.0, + "step": 18057 + }, + { + "epoch": 2.297163210787432, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.73110294342041, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8827872276306152, + "num_tokens": 701218000.0, + "step": 18058 + }, + { + "epoch": 2.2972904210660223, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.716324806213379, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8857974410057068, + "num_tokens": 701256598.0, + "step": 18059 + }, + { + "epoch": 2.2974176313446124, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.664542198181152, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.904466986656189, + "num_tokens": 701296209.0, + "step": 18060 + }, + { + "epoch": 2.2975448416232034, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.9415864944458, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8826454281806946, + "num_tokens": 701328350.0, + "step": 18061 + }, + { + "epoch": 2.2976720519017935, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.627920150756836, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9070544838905334, + "num_tokens": 701368276.0, + "step": 18062 + }, + { + "epoch": 2.297799262180384, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.153915405273438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.816248893737793, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8935414552688599, + "num_tokens": 701406011.0, + "step": 18063 + }, + { + "epoch": 2.2979264724589745, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.625239372253418, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.901835024356842, + "num_tokens": 701444664.0, + "step": 18064 + }, + { + "epoch": 2.298053682737565, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.691107749938965, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.912861168384552, + "num_tokens": 701477017.0, + "step": 18065 + }, + { + "epoch": 2.2981808930161556, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.688387870788574, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.9057549834251404, + "num_tokens": 701511715.0, + "step": 18066 + }, + { + "epoch": 2.298308103294746, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.687737464904785, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8899526596069336, + "num_tokens": 701553293.0, + "step": 18067 + }, + { + "epoch": 2.2984353135733366, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.756674766540527, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8894770741462708, + "num_tokens": 701589528.0, + "step": 18068 + }, + { + "epoch": 2.298562523851927, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.735188484191895, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.9011939764022827, + "num_tokens": 701624840.0, + "step": 18069 + }, + { + "epoch": 2.2986897341305177, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.657587051391602, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8948047161102295, + "num_tokens": 701666905.0, + "step": 18070 + }, + { + "epoch": 2.298816944409108, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.720824241638184, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8786798119544983, + "num_tokens": 701704943.0, + "step": 18071 + }, + { + "epoch": 2.2989441546876987, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.664918899536133, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8996757864952087, + "num_tokens": 701743863.0, + "step": 18072 + }, + { + "epoch": 2.2990713649662893, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.614925384521484, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.908200740814209, + "num_tokens": 701779157.0, + "step": 18073 + }, + { + "epoch": 2.29919857524488, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.632919311523438, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9088461399078369, + "num_tokens": 701819799.0, + "step": 18074 + }, + { + "epoch": 2.2993257855234703, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.605159759521484, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8904592990875244, + "num_tokens": 701867451.0, + "step": 18075 + }, + { + "epoch": 2.299452995802061, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.698010444641113, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8948744535446167, + "num_tokens": 701909898.0, + "step": 18076 + }, + { + "epoch": 2.2995802060806514, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.664344787597656, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9054902791976929, + "num_tokens": 701951361.0, + "step": 18077 + }, + { + "epoch": 2.299707416359242, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.676094055175781, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9081354141235352, + "num_tokens": 701986499.0, + "step": 18078 + }, + { + "epoch": 2.2998346266378324, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.835289001464844, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8985173106193542, + "num_tokens": 702019676.0, + "step": 18079 + }, + { + "epoch": 2.299961836916423, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.658528327941895, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.9032601118087769, + "num_tokens": 702060159.0, + "step": 18080 + }, + { + "epoch": 2.3000890471950135, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.69786262512207, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8949504494667053, + "num_tokens": 702098928.0, + "step": 18081 + }, + { + "epoch": 2.300216257473604, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.704306602478027, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.9002175331115723, + "num_tokens": 702138594.0, + "step": 18082 + }, + { + "epoch": 2.3003434677521946, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.728865623474121, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9069602489471436, + "num_tokens": 702172662.0, + "step": 18083 + }, + { + "epoch": 2.300470678030785, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.682982444763184, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.9030722379684448, + "num_tokens": 702216035.0, + "step": 18084 + }, + { + "epoch": 2.300597888309375, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.716447830200195, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8924992084503174, + "num_tokens": 702261032.0, + "step": 18085 + }, + { + "epoch": 2.300725098587966, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.707326889038086, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9078421592712402, + "num_tokens": 702295313.0, + "step": 18086 + }, + { + "epoch": 2.3008523088665562, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.704471588134766, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8994725942611694, + "num_tokens": 702338993.0, + "step": 18087 + }, + { + "epoch": 2.3009795191451468, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.679100036621094, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8978396654129028, + "num_tokens": 702384332.0, + "step": 18088 + }, + { + "epoch": 2.3011067294237373, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.716197967529297, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9033877849578857, + "num_tokens": 702419784.0, + "step": 18089 + }, + { + "epoch": 2.301233939702328, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.629012107849121, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8992311358451843, + "num_tokens": 702466387.0, + "step": 18090 + }, + { + "epoch": 2.3013611499809183, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.795008659362793, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.899354100227356, + "num_tokens": 702501711.0, + "step": 18091 + }, + { + "epoch": 2.301488360259509, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.638906478881836, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.9002306461334229, + "num_tokens": 702544617.0, + "step": 18092 + }, + { + "epoch": 2.3016155705380994, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.703207015991211, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.9012399911880493, + "num_tokens": 702590330.0, + "step": 18093 + }, + { + "epoch": 2.30174278081669, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.773411750793457, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8965952396392822, + "num_tokens": 702631158.0, + "step": 18094 + }, + { + "epoch": 2.3018699910952805, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.744468688964844, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8952723741531372, + "num_tokens": 702667780.0, + "step": 18095 + }, + { + "epoch": 2.301997201373871, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.669048309326172, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8962292671203613, + "num_tokens": 702711644.0, + "step": 18096 + }, + { + "epoch": 2.3021244116524615, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.642424583435059, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9050959944725037, + "num_tokens": 702749284.0, + "step": 18097 + }, + { + "epoch": 2.302251621931052, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.773601531982422, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.9001216888427734, + "num_tokens": 702783848.0, + "step": 18098 + }, + { + "epoch": 2.3023788322096426, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.63196849822998, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9057614207267761, + "num_tokens": 702825772.0, + "step": 18099 + }, + { + "epoch": 2.302506042488233, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.766606330871582, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8973771929740906, + "num_tokens": 702864794.0, + "step": 18100 + }, + { + "epoch": 2.3026332527668236, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.81954574584961, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8785418272018433, + "num_tokens": 702900722.0, + "step": 18101 + }, + { + "epoch": 2.302760463045414, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.645736694335938, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.9047551155090332, + "num_tokens": 702941572.0, + "step": 18102 + }, + { + "epoch": 2.3028876733240047, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.655291557312012, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.9007513523101807, + "num_tokens": 702983946.0, + "step": 18103 + }, + { + "epoch": 2.303014883602595, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.838937759399414, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8954137563705444, + "num_tokens": 703016787.0, + "step": 18104 + }, + { + "epoch": 2.3031420938811857, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.698043823242188, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8934201002120972, + "num_tokens": 703055341.0, + "step": 18105 + }, + { + "epoch": 2.3032693041597763, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.719972610473633, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9083695411682129, + "num_tokens": 703089286.0, + "step": 18106 + }, + { + "epoch": 2.303396514438367, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.733393669128418, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8935543298721313, + "num_tokens": 703127634.0, + "step": 18107 + }, + { + "epoch": 2.3035237247169573, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.68071460723877, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8998430371284485, + "num_tokens": 703166136.0, + "step": 18108 + }, + { + "epoch": 2.303650934995548, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.720273971557617, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.9063296318054199, + "num_tokens": 703208370.0, + "step": 18109 + }, + { + "epoch": 2.303778145274138, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.766524314880371, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8987341523170471, + "num_tokens": 703243598.0, + "step": 18110 + }, + { + "epoch": 2.303905355552729, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.687067031860352, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.9096344709396362, + "num_tokens": 703280979.0, + "step": 18111 + }, + { + "epoch": 2.304032565831319, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.819558143615723, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8854823112487793, + "num_tokens": 703319377.0, + "step": 18112 + }, + { + "epoch": 2.3041597761099095, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.869598388671875, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8841879963874817, + "num_tokens": 703348957.0, + "step": 18113 + }, + { + "epoch": 2.3042869863885, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.774323463439941, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.9041016101837158, + "num_tokens": 703380854.0, + "step": 18114 + }, + { + "epoch": 2.3044141966670906, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.699030876159668, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8926617503166199, + "num_tokens": 703419848.0, + "step": 18115 + }, + { + "epoch": 2.304541406945681, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.629825592041016, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8990743160247803, + "num_tokens": 703459243.0, + "step": 18116 + }, + { + "epoch": 2.3046686172242716, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.674421310424805, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8993271589279175, + "num_tokens": 703502647.0, + "step": 18117 + }, + { + "epoch": 2.304795827502862, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.675933837890625, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.9012148380279541, + "num_tokens": 703544601.0, + "step": 18118 + }, + { + "epoch": 2.3049230377814527, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.64294147491455, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.9020824432373047, + "num_tokens": 703586973.0, + "step": 18119 + }, + { + "epoch": 2.305050248060043, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.770418167114258, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8920218348503113, + "num_tokens": 703627153.0, + "step": 18120 + }, + { + "epoch": 2.3051774583386337, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.763749122619629, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8910366892814636, + "num_tokens": 703665743.0, + "step": 18121 + }, + { + "epoch": 2.3053046686172243, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.74560832977295, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8958141803741455, + "num_tokens": 703706918.0, + "step": 18122 + }, + { + "epoch": 2.305431878895815, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.775108337402344, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8960616588592529, + "num_tokens": 703743097.0, + "step": 18123 + }, + { + "epoch": 2.3055590891744053, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.718587875366211, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8978290557861328, + "num_tokens": 703782724.0, + "step": 18124 + }, + { + "epoch": 2.305686299452996, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.722789764404297, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8956935405731201, + "num_tokens": 703819687.0, + "step": 18125 + }, + { + "epoch": 2.3058135097315864, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.751948356628418, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.9008802175521851, + "num_tokens": 703855990.0, + "step": 18126 + }, + { + "epoch": 2.305940720010177, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.694478034973145, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.9058968424797058, + "num_tokens": 703895415.0, + "step": 18127 + }, + { + "epoch": 2.3060679302887674, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.700339317321777, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8977068066596985, + "num_tokens": 703937952.0, + "step": 18128 + }, + { + "epoch": 2.306195140567358, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.611335754394531, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9077386260032654, + "num_tokens": 703977623.0, + "step": 18129 + }, + { + "epoch": 2.3063223508459485, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.824265480041504, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8981747627258301, + "num_tokens": 704012517.0, + "step": 18130 + }, + { + "epoch": 2.306449561124539, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.789807319641113, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8903993368148804, + "num_tokens": 704048610.0, + "step": 18131 + }, + { + "epoch": 2.3065767714031296, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.779118537902832, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8849085569381714, + "num_tokens": 704086818.0, + "step": 18132 + }, + { + "epoch": 2.3067039816817196, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.790032386779785, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8958480358123779, + "num_tokens": 704121837.0, + "step": 18133 + }, + { + "epoch": 2.3068311919603106, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.728002548217773, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8965669870376587, + "num_tokens": 704157724.0, + "step": 18134 + }, + { + "epoch": 2.3069584022389007, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.663228988647461, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.9022327065467834, + "num_tokens": 704203578.0, + "step": 18135 + }, + { + "epoch": 2.3070856125174912, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.789016723632812, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.9029145240783691, + "num_tokens": 704237715.0, + "step": 18136 + }, + { + "epoch": 2.3072128227960818, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.704097747802734, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.9011098146438599, + "num_tokens": 704275774.0, + "step": 18137 + }, + { + "epoch": 2.3073400330746723, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.761170387268066, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9050389528274536, + "num_tokens": 704311023.0, + "step": 18138 + }, + { + "epoch": 2.307467243353263, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.687764167785645, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.901925802230835, + "num_tokens": 704353092.0, + "step": 18139 + }, + { + "epoch": 2.3075944536318533, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.797687530517578, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8891869783401489, + "num_tokens": 704392879.0, + "step": 18140 + }, + { + "epoch": 2.307721663910444, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.539937019348145, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9151731133460999, + "num_tokens": 704437616.0, + "step": 18141 + }, + { + "epoch": 2.3078488741890344, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.766098022460938, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8921613693237305, + "num_tokens": 704474364.0, + "step": 18142 + }, + { + "epoch": 2.307976084467625, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.782621383666992, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.88560950756073, + "num_tokens": 704515828.0, + "step": 18143 + }, + { + "epoch": 2.3081032947462155, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.782586097717285, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8793126344680786, + "num_tokens": 704556238.0, + "step": 18144 + }, + { + "epoch": 2.308230505024806, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.863940238952637, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8844056129455566, + "num_tokens": 704587678.0, + "step": 18145 + }, + { + "epoch": 2.3083577153033965, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.758193969726562, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8972632884979248, + "num_tokens": 704620002.0, + "step": 18146 + }, + { + "epoch": 2.308484925581987, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.631512641906738, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8843805193901062, + "num_tokens": 704660899.0, + "step": 18147 + }, + { + "epoch": 2.3086121358605776, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.700439453125, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8961620330810547, + "num_tokens": 704703315.0, + "step": 18148 + }, + { + "epoch": 2.308739346139168, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.703224182128906, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8980261087417603, + "num_tokens": 704737835.0, + "step": 18149 + }, + { + "epoch": 2.3088665564177586, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.734460830688477, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8956043124198914, + "num_tokens": 704772644.0, + "step": 18150 + }, + { + "epoch": 2.308993766696349, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.676056861877441, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8939501643180847, + "num_tokens": 704820469.0, + "step": 18151 + }, + { + "epoch": 2.3091209769749397, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.758502006530762, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9039276838302612, + "num_tokens": 704852680.0, + "step": 18152 + }, + { + "epoch": 2.30924818725353, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.712599754333496, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8979507684707642, + "num_tokens": 704892098.0, + "step": 18153 + }, + { + "epoch": 2.3093753975321207, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.778044700622559, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8881096839904785, + "num_tokens": 704926409.0, + "step": 18154 + }, + { + "epoch": 2.3095026078107113, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.638763427734375, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8990683555603027, + "num_tokens": 704964849.0, + "step": 18155 + }, + { + "epoch": 2.309629818089302, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.728416442871094, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.9012182950973511, + "num_tokens": 705002515.0, + "step": 18156 + }, + { + "epoch": 2.3097570283678923, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.734719276428223, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8939376473426819, + "num_tokens": 705042548.0, + "step": 18157 + }, + { + "epoch": 2.3098842386464824, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.814386367797852, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8851478099822998, + "num_tokens": 705078524.0, + "step": 18158 + }, + { + "epoch": 2.3100114489250734, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.747937202453613, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8956030607223511, + "num_tokens": 705117102.0, + "step": 18159 + }, + { + "epoch": 2.3101386592036635, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.759479522705078, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8930227756500244, + "num_tokens": 705155683.0, + "step": 18160 + }, + { + "epoch": 2.310265869482254, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.741576194763184, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.9044620990753174, + "num_tokens": 705191046.0, + "step": 18161 + }, + { + "epoch": 2.3103930797608445, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.730688095092773, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8958784341812134, + "num_tokens": 705234642.0, + "step": 18162 + }, + { + "epoch": 2.310520290039435, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.743880271911621, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9070947170257568, + "num_tokens": 705271358.0, + "step": 18163 + }, + { + "epoch": 2.3106475003180256, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.687077522277832, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8996216058731079, + "num_tokens": 705313664.0, + "step": 18164 + }, + { + "epoch": 2.310774710596616, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.738738059997559, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9088008403778076, + "num_tokens": 705349937.0, + "step": 18165 + }, + { + "epoch": 2.3109019208752066, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.649826049804688, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8972296118736267, + "num_tokens": 705392234.0, + "step": 18166 + }, + { + "epoch": 2.311029131153797, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.736092567443848, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.9014937877655029, + "num_tokens": 705427280.0, + "step": 18167 + }, + { + "epoch": 2.3111563414323877, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.746429443359375, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8963642716407776, + "num_tokens": 705465477.0, + "step": 18168 + }, + { + "epoch": 2.311283551710978, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.689300537109375, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8891441226005554, + "num_tokens": 705508340.0, + "step": 18169 + }, + { + "epoch": 2.3114107619895687, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.677769660949707, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.9020735621452332, + "num_tokens": 705553754.0, + "step": 18170 + }, + { + "epoch": 2.3115379722681593, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.57767391204834, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9051996469497681, + "num_tokens": 705599371.0, + "step": 18171 + }, + { + "epoch": 2.31166518254675, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.900619506835938, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8931276798248291, + "num_tokens": 705638859.0, + "step": 18172 + }, + { + "epoch": 2.3117923928253403, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.71216869354248, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8932210803031921, + "num_tokens": 705675939.0, + "step": 18173 + }, + { + "epoch": 2.311919603103931, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.75967788696289, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8948788642883301, + "num_tokens": 705717426.0, + "step": 18174 + }, + { + "epoch": 2.3120468133825214, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.693644523620605, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8984794020652771, + "num_tokens": 705753670.0, + "step": 18175 + }, + { + "epoch": 2.312174023661112, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.753689765930176, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8933117985725403, + "num_tokens": 705795275.0, + "step": 18176 + }, + { + "epoch": 2.3123012339397024, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.79362964630127, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8965036273002625, + "num_tokens": 705830541.0, + "step": 18177 + }, + { + "epoch": 2.312428444218293, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.675247192382812, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.9192816615104675, + "num_tokens": 705869938.0, + "step": 18178 + }, + { + "epoch": 2.3125556544968835, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.651087760925293, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.900383710861206, + "num_tokens": 705913377.0, + "step": 18179 + }, + { + "epoch": 2.312682864775474, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.749157905578613, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.9015390276908875, + "num_tokens": 705953494.0, + "step": 18180 + }, + { + "epoch": 2.3128100750540646, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.708205223083496, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.9007248282432556, + "num_tokens": 705998350.0, + "step": 18181 + }, + { + "epoch": 2.312937285332655, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.673416137695312, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8978925347328186, + "num_tokens": 706041706.0, + "step": 18182 + }, + { + "epoch": 2.313064495611245, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.766603469848633, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.9003796577453613, + "num_tokens": 706077136.0, + "step": 18183 + }, + { + "epoch": 2.313191705889836, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.776931762695312, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8952716588973999, + "num_tokens": 706115873.0, + "step": 18184 + }, + { + "epoch": 2.313318916168426, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.757139205932617, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.899020254611969, + "num_tokens": 706149613.0, + "step": 18185 + }, + { + "epoch": 2.3134461264470167, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.882841110229492, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8974445462226868, + "num_tokens": 706182337.0, + "step": 18186 + }, + { + "epoch": 2.3135733367256073, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.614816665649414, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8959115147590637, + "num_tokens": 706227545.0, + "step": 18187 + }, + { + "epoch": 2.313700547004198, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.789379119873047, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8864946365356445, + "num_tokens": 706265928.0, + "step": 18188 + }, + { + "epoch": 2.3138277572827883, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.816120147705078, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.9046934247016907, + "num_tokens": 706301260.0, + "step": 18189 + }, + { + "epoch": 2.313954967561379, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.646008491516113, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8939206600189209, + "num_tokens": 706343096.0, + "step": 18190 + }, + { + "epoch": 2.3140821778399694, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.671780586242676, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9013583660125732, + "num_tokens": 706386432.0, + "step": 18191 + }, + { + "epoch": 2.31420938811856, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.695551872253418, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8881072402000427, + "num_tokens": 706432192.0, + "step": 18192 + }, + { + "epoch": 2.3143365983971504, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.63681697845459, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.9058194160461426, + "num_tokens": 706471902.0, + "step": 18193 + }, + { + "epoch": 2.314463808675741, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.662457466125488, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8992840647697449, + "num_tokens": 706516276.0, + "step": 18194 + }, + { + "epoch": 2.3145910189543315, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.710365295410156, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.9026192426681519, + "num_tokens": 706555428.0, + "step": 18195 + }, + { + "epoch": 2.314718229232922, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.751273155212402, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9034792184829712, + "num_tokens": 706593121.0, + "step": 18196 + }, + { + "epoch": 2.3148454395115126, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.779454231262207, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8968008756637573, + "num_tokens": 706629157.0, + "step": 18197 + }, + { + "epoch": 2.314972649790103, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.621013641357422, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8946473598480225, + "num_tokens": 706678578.0, + "step": 18198 + }, + { + "epoch": 2.3150998600686936, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.604111671447754, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.9018901586532593, + "num_tokens": 706725033.0, + "step": 18199 + }, + { + "epoch": 2.315227070347284, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.698458671569824, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.9015371799468994, + "num_tokens": 706767938.0, + "step": 18200 + }, + { + "epoch": 2.3153542806258747, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.619805335998535, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.9006226658821106, + "num_tokens": 706808591.0, + "step": 18201 + }, + { + "epoch": 2.315481490904465, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.758092880249023, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8976075649261475, + "num_tokens": 706846666.0, + "step": 18202 + }, + { + "epoch": 2.3156087011830557, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.705004692077637, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8938930034637451, + "num_tokens": 706886155.0, + "step": 18203 + }, + { + "epoch": 2.3157359114616463, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.73438549041748, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.9030230045318604, + "num_tokens": 706928119.0, + "step": 18204 + }, + { + "epoch": 2.315863121740237, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.702757835388184, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9052252769470215, + "num_tokens": 706964774.0, + "step": 18205 + }, + { + "epoch": 2.3159903320188273, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.754864692687988, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.895455002784729, + "num_tokens": 707005592.0, + "step": 18206 + }, + { + "epoch": 2.316117542297418, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.72815990447998, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.9065531492233276, + "num_tokens": 707037188.0, + "step": 18207 + }, + { + "epoch": 2.316244752576008, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.798727989196777, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8990035653114319, + "num_tokens": 707080313.0, + "step": 18208 + }, + { + "epoch": 2.316371962854599, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.683356285095215, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8942518830299377, + "num_tokens": 707121692.0, + "step": 18209 + }, + { + "epoch": 2.316499173133189, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.79770278930664, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8895195722579956, + "num_tokens": 707158851.0, + "step": 18210 + }, + { + "epoch": 2.3166263834117795, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.801063537597656, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.890485405921936, + "num_tokens": 707192126.0, + "step": 18211 + }, + { + "epoch": 2.31675359369037, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.630472183227539, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8961336612701416, + "num_tokens": 707234230.0, + "step": 18212 + }, + { + "epoch": 2.3168808039689606, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.820478439331055, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8936077952384949, + "num_tokens": 707270452.0, + "step": 18213 + }, + { + "epoch": 2.317008014247551, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.76070785522461, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8938587307929993, + "num_tokens": 707308461.0, + "step": 18214 + }, + { + "epoch": 2.3171352245261416, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.766183853149414, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8896005749702454, + "num_tokens": 707345561.0, + "step": 18215 + }, + { + "epoch": 2.317262434804732, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.707877159118652, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8986129760742188, + "num_tokens": 707385288.0, + "step": 18216 + }, + { + "epoch": 2.3173896450833227, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.753373146057129, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8946212530136108, + "num_tokens": 707422853.0, + "step": 18217 + }, + { + "epoch": 2.317516855361913, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.711408615112305, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9145869612693787, + "num_tokens": 707460063.0, + "step": 18218 + }, + { + "epoch": 2.3176440656405037, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.69975471496582, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.9001550674438477, + "num_tokens": 707500701.0, + "step": 18219 + }, + { + "epoch": 2.3177712759190943, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.789326667785645, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8980218768119812, + "num_tokens": 707536539.0, + "step": 18220 + }, + { + "epoch": 2.317898486197685, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.78089427947998, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.906617283821106, + "num_tokens": 707571691.0, + "step": 18221 + }, + { + "epoch": 2.3180256964762753, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.716721534729004, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8912431597709656, + "num_tokens": 707609156.0, + "step": 18222 + }, + { + "epoch": 2.318152906754866, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.76513671875, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.896995484828949, + "num_tokens": 707647619.0, + "step": 18223 + }, + { + "epoch": 2.3182801170334564, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.696102142333984, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8957864046096802, + "num_tokens": 707687016.0, + "step": 18224 + }, + { + "epoch": 2.318407327312047, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.714922904968262, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.9016715288162231, + "num_tokens": 707722174.0, + "step": 18225 + }, + { + "epoch": 2.3185345375906374, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.880069732666016, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8981835842132568, + "num_tokens": 707754458.0, + "step": 18226 + }, + { + "epoch": 2.318661747869228, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.6881685256958, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8846839666366577, + "num_tokens": 707798417.0, + "step": 18227 + }, + { + "epoch": 2.3187889581478185, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.749099731445312, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8899731040000916, + "num_tokens": 707842901.0, + "step": 18228 + }, + { + "epoch": 2.318916168426409, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.789846420288086, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8918653726577759, + "num_tokens": 707880354.0, + "step": 18229 + }, + { + "epoch": 2.3190433787049995, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.75413990020752, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8972932696342468, + "num_tokens": 707919876.0, + "step": 18230 + }, + { + "epoch": 2.3191705889835896, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.731375694274902, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.9054486155509949, + "num_tokens": 707960783.0, + "step": 18231 + }, + { + "epoch": 2.3192977992621806, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.804116249084473, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8981515169143677, + "num_tokens": 707999324.0, + "step": 18232 + }, + { + "epoch": 2.3194250095407707, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.691096305847168, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9088169932365417, + "num_tokens": 708036409.0, + "step": 18233 + }, + { + "epoch": 2.319552219819361, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.684564590454102, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8959285020828247, + "num_tokens": 708081608.0, + "step": 18234 + }, + { + "epoch": 2.3196794300979517, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.656962394714355, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.912664532661438, + "num_tokens": 708117780.0, + "step": 18235 + }, + { + "epoch": 2.3198066403765423, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.702597618103027, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.9048742055892944, + "num_tokens": 708157822.0, + "step": 18236 + }, + { + "epoch": 2.319933850655133, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.665939331054688, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9033078551292419, + "num_tokens": 708199427.0, + "step": 18237 + }, + { + "epoch": 2.3200610609337233, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.78226375579834, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8841817378997803, + "num_tokens": 708238053.0, + "step": 18238 + }, + { + "epoch": 2.320188271212314, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.71240520477295, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8932009935379028, + "num_tokens": 708277025.0, + "step": 18239 + }, + { + "epoch": 2.3203154814909044, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.752948760986328, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.9040440917015076, + "num_tokens": 708307286.0, + "step": 18240 + }, + { + "epoch": 2.320442691769495, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.764300346374512, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.893363356590271, + "num_tokens": 708345026.0, + "step": 18241 + }, + { + "epoch": 2.3205699020480854, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.663128852844238, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8994982838630676, + "num_tokens": 708389448.0, + "step": 18242 + }, + { + "epoch": 2.320697112326676, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.824917793273926, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8943459391593933, + "num_tokens": 708425484.0, + "step": 18243 + }, + { + "epoch": 2.3208243226052665, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.665599822998047, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9101699590682983, + "num_tokens": 708465012.0, + "step": 18244 + }, + { + "epoch": 2.320951532883857, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.643661499023438, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8897382020950317, + "num_tokens": 708507743.0, + "step": 18245 + }, + { + "epoch": 2.3210787431624476, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.702533721923828, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.894271969795227, + "num_tokens": 708545054.0, + "step": 18246 + }, + { + "epoch": 2.321205953441038, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.780762672424316, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8994183540344238, + "num_tokens": 708581311.0, + "step": 18247 + }, + { + "epoch": 2.3213331637196286, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.736322402954102, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8887976408004761, + "num_tokens": 708621559.0, + "step": 18248 + }, + { + "epoch": 2.321460373998219, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.705556869506836, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.9121014475822449, + "num_tokens": 708657353.0, + "step": 18249 + }, + { + "epoch": 2.3215875842768097, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.724175453186035, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8895901441574097, + "num_tokens": 708694922.0, + "step": 18250 + }, + { + "epoch": 2.3217147945554, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.743230819702148, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8990302085876465, + "num_tokens": 708732639.0, + "step": 18251 + }, + { + "epoch": 2.3218420048339907, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.768965721130371, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8842265009880066, + "num_tokens": 708768132.0, + "step": 18252 + }, + { + "epoch": 2.3219692151125813, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.740496635437012, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.895004391670227, + "num_tokens": 708806199.0, + "step": 18253 + }, + { + "epoch": 2.322096425391172, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.639945983886719, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9071810245513916, + "num_tokens": 708850313.0, + "step": 18254 + }, + { + "epoch": 2.3222236356697623, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.569002151489258, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9122470617294312, + "num_tokens": 708894050.0, + "step": 18255 + }, + { + "epoch": 2.3223508459483524, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.888197898864746, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8892742395401001, + "num_tokens": 708928476.0, + "step": 18256 + }, + { + "epoch": 2.3224780562269434, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.63159465789795, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9023035168647766, + "num_tokens": 708971216.0, + "step": 18257 + }, + { + "epoch": 2.3226052665055335, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 10.06454086303711, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.888416051864624, + "num_tokens": 709012344.0, + "step": 18258 + }, + { + "epoch": 2.322732476784124, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.69485092163086, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.888125479221344, + "num_tokens": 709053156.0, + "step": 18259 + }, + { + "epoch": 2.3228596870627145, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.57215404510498, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.9068714380264282, + "num_tokens": 709092083.0, + "step": 18260 + }, + { + "epoch": 2.322986897341305, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.20159912109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.769125938415527, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8954604864120483, + "num_tokens": 709130885.0, + "step": 18261 + }, + { + "epoch": 2.3231141076198956, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.579063415527344, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.9007821083068848, + "num_tokens": 709174560.0, + "step": 18262 + }, + { + "epoch": 2.323241317898486, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.711698532104492, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8889349699020386, + "num_tokens": 709211925.0, + "step": 18263 + }, + { + "epoch": 2.3233685281770766, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.665578842163086, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8889517784118652, + "num_tokens": 709254552.0, + "step": 18264 + }, + { + "epoch": 2.323495738455667, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.688529968261719, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9040915966033936, + "num_tokens": 709290571.0, + "step": 18265 + }, + { + "epoch": 2.3236229487342577, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.735346794128418, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9110055565834045, + "num_tokens": 709331464.0, + "step": 18266 + }, + { + "epoch": 2.323750159012848, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.685503005981445, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.897350013256073, + "num_tokens": 709371129.0, + "step": 18267 + }, + { + "epoch": 2.3238773692914387, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.610791206359863, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.897667646408081, + "num_tokens": 709415614.0, + "step": 18268 + }, + { + "epoch": 2.3240045795700293, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.972290992736816, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8790595531463623, + "num_tokens": 709455099.0, + "step": 18269 + }, + { + "epoch": 2.32413178984862, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.655014038085938, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9039009809494019, + "num_tokens": 709493760.0, + "step": 18270 + }, + { + "epoch": 2.3242590001272103, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.685802459716797, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9049762487411499, + "num_tokens": 709535315.0, + "step": 18271 + }, + { + "epoch": 2.324386210405801, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.742644309997559, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9063578248023987, + "num_tokens": 709574870.0, + "step": 18272 + }, + { + "epoch": 2.3245134206843914, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.680490493774414, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9150735139846802, + "num_tokens": 709609902.0, + "step": 18273 + }, + { + "epoch": 2.324640630962982, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.648017883300781, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8890881538391113, + "num_tokens": 709656945.0, + "step": 18274 + }, + { + "epoch": 2.3247678412415724, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.703787803649902, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.899822473526001, + "num_tokens": 709702019.0, + "step": 18275 + }, + { + "epoch": 2.324895051520163, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.759023666381836, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.897441029548645, + "num_tokens": 709741858.0, + "step": 18276 + }, + { + "epoch": 2.3250222617987535, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.778825759887695, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8857541084289551, + "num_tokens": 709779546.0, + "step": 18277 + }, + { + "epoch": 2.325149472077344, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.795586585998535, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8930219411849976, + "num_tokens": 709817629.0, + "step": 18278 + }, + { + "epoch": 2.3252766823559345, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.753867149353027, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.9016700983047485, + "num_tokens": 709852327.0, + "step": 18279 + }, + { + "epoch": 2.325403892634525, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.806686401367188, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.9029377698898315, + "num_tokens": 709886635.0, + "step": 18280 + }, + { + "epoch": 2.325531102913115, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.699586868286133, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9069924354553223, + "num_tokens": 709926651.0, + "step": 18281 + }, + { + "epoch": 2.325658313191706, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.850935935974121, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8876116275787354, + "num_tokens": 709961644.0, + "step": 18282 + }, + { + "epoch": 2.325785523470296, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.737719535827637, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8870836496353149, + "num_tokens": 710004503.0, + "step": 18283 + }, + { + "epoch": 2.3259127337488867, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.636865615844727, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.906558632850647, + "num_tokens": 710045397.0, + "step": 18284 + }, + { + "epoch": 2.3260399440274773, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.837494850158691, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8925318717956543, + "num_tokens": 710080518.0, + "step": 18285 + }, + { + "epoch": 2.326167154306068, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.67681884765625, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8854315876960754, + "num_tokens": 710120282.0, + "step": 18286 + }, + { + "epoch": 2.3262943645846583, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.725289344787598, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.890905499458313, + "num_tokens": 710161965.0, + "step": 18287 + }, + { + "epoch": 2.326421574863249, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.64857292175293, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9056714773178101, + "num_tokens": 710207484.0, + "step": 18288 + }, + { + "epoch": 2.3265487851418394, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.67371654510498, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.9008331298828125, + "num_tokens": 710245823.0, + "step": 18289 + }, + { + "epoch": 2.32667599542043, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.729143142700195, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8817367553710938, + "num_tokens": 710292641.0, + "step": 18290 + }, + { + "epoch": 2.3268032056990204, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.725805282592773, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9083974361419678, + "num_tokens": 710327262.0, + "step": 18291 + }, + { + "epoch": 2.326930415977611, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.741730690002441, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9072409868240356, + "num_tokens": 710364078.0, + "step": 18292 + }, + { + "epoch": 2.3270576262562015, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.747208595275879, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8905609250068665, + "num_tokens": 710401873.0, + "step": 18293 + }, + { + "epoch": 2.327184836534792, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.801082611083984, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8955620527267456, + "num_tokens": 710434399.0, + "step": 18294 + }, + { + "epoch": 2.3273120468133826, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.699808120727539, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9038448929786682, + "num_tokens": 710469824.0, + "step": 18295 + }, + { + "epoch": 2.327439257091973, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.744760513305664, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8990402221679688, + "num_tokens": 710506211.0, + "step": 18296 + }, + { + "epoch": 2.3275664673705636, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.795256614685059, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8929036855697632, + "num_tokens": 710543557.0, + "step": 18297 + }, + { + "epoch": 2.327693677649154, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.649680137634277, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8949030637741089, + "num_tokens": 710584721.0, + "step": 18298 + }, + { + "epoch": 2.3278208879277447, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.786639213562012, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8950926065444946, + "num_tokens": 710621628.0, + "step": 18299 + }, + { + "epoch": 2.327948098206335, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.653417587280273, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8954613208770752, + "num_tokens": 710664456.0, + "step": 18300 + }, + { + "epoch": 2.3280753084849257, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.680103302001953, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9090369939804077, + "num_tokens": 710703044.0, + "step": 18301 + }, + { + "epoch": 2.3282025187635162, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.888949394226074, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8962200880050659, + "num_tokens": 710732632.0, + "step": 18302 + }, + { + "epoch": 2.3283297290421068, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.713408470153809, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8935946226119995, + "num_tokens": 710778453.0, + "step": 18303 + }, + { + "epoch": 2.3284569393206973, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.766046524047852, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8880866765975952, + "num_tokens": 710820057.0, + "step": 18304 + }, + { + "epoch": 2.328584149599288, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.729755401611328, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.899604082107544, + "num_tokens": 710859589.0, + "step": 18305 + }, + { + "epoch": 2.328711359877878, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.698373794555664, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9118900299072266, + "num_tokens": 710897298.0, + "step": 18306 + }, + { + "epoch": 2.328838570156469, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.725656509399414, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.894902229309082, + "num_tokens": 710934006.0, + "step": 18307 + }, + { + "epoch": 2.328965780435059, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.868332862854004, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8955122232437134, + "num_tokens": 710967561.0, + "step": 18308 + }, + { + "epoch": 2.3290929907136495, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.580849647521973, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.9023929238319397, + "num_tokens": 711009734.0, + "step": 18309 + }, + { + "epoch": 2.32922020099224, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.759662628173828, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8966790437698364, + "num_tokens": 711046050.0, + "step": 18310 + }, + { + "epoch": 2.3293474112708306, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.723718643188477, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8874919414520264, + "num_tokens": 711090177.0, + "step": 18311 + }, + { + "epoch": 2.329474621549421, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.813525199890137, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8827991485595703, + "num_tokens": 711128254.0, + "step": 18312 + }, + { + "epoch": 2.3296018318280116, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.778542518615723, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9048765897750854, + "num_tokens": 711159173.0, + "step": 18313 + }, + { + "epoch": 2.329729042106602, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.777764320373535, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9056390523910522, + "num_tokens": 711200763.0, + "step": 18314 + }, + { + "epoch": 2.3298562523851927, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.719534873962402, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.9009340405464172, + "num_tokens": 711240457.0, + "step": 18315 + }, + { + "epoch": 2.329983462663783, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.787728309631348, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8947421312332153, + "num_tokens": 711277824.0, + "step": 18316 + }, + { + "epoch": 2.3301106729423737, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.716734886169434, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.894715428352356, + "num_tokens": 711315450.0, + "step": 18317 + }, + { + "epoch": 2.3302378832209643, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.73668098449707, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9012658596038818, + "num_tokens": 711350234.0, + "step": 18318 + }, + { + "epoch": 2.330365093499555, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.837464332580566, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8840446472167969, + "num_tokens": 711388994.0, + "step": 18319 + }, + { + "epoch": 2.3304923037781453, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.7296724319458, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.9035872220993042, + "num_tokens": 711428747.0, + "step": 18320 + }, + { + "epoch": 2.330619514056736, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.87375545501709, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8885953426361084, + "num_tokens": 711461755.0, + "step": 18321 + }, + { + "epoch": 2.3307467243353264, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.784514427185059, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8879464268684387, + "num_tokens": 711498815.0, + "step": 18322 + }, + { + "epoch": 2.330873934613917, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.802597999572754, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9100584983825684, + "num_tokens": 711528323.0, + "step": 18323 + }, + { + "epoch": 2.3310011448925074, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.755224227905273, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9109325408935547, + "num_tokens": 711565994.0, + "step": 18324 + }, + { + "epoch": 2.331128355171098, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.731793403625488, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8860963582992554, + "num_tokens": 711604799.0, + "step": 18325 + }, + { + "epoch": 2.3312555654496885, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.595263481140137, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.9039778709411621, + "num_tokens": 711645019.0, + "step": 18326 + }, + { + "epoch": 2.331382775728279, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.824663162231445, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9057691097259521, + "num_tokens": 711679576.0, + "step": 18327 + }, + { + "epoch": 2.3315099860068695, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.653325080871582, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.9006752967834473, + "num_tokens": 711719896.0, + "step": 18328 + }, + { + "epoch": 2.3316371962854596, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.753494262695312, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8870278596878052, + "num_tokens": 711757856.0, + "step": 18329 + }, + { + "epoch": 2.3317644065640506, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.691540718078613, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9014690518379211, + "num_tokens": 711798090.0, + "step": 18330 + }, + { + "epoch": 2.3318916168426407, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.755059242248535, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8893924951553345, + "num_tokens": 711834060.0, + "step": 18331 + }, + { + "epoch": 2.332018827121231, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.687943458557129, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8995435237884521, + "num_tokens": 711878561.0, + "step": 18332 + }, + { + "epoch": 2.3321460373998217, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.722111701965332, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8988136053085327, + "num_tokens": 711914433.0, + "step": 18333 + }, + { + "epoch": 2.3322732476784123, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.746713638305664, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9121357798576355, + "num_tokens": 711950150.0, + "step": 18334 + }, + { + "epoch": 2.332400457957003, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.710251808166504, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.895710289478302, + "num_tokens": 711989757.0, + "step": 18335 + }, + { + "epoch": 2.3325276682355933, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.703439712524414, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.876858115196228, + "num_tokens": 712036789.0, + "step": 18336 + }, + { + "epoch": 2.332654878514184, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.797441482543945, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8999485969543457, + "num_tokens": 712071879.0, + "step": 18337 + }, + { + "epoch": 2.3327820887927744, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.741819381713867, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8953953981399536, + "num_tokens": 712106666.0, + "step": 18338 + }, + { + "epoch": 2.332909299071365, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.778318405151367, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8954614400863647, + "num_tokens": 712144551.0, + "step": 18339 + }, + { + "epoch": 2.3330365093499554, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.685482025146484, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8990979194641113, + "num_tokens": 712187635.0, + "step": 18340 + }, + { + "epoch": 2.333163719628546, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.729340553283691, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.910798966884613, + "num_tokens": 712225342.0, + "step": 18341 + }, + { + "epoch": 2.3332909299071365, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.746606826782227, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8863790035247803, + "num_tokens": 712265062.0, + "step": 18342 + }, + { + "epoch": 2.333418140185727, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.8817138671875, + "learning_rate": 1e-06, + "loss": 0.3761, + "mean_token_accuracy": 0.8720364570617676, + "num_tokens": 712307104.0, + "step": 18343 + }, + { + "epoch": 2.3335453504643175, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.824289321899414, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8953621983528137, + "num_tokens": 712339081.0, + "step": 18344 + }, + { + "epoch": 2.333672560742908, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.693037986755371, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8980181813240051, + "num_tokens": 712377394.0, + "step": 18345 + }, + { + "epoch": 2.3337997710214986, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.694130897521973, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8932151794433594, + "num_tokens": 712416674.0, + "step": 18346 + }, + { + "epoch": 2.333926981300089, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.67629337310791, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9036002159118652, + "num_tokens": 712457565.0, + "step": 18347 + }, + { + "epoch": 2.3340541915786797, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.764792442321777, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8830657005310059, + "num_tokens": 712494865.0, + "step": 18348 + }, + { + "epoch": 2.33418140185727, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.616189002990723, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9044099450111389, + "num_tokens": 712534497.0, + "step": 18349 + }, + { + "epoch": 2.3343086121358607, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.855920791625977, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8974078297615051, + "num_tokens": 712568805.0, + "step": 18350 + }, + { + "epoch": 2.3344358224144512, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.705150604248047, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9194327592849731, + "num_tokens": 712605368.0, + "step": 18351 + }, + { + "epoch": 2.3345630326930418, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.76070785522461, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.896273136138916, + "num_tokens": 712639536.0, + "step": 18352 + }, + { + "epoch": 2.3346902429716323, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.718539237976074, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9016972780227661, + "num_tokens": 712674596.0, + "step": 18353 + }, + { + "epoch": 2.3348174532502224, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.73688793182373, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8888756036758423, + "num_tokens": 712711776.0, + "step": 18354 + }, + { + "epoch": 2.3349446635288134, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.715660095214844, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9077529907226562, + "num_tokens": 712752103.0, + "step": 18355 + }, + { + "epoch": 2.3350718738074034, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.739630699157715, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9080139398574829, + "num_tokens": 712787868.0, + "step": 18356 + }, + { + "epoch": 2.335199084085994, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.768132209777832, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9010648727416992, + "num_tokens": 712824560.0, + "step": 18357 + }, + { + "epoch": 2.3353262943645845, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.666525840759277, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.9020757079124451, + "num_tokens": 712864544.0, + "step": 18358 + }, + { + "epoch": 2.335453504643175, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.716538429260254, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.904258668422699, + "num_tokens": 712899150.0, + "step": 18359 + }, + { + "epoch": 2.3355807149217656, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.621204376220703, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.902133047580719, + "num_tokens": 712945891.0, + "step": 18360 + }, + { + "epoch": 2.335707925200356, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.715924263000488, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.9028140306472778, + "num_tokens": 712985209.0, + "step": 18361 + }, + { + "epoch": 2.3358351354789466, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.719768524169922, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.9025713205337524, + "num_tokens": 713026194.0, + "step": 18362 + }, + { + "epoch": 2.335962345757537, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.708736419677734, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.9028525352478027, + "num_tokens": 713069093.0, + "step": 18363 + }, + { + "epoch": 2.3360895560361277, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.786698341369629, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8989067673683167, + "num_tokens": 713107586.0, + "step": 18364 + }, + { + "epoch": 2.336216766314718, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.672614097595215, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.895311713218689, + "num_tokens": 713151360.0, + "step": 18365 + }, + { + "epoch": 2.3363439765933087, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.810319900512695, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8934844136238098, + "num_tokens": 713189558.0, + "step": 18366 + }, + { + "epoch": 2.3364711868718993, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.773892402648926, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8976994752883911, + "num_tokens": 713225202.0, + "step": 18367 + }, + { + "epoch": 2.33659839715049, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.690345764160156, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9031226634979248, + "num_tokens": 713265860.0, + "step": 18368 + }, + { + "epoch": 2.3367256074290803, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 10.121983528137207, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8970977067947388, + "num_tokens": 713303554.0, + "step": 18369 + }, + { + "epoch": 2.336852817707671, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.738068580627441, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.896576464176178, + "num_tokens": 713339735.0, + "step": 18370 + }, + { + "epoch": 2.3369800279862614, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.692439079284668, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8996713161468506, + "num_tokens": 713378399.0, + "step": 18371 + }, + { + "epoch": 2.337107238264852, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.724979400634766, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8933563828468323, + "num_tokens": 713417563.0, + "step": 18372 + }, + { + "epoch": 2.3372344485434424, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.783773422241211, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8942762017250061, + "num_tokens": 713453104.0, + "step": 18373 + }, + { + "epoch": 2.337361658822033, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.696723937988281, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8996407985687256, + "num_tokens": 713491388.0, + "step": 18374 + }, + { + "epoch": 2.3374888691006235, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.711332321166992, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.9027875065803528, + "num_tokens": 713531626.0, + "step": 18375 + }, + { + "epoch": 2.337616079379214, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.817424774169922, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8939230442047119, + "num_tokens": 713567375.0, + "step": 18376 + }, + { + "epoch": 2.3377432896578045, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.817039489746094, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8806111216545105, + "num_tokens": 713609049.0, + "step": 18377 + }, + { + "epoch": 2.337870499936395, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 10.01431655883789, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8961702585220337, + "num_tokens": 713650400.0, + "step": 18378 + }, + { + "epoch": 2.337997710214985, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.757719039916992, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.9018265008926392, + "num_tokens": 713685327.0, + "step": 18379 + }, + { + "epoch": 2.338124920493576, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.70805549621582, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8770442008972168, + "num_tokens": 713727427.0, + "step": 18380 + }, + { + "epoch": 2.338252130772166, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.696574211120605, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8966752886772156, + "num_tokens": 713766380.0, + "step": 18381 + }, + { + "epoch": 2.3383793410507567, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.742850303649902, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.898772120475769, + "num_tokens": 713800108.0, + "step": 18382 + }, + { + "epoch": 2.3385065513293473, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.637544631958008, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.9019150733947754, + "num_tokens": 713841583.0, + "step": 18383 + }, + { + "epoch": 2.338633761607938, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.651222229003906, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9086943864822388, + "num_tokens": 713886574.0, + "step": 18384 + }, + { + "epoch": 2.3387609718865283, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.700468063354492, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8936992883682251, + "num_tokens": 713928360.0, + "step": 18385 + }, + { + "epoch": 2.338888182165119, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.82025146484375, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8959077596664429, + "num_tokens": 713961192.0, + "step": 18386 + }, + { + "epoch": 2.3390153924437094, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.650350570678711, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8946418762207031, + "num_tokens": 714004637.0, + "step": 18387 + }, + { + "epoch": 2.3391426027223, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.778611183166504, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8948809504508972, + "num_tokens": 714040550.0, + "step": 18388 + }, + { + "epoch": 2.3392698130008904, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.808903694152832, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.9006926417350769, + "num_tokens": 714075272.0, + "step": 18389 + }, + { + "epoch": 2.339397023279481, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.64759349822998, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8972219824790955, + "num_tokens": 714116877.0, + "step": 18390 + }, + { + "epoch": 2.3395242335580715, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.756538391113281, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8698325157165527, + "num_tokens": 714156410.0, + "step": 18391 + }, + { + "epoch": 2.339651443836662, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.705263137817383, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8917018175125122, + "num_tokens": 714197977.0, + "step": 18392 + }, + { + "epoch": 2.3397786541152525, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.761055946350098, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.9030719995498657, + "num_tokens": 714239753.0, + "step": 18393 + }, + { + "epoch": 2.339905864393843, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.705143928527832, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8946621417999268, + "num_tokens": 714279228.0, + "step": 18394 + }, + { + "epoch": 2.3400330746724336, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.69443416595459, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8910725116729736, + "num_tokens": 714317728.0, + "step": 18395 + }, + { + "epoch": 2.340160284951024, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.72773551940918, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.9020673036575317, + "num_tokens": 714357311.0, + "step": 18396 + }, + { + "epoch": 2.3402874952296147, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.681784629821777, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9096776247024536, + "num_tokens": 714391973.0, + "step": 18397 + }, + { + "epoch": 2.340414705508205, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.296966552734375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.65917682647705, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9066163301467896, + "num_tokens": 714434347.0, + "step": 18398 + }, + { + "epoch": 2.3405419157867957, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.296966552734375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.759763717651367, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8968163132667542, + "num_tokens": 714479710.0, + "step": 18399 + }, + { + "epoch": 2.3406691260653862, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.296966552734375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.8143892288208, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8921605348587036, + "num_tokens": 714513956.0, + "step": 18400 + }, + { + "epoch": 2.3407963363439768, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.296966552734375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.619894981384277, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9057363867759705, + "num_tokens": 714557996.0, + "step": 18401 + }, + { + "epoch": 2.3409235466225673, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.683032035827637, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9058271646499634, + "num_tokens": 714598273.0, + "step": 18402 + }, + { + "epoch": 2.341050756901158, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 10.028515815734863, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8923873901367188, + "num_tokens": 714628359.0, + "step": 18403 + }, + { + "epoch": 2.341177967179748, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.789469718933105, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.9033491611480713, + "num_tokens": 714659984.0, + "step": 18404 + }, + { + "epoch": 2.341305177458339, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.755550384521484, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9026100635528564, + "num_tokens": 714696046.0, + "step": 18405 + }, + { + "epoch": 2.341432387736929, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.729808807373047, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9080681800842285, + "num_tokens": 714732902.0, + "step": 18406 + }, + { + "epoch": 2.3415595980155195, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.914765357971191, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.889316976070404, + "num_tokens": 714770524.0, + "step": 18407 + }, + { + "epoch": 2.34168680829411, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.66201400756836, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.900812029838562, + "num_tokens": 714808550.0, + "step": 18408 + }, + { + "epoch": 2.3418140185727006, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.804337501525879, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.893140435218811, + "num_tokens": 714840733.0, + "step": 18409 + }, + { + "epoch": 2.341941228851291, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.78837776184082, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9044645428657532, + "num_tokens": 714878757.0, + "step": 18410 + }, + { + "epoch": 2.3420684391298816, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.682151794433594, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.9035265445709229, + "num_tokens": 714914864.0, + "step": 18411 + }, + { + "epoch": 2.342195649408472, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.702308654785156, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8932404518127441, + "num_tokens": 714958733.0, + "step": 18412 + }, + { + "epoch": 2.3423228596870627, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.694635391235352, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8921604752540588, + "num_tokens": 714999162.0, + "step": 18413 + }, + { + "epoch": 2.342450069965653, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.695086479187012, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8990557193756104, + "num_tokens": 715038838.0, + "step": 18414 + }, + { + "epoch": 2.3425772802442437, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.711718559265137, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8928026556968689, + "num_tokens": 715082688.0, + "step": 18415 + }, + { + "epoch": 2.3427044905228342, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.638284683227539, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9095662832260132, + "num_tokens": 715125822.0, + "step": 18416 + }, + { + "epoch": 2.3428317008014248, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.69495677947998, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9042707085609436, + "num_tokens": 715165023.0, + "step": 18417 + }, + { + "epoch": 2.3429589110800153, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.717628479003906, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.906640887260437, + "num_tokens": 715202742.0, + "step": 18418 + }, + { + "epoch": 2.343086121358606, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.71365737915039, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.896307110786438, + "num_tokens": 715245025.0, + "step": 18419 + }, + { + "epoch": 2.3432133316371964, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.686237335205078, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9028488993644714, + "num_tokens": 715283623.0, + "step": 18420 + }, + { + "epoch": 2.343340541915787, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.704739570617676, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8921324014663696, + "num_tokens": 715325815.0, + "step": 18421 + }, + { + "epoch": 2.3434677521943774, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.74802017211914, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8940433859825134, + "num_tokens": 715366208.0, + "step": 18422 + }, + { + "epoch": 2.343594962472968, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.69075870513916, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8905529379844666, + "num_tokens": 715410394.0, + "step": 18423 + }, + { + "epoch": 2.3437221727515585, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.801087379455566, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8917038440704346, + "num_tokens": 715452437.0, + "step": 18424 + }, + { + "epoch": 2.343849383030149, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.845762252807617, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8968740701675415, + "num_tokens": 715486630.0, + "step": 18425 + }, + { + "epoch": 2.3439765933087395, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.72596263885498, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8964865207672119, + "num_tokens": 715523680.0, + "step": 18426 + }, + { + "epoch": 2.3441038035873296, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.8306245803833, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8834273815155029, + "num_tokens": 715558730.0, + "step": 18427 + }, + { + "epoch": 2.3442310138659206, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.744048118591309, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.899285078048706, + "num_tokens": 715593964.0, + "step": 18428 + }, + { + "epoch": 2.3443582241445107, + "ewc_loss": 0.019775390625, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.729286193847656, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9097981452941895, + "num_tokens": 715630943.0, + "step": 18429 + }, + { + "epoch": 2.344485434423101, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.702064514160156, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.896996259689331, + "num_tokens": 715667639.0, + "step": 18430 + }, + { + "epoch": 2.3446126447016917, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.690062522888184, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8982973098754883, + "num_tokens": 715709982.0, + "step": 18431 + }, + { + "epoch": 2.3447398549802823, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.681843757629395, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.907550573348999, + "num_tokens": 715744359.0, + "step": 18432 + }, + { + "epoch": 2.344867065258873, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.673648834228516, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.906914234161377, + "num_tokens": 715782581.0, + "step": 18433 + }, + { + "epoch": 2.3449942755374633, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.64319896697998, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9051307439804077, + "num_tokens": 715820298.0, + "step": 18434 + }, + { + "epoch": 2.345121485816054, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.249282836914062e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.691474914550781, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8981786966323853, + "num_tokens": 715857753.0, + "step": 18435 + }, + { + "epoch": 2.3452486960946444, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.676473617553711, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.9012336730957031, + "num_tokens": 715899874.0, + "step": 18436 + }, + { + "epoch": 2.345375906373235, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.706327438354492, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9086496829986572, + "num_tokens": 715938633.0, + "step": 18437 + }, + { + "epoch": 2.3455031166518254, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.635228157043457, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.9078929424285889, + "num_tokens": 715981113.0, + "step": 18438 + }, + { + "epoch": 2.345630326930416, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.705473899841309, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.9057930111885071, + "num_tokens": 716023464.0, + "step": 18439 + }, + { + "epoch": 2.3457575372090065, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.761126518249512, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.894416868686676, + "num_tokens": 716060752.0, + "step": 18440 + }, + { + "epoch": 2.345884747487597, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.821593284606934, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.910200834274292, + "num_tokens": 716092347.0, + "step": 18441 + }, + { + "epoch": 2.3460119577661875, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.629273414611816, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.902340292930603, + "num_tokens": 716133708.0, + "step": 18442 + }, + { + "epoch": 2.346139168044778, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.716694831848145, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8972179889678955, + "num_tokens": 716168059.0, + "step": 18443 + }, + { + "epoch": 2.3462663783233686, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.633633613586426, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8986472487449646, + "num_tokens": 716209411.0, + "step": 18444 + }, + { + "epoch": 2.346393588601959, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.621835708618164, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9063664674758911, + "num_tokens": 716250949.0, + "step": 18445 + }, + { + "epoch": 2.3465207988805497, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.717886924743652, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.9019862413406372, + "num_tokens": 716285764.0, + "step": 18446 + }, + { + "epoch": 2.34664800915914, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.677462577819824, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.9005183577537537, + "num_tokens": 716324151.0, + "step": 18447 + }, + { + "epoch": 2.3467752194377307, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.613876342773438, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8962960243225098, + "num_tokens": 716363614.0, + "step": 18448 + }, + { + "epoch": 2.3469024297163212, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.768594741821289, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8990586400032043, + "num_tokens": 716403281.0, + "step": 18449 + }, + { + "epoch": 2.3470296399949118, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.701970100402832, + "learning_rate": 1e-06, + "loss": 0.2575, + "mean_token_accuracy": 0.9124316573143005, + "num_tokens": 716436833.0, + "step": 18450 + }, + { + "epoch": 2.3471568502735023, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.728906631469727, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8891879916191101, + "num_tokens": 716471693.0, + "step": 18451 + }, + { + "epoch": 2.3472840605520924, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.678131103515625, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9045255184173584, + "num_tokens": 716512584.0, + "step": 18452 + }, + { + "epoch": 2.3474112708306833, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.717386245727539, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9020461440086365, + "num_tokens": 716547218.0, + "step": 18453 + }, + { + "epoch": 2.3475384811092734, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.739903450012207, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9125328063964844, + "num_tokens": 716582564.0, + "step": 18454 + }, + { + "epoch": 2.347665691387864, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.74643611907959, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8952764272689819, + "num_tokens": 716619076.0, + "step": 18455 + }, + { + "epoch": 2.3477929016664545, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.827284812927246, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8939651846885681, + "num_tokens": 716655820.0, + "step": 18456 + }, + { + "epoch": 2.347920111945045, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.59377670288086, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9052364230155945, + "num_tokens": 716693910.0, + "step": 18457 + }, + { + "epoch": 2.3480473222236355, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.751897811889648, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8850496411323547, + "num_tokens": 716737252.0, + "step": 18458 + }, + { + "epoch": 2.348174532502226, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.754524230957031, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9132212400436401, + "num_tokens": 716770298.0, + "step": 18459 + }, + { + "epoch": 2.3483017427808166, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.702337265014648, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.9053246974945068, + "num_tokens": 716804768.0, + "step": 18460 + }, + { + "epoch": 2.348428953059407, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.757977485656738, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9085665941238403, + "num_tokens": 716840696.0, + "step": 18461 + }, + { + "epoch": 2.3485561633379977, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.668588638305664, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8974438905715942, + "num_tokens": 716885091.0, + "step": 18462 + }, + { + "epoch": 2.348683373616588, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.689935684204102, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.9045088887214661, + "num_tokens": 716923145.0, + "step": 18463 + }, + { + "epoch": 2.3488105838951787, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.719049453735352, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.900429368019104, + "num_tokens": 716964174.0, + "step": 18464 + }, + { + "epoch": 2.3489377941737692, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.701082229614258, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9089827537536621, + "num_tokens": 717005134.0, + "step": 18465 + }, + { + "epoch": 2.3490650044523598, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.722079277038574, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.910513162612915, + "num_tokens": 717040136.0, + "step": 18466 + }, + { + "epoch": 2.3491922147309503, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.75048828125, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.9007041454315186, + "num_tokens": 717078326.0, + "step": 18467 + }, + { + "epoch": 2.349319425009541, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.666019439697266, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.906426191329956, + "num_tokens": 717119866.0, + "step": 18468 + }, + { + "epoch": 2.3494466352881314, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.925737380981445, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8922748565673828, + "num_tokens": 717152607.0, + "step": 18469 + }, + { + "epoch": 2.349573845566722, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.75675106048584, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.89811110496521, + "num_tokens": 717192783.0, + "step": 18470 + }, + { + "epoch": 2.3497010558453124, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.531107902526855, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9191152453422546, + "num_tokens": 717236993.0, + "step": 18471 + }, + { + "epoch": 2.349828266123903, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.818925857543945, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8794089555740356, + "num_tokens": 717280453.0, + "step": 18472 + }, + { + "epoch": 2.3499554764024935, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.678506851196289, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8963910341262817, + "num_tokens": 717319271.0, + "step": 18473 + }, + { + "epoch": 2.350082686681084, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.724556922912598, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8994020819664001, + "num_tokens": 717359906.0, + "step": 18474 + }, + { + "epoch": 2.3502098969596745, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.767316818237305, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8885214328765869, + "num_tokens": 717398916.0, + "step": 18475 + }, + { + "epoch": 2.350337107238265, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.780698776245117, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.888509213924408, + "num_tokens": 717437414.0, + "step": 18476 + }, + { + "epoch": 2.350464317516855, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.637029647827148, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9129563570022583, + "num_tokens": 717475863.0, + "step": 18477 + }, + { + "epoch": 2.350591527795446, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.737405776977539, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9065027236938477, + "num_tokens": 717509374.0, + "step": 18478 + }, + { + "epoch": 2.350718738074036, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.683287620544434, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.9001097679138184, + "num_tokens": 717552657.0, + "step": 18479 + }, + { + "epoch": 2.3508459483526267, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.71728515625, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9141637086868286, + "num_tokens": 717590389.0, + "step": 18480 + }, + { + "epoch": 2.3509731586312173, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.884989738464355, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8821666240692139, + "num_tokens": 717625132.0, + "step": 18481 + }, + { + "epoch": 2.351100368909808, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.705316543579102, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.9012203812599182, + "num_tokens": 717664465.0, + "step": 18482 + }, + { + "epoch": 2.3512275791883983, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.700448989868164, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9090150594711304, + "num_tokens": 717699898.0, + "step": 18483 + }, + { + "epoch": 2.351354789466989, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.767394065856934, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8984789848327637, + "num_tokens": 717738988.0, + "step": 18484 + }, + { + "epoch": 2.3514819997455794, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.809913635253906, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8878099918365479, + "num_tokens": 717773960.0, + "step": 18485 + }, + { + "epoch": 2.35160921002417, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.810938835144043, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8999996185302734, + "num_tokens": 717812994.0, + "step": 18486 + }, + { + "epoch": 2.3517364203027604, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.745923042297363, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8960685729980469, + "num_tokens": 717846630.0, + "step": 18487 + }, + { + "epoch": 2.351863630581351, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.79860782623291, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8950198292732239, + "num_tokens": 717883972.0, + "step": 18488 + }, + { + "epoch": 2.3519908408599415, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.66909408569336, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.9006595611572266, + "num_tokens": 717925934.0, + "step": 18489 + }, + { + "epoch": 2.352118051138532, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.67345142364502, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8963647484779358, + "num_tokens": 717970474.0, + "step": 18490 + }, + { + "epoch": 2.3522452614171225, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 10.005254745483398, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.893634557723999, + "num_tokens": 718009236.0, + "step": 18491 + }, + { + "epoch": 2.352372471695713, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.741927146911621, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8966844081878662, + "num_tokens": 718048138.0, + "step": 18492 + }, + { + "epoch": 2.3524996819743036, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.777466773986816, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8948268890380859, + "num_tokens": 718082536.0, + "step": 18493 + }, + { + "epoch": 2.352626892252894, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.71707534790039, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8982521295547485, + "num_tokens": 718123342.0, + "step": 18494 + }, + { + "epoch": 2.3527541025314846, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.678166389465332, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.905531644821167, + "num_tokens": 718164286.0, + "step": 18495 + }, + { + "epoch": 2.352881312810075, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.632010459899902, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.9030962586402893, + "num_tokens": 718207785.0, + "step": 18496 + }, + { + "epoch": 2.3530085230886657, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.720996856689453, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8983943462371826, + "num_tokens": 718244941.0, + "step": 18497 + }, + { + "epoch": 2.3531357333672562, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.696824073791504, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8982300162315369, + "num_tokens": 718284755.0, + "step": 18498 + }, + { + "epoch": 2.3532629436458468, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.703651428222656, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8948944807052612, + "num_tokens": 718327148.0, + "step": 18499 + }, + { + "epoch": 2.353390153924437, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.679609298706055, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8945189714431763, + "num_tokens": 718374333.0, + "step": 18500 + }, + { + "epoch": 2.353517364203028, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.559471130371094, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8983435034751892, + "num_tokens": 718421415.0, + "step": 18501 + }, + { + "epoch": 2.353644574481618, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.717976570129395, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.887934684753418, + "num_tokens": 718456474.0, + "step": 18502 + }, + { + "epoch": 2.353771784760209, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.774810791015625, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8927862644195557, + "num_tokens": 718491090.0, + "step": 18503 + }, + { + "epoch": 2.353898995038799, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.596281051635742, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9139115810394287, + "num_tokens": 718530883.0, + "step": 18504 + }, + { + "epoch": 2.3540262053173895, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.72308349609375, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.886806309223175, + "num_tokens": 718573668.0, + "step": 18505 + }, + { + "epoch": 2.35415341559598, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.689521789550781, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.9005071520805359, + "num_tokens": 718613791.0, + "step": 18506 + }, + { + "epoch": 2.3542806258745705, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.607624053955078, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9065041542053223, + "num_tokens": 718654468.0, + "step": 18507 + }, + { + "epoch": 2.354407836153161, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.811134338378906, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8896069526672363, + "num_tokens": 718689386.0, + "step": 18508 + }, + { + "epoch": 2.3545350464317516, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.605509757995605, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.9075483083724976, + "num_tokens": 718733966.0, + "step": 18509 + }, + { + "epoch": 2.354662256710342, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.868243217468262, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8959325551986694, + "num_tokens": 718769560.0, + "step": 18510 + }, + { + "epoch": 2.3547894669889327, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.709746360778809, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.9048632383346558, + "num_tokens": 718806440.0, + "step": 18511 + }, + { + "epoch": 2.354916677267523, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.344650268554688e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.661096572875977, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9098546504974365, + "num_tokens": 718847349.0, + "step": 18512 + }, + { + "epoch": 2.3550438875461137, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.640908241271973, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8972815871238708, + "num_tokens": 718888166.0, + "step": 18513 + }, + { + "epoch": 2.3551710978247042, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.780570983886719, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8972223997116089, + "num_tokens": 718929313.0, + "step": 18514 + }, + { + "epoch": 2.3552983081032948, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.73213005065918, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.9040130972862244, + "num_tokens": 718967775.0, + "step": 18515 + }, + { + "epoch": 2.3554255183818853, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.772544860839844, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8911942839622498, + "num_tokens": 719007319.0, + "step": 18516 + }, + { + "epoch": 2.355552728660476, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.832600593566895, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8922010660171509, + "num_tokens": 719042514.0, + "step": 18517 + }, + { + "epoch": 2.3556799389390664, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.74701976776123, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8904817700386047, + "num_tokens": 719077744.0, + "step": 18518 + }, + { + "epoch": 2.355807149217657, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.750643730163574, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.9007854461669922, + "num_tokens": 719116607.0, + "step": 18519 + }, + { + "epoch": 2.3559343594962474, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.559874534606934, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.9179750084877014, + "num_tokens": 719156282.0, + "step": 18520 + }, + { + "epoch": 2.356061569774838, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.779073715209961, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8879728317260742, + "num_tokens": 719193221.0, + "step": 18521 + }, + { + "epoch": 2.3561887800534285, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.662904739379883, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8970789909362793, + "num_tokens": 719236253.0, + "step": 18522 + }, + { + "epoch": 2.356315990332019, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.611115455627441, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.9000746607780457, + "num_tokens": 719282694.0, + "step": 18523 + }, + { + "epoch": 2.3564432006106095, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.680887222290039, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8970742225646973, + "num_tokens": 719325645.0, + "step": 18524 + }, + { + "epoch": 2.3565704108891996, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.715314865112305, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8982516527175903, + "num_tokens": 719362942.0, + "step": 18525 + }, + { + "epoch": 2.3566976211677906, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.892341613769531, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.898963451385498, + "num_tokens": 719405124.0, + "step": 18526 + }, + { + "epoch": 2.3568248314463807, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.622735977172852, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.9005617499351501, + "num_tokens": 719449007.0, + "step": 18527 + }, + { + "epoch": 2.356952041724971, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.668768882751465, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.9022305011749268, + "num_tokens": 719488862.0, + "step": 18528 + }, + { + "epoch": 2.3570792520035617, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.736745834350586, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9072995781898499, + "num_tokens": 719526259.0, + "step": 18529 + }, + { + "epoch": 2.3572064622821522, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.835742950439453, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8952924609184265, + "num_tokens": 719560556.0, + "step": 18530 + }, + { + "epoch": 2.3573336725607428, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.68998908996582, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9050273895263672, + "num_tokens": 719598072.0, + "step": 18531 + }, + { + "epoch": 2.3574608828393333, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.709609031677246, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9061467051506042, + "num_tokens": 719638028.0, + "step": 18532 + }, + { + "epoch": 2.357588093117924, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.71968936920166, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9043309688568115, + "num_tokens": 719671886.0, + "step": 18533 + }, + { + "epoch": 2.3577153033965144, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.739301681518555, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9012145400047302, + "num_tokens": 719705727.0, + "step": 18534 + }, + { + "epoch": 2.357842513675105, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.792040824890137, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8851102590560913, + "num_tokens": 719740611.0, + "step": 18535 + }, + { + "epoch": 2.3579697239536954, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.66514778137207, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.9006626605987549, + "num_tokens": 719782558.0, + "step": 18536 + }, + { + "epoch": 2.358096934232286, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.692587852478027, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.9054522514343262, + "num_tokens": 719826389.0, + "step": 18537 + }, + { + "epoch": 2.3582241445108765, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.70113468170166, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8958603143692017, + "num_tokens": 719869489.0, + "step": 18538 + }, + { + "epoch": 2.358351354789467, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.754669189453125, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8878254294395447, + "num_tokens": 719908651.0, + "step": 18539 + }, + { + "epoch": 2.3584785650680575, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.785360336303711, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8951707482337952, + "num_tokens": 719943542.0, + "step": 18540 + }, + { + "epoch": 2.358605775346648, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.683181762695312, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8997750878334045, + "num_tokens": 719985462.0, + "step": 18541 + }, + { + "epoch": 2.3587329856252386, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.86369514465332, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8794866800308228, + "num_tokens": 720018471.0, + "step": 18542 + }, + { + "epoch": 2.358860195903829, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.866963386535645, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.9034333229064941, + "num_tokens": 720049434.0, + "step": 18543 + }, + { + "epoch": 2.3589874061824196, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.602889060974121, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.9024454355239868, + "num_tokens": 720088271.0, + "step": 18544 + }, + { + "epoch": 2.35911461646101, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.672540664672852, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.9005746841430664, + "num_tokens": 720132080.0, + "step": 18545 + }, + { + "epoch": 2.3592418267396007, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.392333984375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.834867477416992, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9057159423828125, + "num_tokens": 720166204.0, + "step": 18546 + }, + { + "epoch": 2.3593690370181912, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.392333984375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.655349731445312, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.9069247841835022, + "num_tokens": 720205740.0, + "step": 18547 + }, + { + "epoch": 2.3594962472967818, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.392333984375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.699533462524414, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8895037174224854, + "num_tokens": 720248248.0, + "step": 18548 + }, + { + "epoch": 2.3596234575753723, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.392333984375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.75246810913086, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.892941951751709, + "num_tokens": 720285211.0, + "step": 18549 + }, + { + "epoch": 2.3597506678539624, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.392333984375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.782052040100098, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.898036003112793, + "num_tokens": 720323403.0, + "step": 18550 + }, + { + "epoch": 2.3598778781325533, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.392333984375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.676311492919922, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.903931736946106, + "num_tokens": 720360463.0, + "step": 18551 + }, + { + "epoch": 2.3600050884111434, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.392333984375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.718987464904785, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9082679748535156, + "num_tokens": 720397681.0, + "step": 18552 + }, + { + "epoch": 2.360132298689734, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.392333984375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.659834861755371, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.9014976620674133, + "num_tokens": 720438972.0, + "step": 18553 + }, + { + "epoch": 2.3602595089683245, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.733733177185059, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8907442092895508, + "num_tokens": 720478734.0, + "step": 18554 + }, + { + "epoch": 2.360386719246915, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.779657363891602, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8945491909980774, + "num_tokens": 720513007.0, + "step": 18555 + }, + { + "epoch": 2.3605139295255055, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.746150016784668, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8993370532989502, + "num_tokens": 720546612.0, + "step": 18556 + }, + { + "epoch": 2.360641139804096, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.723827362060547, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8951313495635986, + "num_tokens": 720590734.0, + "step": 18557 + }, + { + "epoch": 2.3607683500826866, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.70775318145752, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9045529961585999, + "num_tokens": 720626866.0, + "step": 18558 + }, + { + "epoch": 2.360895560361277, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.78740406036377, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.9043062925338745, + "num_tokens": 720663218.0, + "step": 18559 + }, + { + "epoch": 2.3610227706398677, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.73641300201416, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9034451246261597, + "num_tokens": 720700781.0, + "step": 18560 + }, + { + "epoch": 2.361149980918458, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.772221565246582, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.9062366485595703, + "num_tokens": 720731396.0, + "step": 18561 + }, + { + "epoch": 2.3612771911970487, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.721395492553711, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8953294157981873, + "num_tokens": 720767721.0, + "step": 18562 + }, + { + "epoch": 2.3614044014756392, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.75629997253418, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8854801058769226, + "num_tokens": 720807179.0, + "step": 18563 + }, + { + "epoch": 2.3615316117542298, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.633283615112305, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.9010821580886841, + "num_tokens": 720852330.0, + "step": 18564 + }, + { + "epoch": 2.3616588220328203, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.879720687866211, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.882102906703949, + "num_tokens": 720888537.0, + "step": 18565 + }, + { + "epoch": 2.361786032311411, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.695921897888184, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9027109146118164, + "num_tokens": 720924540.0, + "step": 18566 + }, + { + "epoch": 2.3619132425900013, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.712894439697266, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9008532762527466, + "num_tokens": 720963769.0, + "step": 18567 + }, + { + "epoch": 2.362040452868592, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.691187858581543, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9100614190101624, + "num_tokens": 720999281.0, + "step": 18568 + }, + { + "epoch": 2.3621676631471824, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.668377876281738, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.899499773979187, + "num_tokens": 721039463.0, + "step": 18569 + }, + { + "epoch": 2.362294873425773, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.897098541259766, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8925068974494934, + "num_tokens": 721072252.0, + "step": 18570 + }, + { + "epoch": 2.3624220837043635, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.621453285217285, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9077457785606384, + "num_tokens": 721111740.0, + "step": 18571 + }, + { + "epoch": 2.362549293982954, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.771814346313477, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8970431089401245, + "num_tokens": 721151376.0, + "step": 18572 + }, + { + "epoch": 2.3626765042615445, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.840559005737305, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8910871744155884, + "num_tokens": 721189676.0, + "step": 18573 + }, + { + "epoch": 2.362803714540135, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.748296737670898, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.9015630483627319, + "num_tokens": 721224077.0, + "step": 18574 + }, + { + "epoch": 2.362930924818725, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.697221755981445, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9086001515388489, + "num_tokens": 721262067.0, + "step": 18575 + }, + { + "epoch": 2.363058135097316, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.857059478759766, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.9025651812553406, + "num_tokens": 721292850.0, + "step": 18576 + }, + { + "epoch": 2.363185345375906, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.753332138061523, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8901302814483643, + "num_tokens": 721332742.0, + "step": 18577 + }, + { + "epoch": 2.3633125556544967, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.712484359741211, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8949023485183716, + "num_tokens": 721374330.0, + "step": 18578 + }, + { + "epoch": 2.3634397659330872, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.768056869506836, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8977251052856445, + "num_tokens": 721411722.0, + "step": 18579 + }, + { + "epoch": 2.3635669762116778, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 10.186246871948242, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8847230076789856, + "num_tokens": 721452314.0, + "step": 18580 + }, + { + "epoch": 2.3636941864902683, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.783053398132324, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8949079513549805, + "num_tokens": 721487882.0, + "step": 18581 + }, + { + "epoch": 2.363821396768859, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.64620304107666, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.9036901593208313, + "num_tokens": 721527900.0, + "step": 18582 + }, + { + "epoch": 2.3639486070474494, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.811387062072754, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9085195064544678, + "num_tokens": 721564616.0, + "step": 18583 + }, + { + "epoch": 2.36407581732604, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.737438201904297, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8967481851577759, + "num_tokens": 721600619.0, + "step": 18584 + }, + { + "epoch": 2.3642030276046304, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.817632675170898, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8950093984603882, + "num_tokens": 721637587.0, + "step": 18585 + }, + { + "epoch": 2.364330237883221, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.828763008117676, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8990808129310608, + "num_tokens": 721670962.0, + "step": 18586 + }, + { + "epoch": 2.3644574481618115, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.719815254211426, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8931539058685303, + "num_tokens": 721712148.0, + "step": 18587 + }, + { + "epoch": 2.364584658440402, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.899609565734863, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8936312198638916, + "num_tokens": 721751557.0, + "step": 18588 + }, + { + "epoch": 2.3647118687189925, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.66860294342041, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9103220105171204, + "num_tokens": 721789519.0, + "step": 18589 + }, + { + "epoch": 2.364839078997583, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.844120025634766, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.909899115562439, + "num_tokens": 721822776.0, + "step": 18590 + }, + { + "epoch": 2.3649662892761736, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.769230842590332, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8961749076843262, + "num_tokens": 721862065.0, + "step": 18591 + }, + { + "epoch": 2.365093499554764, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.692941665649414, + "learning_rate": 1e-06, + "loss": 0.2393, + "mean_token_accuracy": 0.9179232716560364, + "num_tokens": 721901331.0, + "step": 18592 + }, + { + "epoch": 2.3652207098333546, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.746516227722168, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8914594054222107, + "num_tokens": 721946205.0, + "step": 18593 + }, + { + "epoch": 2.365347920111945, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.760824203491211, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8999963998794556, + "num_tokens": 721983893.0, + "step": 18594 + }, + { + "epoch": 2.3654751303905357, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.912554740905762, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8927363753318787, + "num_tokens": 722019597.0, + "step": 18595 + }, + { + "epoch": 2.3656023406691262, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.786368370056152, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.9009412527084351, + "num_tokens": 722061861.0, + "step": 18596 + }, + { + "epoch": 2.3657295509477168, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.788082122802734, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8979324698448181, + "num_tokens": 722098204.0, + "step": 18597 + }, + { + "epoch": 2.365856761226307, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.695575714111328, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8966811895370483, + "num_tokens": 722137386.0, + "step": 18598 + }, + { + "epoch": 2.365983971504898, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.708340644836426, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9119299650192261, + "num_tokens": 722178486.0, + "step": 18599 + }, + { + "epoch": 2.366111181783488, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.657978057861328, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.909801721572876, + "num_tokens": 722218687.0, + "step": 18600 + }, + { + "epoch": 2.3662383920620784, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.677207946777344, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8947395086288452, + "num_tokens": 722263154.0, + "step": 18601 + }, + { + "epoch": 2.366365602340669, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.737969398498535, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8963915109634399, + "num_tokens": 722304423.0, + "step": 18602 + }, + { + "epoch": 2.3664928126192595, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.644678115844727, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.900907039642334, + "num_tokens": 722349559.0, + "step": 18603 + }, + { + "epoch": 2.36662002289785, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.69350814819336, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8968690633773804, + "num_tokens": 722391702.0, + "step": 18604 + }, + { + "epoch": 2.3667472331764405, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.729957580566406, + "learning_rate": 1e-06, + "loss": 0.2438, + "mean_token_accuracy": 0.9175058603286743, + "num_tokens": 722427457.0, + "step": 18605 + }, + { + "epoch": 2.366874443455031, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.750483512878418, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.908667802810669, + "num_tokens": 722461699.0, + "step": 18606 + }, + { + "epoch": 2.3670016537336216, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.726977348327637, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9116246700286865, + "num_tokens": 722501773.0, + "step": 18607 + }, + { + "epoch": 2.367128864012212, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.73873519897461, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9106266498565674, + "num_tokens": 722537767.0, + "step": 18608 + }, + { + "epoch": 2.3672560742908026, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.738517761230469, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8866322040557861, + "num_tokens": 722577586.0, + "step": 18609 + }, + { + "epoch": 2.367383284569393, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.661481857299805, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9006955027580261, + "num_tokens": 722621684.0, + "step": 18610 + }, + { + "epoch": 2.3675104948479837, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.804706573486328, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8879073262214661, + "num_tokens": 722657038.0, + "step": 18611 + }, + { + "epoch": 2.3676377051265742, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.800204277038574, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8967509269714355, + "num_tokens": 722695368.0, + "step": 18612 + }, + { + "epoch": 2.3677649154051648, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.699759483337402, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.9029537439346313, + "num_tokens": 722732362.0, + "step": 18613 + }, + { + "epoch": 2.3678921256837553, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.805423736572266, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9047755002975464, + "num_tokens": 722766920.0, + "step": 18614 + }, + { + "epoch": 2.368019335962346, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.731209754943848, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9024134278297424, + "num_tokens": 722804939.0, + "step": 18615 + }, + { + "epoch": 2.3681465462409363, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.699670791625977, + "learning_rate": 1e-06, + "loss": 0.2565, + "mean_token_accuracy": 0.9117901921272278, + "num_tokens": 722845624.0, + "step": 18616 + }, + { + "epoch": 2.368273756519527, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.767230987548828, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8983964323997498, + "num_tokens": 722882289.0, + "step": 18617 + }, + { + "epoch": 2.3684009667981174, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.66904354095459, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8938910961151123, + "num_tokens": 722930824.0, + "step": 18618 + }, + { + "epoch": 2.368528177076708, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.853099822998047, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.9003844261169434, + "num_tokens": 722963716.0, + "step": 18619 + }, + { + "epoch": 2.3686553873552985, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.796809196472168, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9041745662689209, + "num_tokens": 722998200.0, + "step": 18620 + }, + { + "epoch": 2.368782597633889, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.71373176574707, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8945654630661011, + "num_tokens": 723035810.0, + "step": 18621 + }, + { + "epoch": 2.3689098079124795, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.77303695678711, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8883275389671326, + "num_tokens": 723079632.0, + "step": 18622 + }, + { + "epoch": 2.3690370181910696, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.700933456420898, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.9016493558883667, + "num_tokens": 723119838.0, + "step": 18623 + }, + { + "epoch": 2.3691642284696606, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.81242847442627, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8982135653495789, + "num_tokens": 723155986.0, + "step": 18624 + }, + { + "epoch": 2.3692914387482507, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.714157104492188, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.903498113155365, + "num_tokens": 723196736.0, + "step": 18625 + }, + { + "epoch": 2.369418649026841, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.75137710571289, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8990356922149658, + "num_tokens": 723233085.0, + "step": 18626 + }, + { + "epoch": 2.3695458593054317, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.882421493530273, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8901167511940002, + "num_tokens": 723271517.0, + "step": 18627 + }, + { + "epoch": 2.3696730695840222, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.731060981750488, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8958852291107178, + "num_tokens": 723309523.0, + "step": 18628 + }, + { + "epoch": 2.3698002798626128, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.661273002624512, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8943092823028564, + "num_tokens": 723349432.0, + "step": 18629 + }, + { + "epoch": 2.3699274901412033, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.689207077026367, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8970268964767456, + "num_tokens": 723389265.0, + "step": 18630 + }, + { + "epoch": 2.370054700419794, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.733112335205078, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8894777297973633, + "num_tokens": 723431616.0, + "step": 18631 + }, + { + "epoch": 2.3701819106983844, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.733026504516602, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8983842134475708, + "num_tokens": 723468905.0, + "step": 18632 + }, + { + "epoch": 2.370309120976975, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.748957633972168, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9048770666122437, + "num_tokens": 723504925.0, + "step": 18633 + }, + { + "epoch": 2.3704363312555654, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.832072257995605, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8971474170684814, + "num_tokens": 723540500.0, + "step": 18634 + }, + { + "epoch": 2.370563541534156, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.879980087280273, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8844675421714783, + "num_tokens": 723575455.0, + "step": 18635 + }, + { + "epoch": 2.3706907518127465, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.818033218383789, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8933880925178528, + "num_tokens": 723613302.0, + "step": 18636 + }, + { + "epoch": 2.370817962091337, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.779508590698242, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.9022715091705322, + "num_tokens": 723649950.0, + "step": 18637 + }, + { + "epoch": 2.3709451723699275, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.760151863098145, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8925908207893372, + "num_tokens": 723689303.0, + "step": 18638 + }, + { + "epoch": 2.371072382648518, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.733478546142578, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.9023301601409912, + "num_tokens": 723728197.0, + "step": 18639 + }, + { + "epoch": 2.3711995929271086, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.66947078704834, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8888423442840576, + "num_tokens": 723775274.0, + "step": 18640 + }, + { + "epoch": 2.371326803205699, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.810684204101562, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8988783359527588, + "num_tokens": 723815258.0, + "step": 18641 + }, + { + "epoch": 2.3714540134842896, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.714682579040527, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.9010157585144043, + "num_tokens": 723859458.0, + "step": 18642 + }, + { + "epoch": 2.37158122376288, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.732522010803223, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.906192421913147, + "num_tokens": 723896263.0, + "step": 18643 + }, + { + "epoch": 2.3717084340414707, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.822676658630371, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8781689405441284, + "num_tokens": 723937125.0, + "step": 18644 + }, + { + "epoch": 2.371835644320061, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.745675086975098, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.9023723602294922, + "num_tokens": 723974914.0, + "step": 18645 + }, + { + "epoch": 2.3719628545986517, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.888509750366211, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8947509527206421, + "num_tokens": 724012524.0, + "step": 18646 + }, + { + "epoch": 2.3720900648772423, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.726048469543457, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9111844301223755, + "num_tokens": 724050312.0, + "step": 18647 + }, + { + "epoch": 2.3722172751558324, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.726223945617676, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8904789090156555, + "num_tokens": 724095374.0, + "step": 18648 + }, + { + "epoch": 2.3723444854344233, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.658281326293945, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9059685468673706, + "num_tokens": 724136242.0, + "step": 18649 + }, + { + "epoch": 2.3724716957130134, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.749966621398926, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8882805109024048, + "num_tokens": 724179220.0, + "step": 18650 + }, + { + "epoch": 2.372598905991604, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.783080101013184, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8994789123535156, + "num_tokens": 724214990.0, + "step": 18651 + }, + { + "epoch": 2.3727261162701945, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.767521858215332, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9114434123039246, + "num_tokens": 724250337.0, + "step": 18652 + }, + { + "epoch": 2.372853326548785, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.677340507507324, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8872385025024414, + "num_tokens": 724290471.0, + "step": 18653 + }, + { + "epoch": 2.3729805368273755, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.845742225646973, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.9013758301734924, + "num_tokens": 724327863.0, + "step": 18654 + }, + { + "epoch": 2.373107747105966, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.801973342895508, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8929784297943115, + "num_tokens": 724365786.0, + "step": 18655 + }, + { + "epoch": 2.3732349573845566, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.761992454528809, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.9050657749176025, + "num_tokens": 724400515.0, + "step": 18656 + }, + { + "epoch": 2.373362167663147, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.72261905670166, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.9001783132553101, + "num_tokens": 724441852.0, + "step": 18657 + }, + { + "epoch": 2.3734893779417376, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.683947563171387, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8934462666511536, + "num_tokens": 724485790.0, + "step": 18658 + }, + { + "epoch": 2.373616588220328, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.678020477294922, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9047609567642212, + "num_tokens": 724533434.0, + "step": 18659 + }, + { + "epoch": 2.3737437984989187, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.72380256652832, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8956615924835205, + "num_tokens": 724575391.0, + "step": 18660 + }, + { + "epoch": 2.3738710087775092, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.741131782531738, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.9012776613235474, + "num_tokens": 724613299.0, + "step": 18661 + }, + { + "epoch": 2.3739982190560998, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.765032768249512, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9146387577056885, + "num_tokens": 724648492.0, + "step": 18662 + }, + { + "epoch": 2.3741254293346903, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.722702980041504, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9084537029266357, + "num_tokens": 724686253.0, + "step": 18663 + }, + { + "epoch": 2.374252639613281, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.762615203857422, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8914192914962769, + "num_tokens": 724725543.0, + "step": 18664 + }, + { + "epoch": 2.3743798498918713, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.751497268676758, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9010056257247925, + "num_tokens": 724762454.0, + "step": 18665 + }, + { + "epoch": 2.374507060170462, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.861316680908203, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8891041278839111, + "num_tokens": 724797109.0, + "step": 18666 + }, + { + "epoch": 2.3746342704490524, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.760275840759277, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.894405722618103, + "num_tokens": 724831723.0, + "step": 18667 + }, + { + "epoch": 2.374761480727643, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.697154998779297, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.9012081623077393, + "num_tokens": 724871061.0, + "step": 18668 + }, + { + "epoch": 2.3748886910062335, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.673784255981445, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.906588077545166, + "num_tokens": 724909699.0, + "step": 18669 + }, + { + "epoch": 2.375015901284824, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.843356132507324, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8912760019302368, + "num_tokens": 724948822.0, + "step": 18670 + }, + { + "epoch": 2.3751431115634145, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.803603172302246, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8952820301055908, + "num_tokens": 724987017.0, + "step": 18671 + }, + { + "epoch": 2.375270321842005, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.72333812713623, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8811283111572266, + "num_tokens": 725032406.0, + "step": 18672 + }, + { + "epoch": 2.375397532120595, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.698046684265137, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8997877836227417, + "num_tokens": 725071820.0, + "step": 18673 + }, + { + "epoch": 2.375524742399186, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.824471473693848, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9042950868606567, + "num_tokens": 725104939.0, + "step": 18674 + }, + { + "epoch": 2.375651952677776, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.673412322998047, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8931549787521362, + "num_tokens": 725144015.0, + "step": 18675 + }, + { + "epoch": 2.3757791629563667, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.739883422851562, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.9029271602630615, + "num_tokens": 725178935.0, + "step": 18676 + }, + { + "epoch": 2.3759063732349572, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.700057983398438, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9046138525009155, + "num_tokens": 725219297.0, + "step": 18677 + }, + { + "epoch": 2.3760335835135478, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.911456108093262, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.8753010034561157, + "num_tokens": 725256076.0, + "step": 18678 + }, + { + "epoch": 2.3761607937921383, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.619963645935059, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9094290733337402, + "num_tokens": 725294899.0, + "step": 18679 + }, + { + "epoch": 2.376288004070729, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.647491455078125, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.897646427154541, + "num_tokens": 725336557.0, + "step": 18680 + }, + { + "epoch": 2.3764152143493193, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.7074556350708, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9057390093803406, + "num_tokens": 725377777.0, + "step": 18681 + }, + { + "epoch": 2.37654242462791, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.678586959838867, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8991466760635376, + "num_tokens": 725421724.0, + "step": 18682 + }, + { + "epoch": 2.3766696349065004, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.815299987792969, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8989213109016418, + "num_tokens": 725458942.0, + "step": 18683 + }, + { + "epoch": 2.376796845185091, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.781128883361816, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.876966655254364, + "num_tokens": 725498986.0, + "step": 18684 + }, + { + "epoch": 2.3769240554636815, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.692130088806152, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.9050894975662231, + "num_tokens": 725541349.0, + "step": 18685 + }, + { + "epoch": 2.377051265742272, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.651447296142578, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9037894606590271, + "num_tokens": 725579082.0, + "step": 18686 + }, + { + "epoch": 2.3771784760208625, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.682336807250977, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.9043436646461487, + "num_tokens": 725623082.0, + "step": 18687 + }, + { + "epoch": 2.377305686299453, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.687950134277344, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8929500579833984, + "num_tokens": 725670817.0, + "step": 18688 + }, + { + "epoch": 2.3774328965780436, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.696355819702148, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.9032979607582092, + "num_tokens": 725710124.0, + "step": 18689 + }, + { + "epoch": 2.377560106856634, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.7377347946167, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8983449935913086, + "num_tokens": 725752246.0, + "step": 18690 + }, + { + "epoch": 2.3776873171352246, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.693839073181152, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8909257650375366, + "num_tokens": 725793034.0, + "step": 18691 + }, + { + "epoch": 2.377814527413815, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.779568672180176, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8910080790519714, + "num_tokens": 725834699.0, + "step": 18692 + }, + { + "epoch": 2.3779417376924057, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.673652648925781, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8985819816589355, + "num_tokens": 725878748.0, + "step": 18693 + }, + { + "epoch": 2.378068947970996, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.726716041564941, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8964584469795227, + "num_tokens": 725918285.0, + "step": 18694 + }, + { + "epoch": 2.3781961582495867, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.751266479492188, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9023774862289429, + "num_tokens": 725961456.0, + "step": 18695 + }, + { + "epoch": 2.378323368528177, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.795621871948242, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8976694345474243, + "num_tokens": 725999971.0, + "step": 18696 + }, + { + "epoch": 2.378450578806768, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.714349746704102, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8948552012443542, + "num_tokens": 726044550.0, + "step": 18697 + }, + { + "epoch": 2.378577789085358, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.747291564941406, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9026623964309692, + "num_tokens": 726077325.0, + "step": 18698 + }, + { + "epoch": 2.3787049993639484, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.70151424407959, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8983361124992371, + "num_tokens": 726120878.0, + "step": 18699 + }, + { + "epoch": 2.378832209642539, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.892104148864746, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8911446332931519, + "num_tokens": 726153288.0, + "step": 18700 + }, + { + "epoch": 2.3789594199211295, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.872684478759766, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8950719237327576, + "num_tokens": 726185230.0, + "step": 18701 + }, + { + "epoch": 2.37908663019972, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.686945915222168, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8970988988876343, + "num_tokens": 726224394.0, + "step": 18702 + }, + { + "epoch": 2.3792138404783105, + "ewc_loss": 0.0198974609375, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.150369644165039e-05, + "grad_norm": 9.66284465789795, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8831958770751953, + "num_tokens": 726268423.0, + "step": 18703 + }, + { + "epoch": 2.379341050756901, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.74759292602539, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.9019245505332947, + "num_tokens": 726310977.0, + "step": 18704 + }, + { + "epoch": 2.3794682610354916, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.995879173278809, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8938730955123901, + "num_tokens": 726358236.0, + "step": 18705 + }, + { + "epoch": 2.379595471314082, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.769455909729004, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.9025445580482483, + "num_tokens": 726396387.0, + "step": 18706 + }, + { + "epoch": 2.3797226815926726, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.656584739685059, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8945467472076416, + "num_tokens": 726437215.0, + "step": 18707 + }, + { + "epoch": 2.379849891871263, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.689916610717773, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.900185227394104, + "num_tokens": 726480055.0, + "step": 18708 + }, + { + "epoch": 2.3799771021498537, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.591858863830566, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.911566972732544, + "num_tokens": 726522127.0, + "step": 18709 + }, + { + "epoch": 2.3801043124284442, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.791666984558105, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8978742361068726, + "num_tokens": 726558063.0, + "step": 18710 + }, + { + "epoch": 2.3802315227070348, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.684272766113281, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9077896475791931, + "num_tokens": 726599802.0, + "step": 18711 + }, + { + "epoch": 2.3803587329856253, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.661676406860352, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8976141810417175, + "num_tokens": 726642087.0, + "step": 18712 + }, + { + "epoch": 2.380485943264216, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.768543243408203, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8950986862182617, + "num_tokens": 726680372.0, + "step": 18713 + }, + { + "epoch": 2.3806131535428063, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.76697063446045, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8987716436386108, + "num_tokens": 726717658.0, + "step": 18714 + }, + { + "epoch": 2.380740363821397, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.764680862426758, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9054861664772034, + "num_tokens": 726755042.0, + "step": 18715 + }, + { + "epoch": 2.3808675740999874, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.822442054748535, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8871042728424072, + "num_tokens": 726796560.0, + "step": 18716 + }, + { + "epoch": 2.380994784378578, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.734726905822754, + "learning_rate": 1e-06, + "loss": 0.2591, + "mean_token_accuracy": 0.9105821251869202, + "num_tokens": 726830891.0, + "step": 18717 + }, + { + "epoch": 2.3811219946571685, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.67805004119873, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.9021816849708557, + "num_tokens": 726872527.0, + "step": 18718 + }, + { + "epoch": 2.381249204935759, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.838926315307617, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8911958336830139, + "num_tokens": 726912342.0, + "step": 18719 + }, + { + "epoch": 2.3813764152143495, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.440017700195312e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.689095497131348, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.9022727012634277, + "num_tokens": 726951482.0, + "step": 18720 + }, + { + "epoch": 2.3815036254929396, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.697134971618652, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8980355262756348, + "num_tokens": 726991003.0, + "step": 18721 + }, + { + "epoch": 2.3816308357715306, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.736550331115723, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8968772888183594, + "num_tokens": 727033206.0, + "step": 18722 + }, + { + "epoch": 2.3817580460501206, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.6704740524292, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9110620617866516, + "num_tokens": 727072512.0, + "step": 18723 + }, + { + "epoch": 2.381885256328711, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.730785369873047, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8914646506309509, + "num_tokens": 727114353.0, + "step": 18724 + }, + { + "epoch": 2.3820124666073017, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.772015571594238, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8952512741088867, + "num_tokens": 727153220.0, + "step": 18725 + }, + { + "epoch": 2.3821396768858922, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.63195514678955, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9103509187698364, + "num_tokens": 727193391.0, + "step": 18726 + }, + { + "epoch": 2.3822668871644828, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.80577564239502, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8987007141113281, + "num_tokens": 727235224.0, + "step": 18727 + }, + { + "epoch": 2.3823940974430733, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.674315452575684, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8943817615509033, + "num_tokens": 727275191.0, + "step": 18728 + }, + { + "epoch": 2.382521307721664, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.679248809814453, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8933042287826538, + "num_tokens": 727316655.0, + "step": 18729 + }, + { + "epoch": 2.3826485180002543, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.724279403686523, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8942071199417114, + "num_tokens": 727358041.0, + "step": 18730 + }, + { + "epoch": 2.382775728278845, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.631495475769043, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.9017969965934753, + "num_tokens": 727403420.0, + "step": 18731 + }, + { + "epoch": 2.3829029385574354, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1563301086425781e-05, + "grad_norm": 9.78381633758545, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.907784104347229, + "num_tokens": 727438884.0, + "step": 18732 + }, + { + "epoch": 2.383030148836026, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.752725601196289, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8928625583648682, + "num_tokens": 727475638.0, + "step": 18733 + }, + { + "epoch": 2.3831573591146165, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.758339881896973, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9019540548324585, + "num_tokens": 727511610.0, + "step": 18734 + }, + { + "epoch": 2.383284569393207, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.840868949890137, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8987205624580383, + "num_tokens": 727542300.0, + "step": 18735 + }, + { + "epoch": 2.3834117796717975, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.77949047088623, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9114016890525818, + "num_tokens": 727576930.0, + "step": 18736 + }, + { + "epoch": 2.383538989950388, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.736534118652344, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.9014860391616821, + "num_tokens": 727617569.0, + "step": 18737 + }, + { + "epoch": 2.3836662002289786, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.81990909576416, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8841991424560547, + "num_tokens": 727655480.0, + "step": 18738 + }, + { + "epoch": 2.383793410507569, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.750547409057617, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.898775577545166, + "num_tokens": 727696020.0, + "step": 18739 + }, + { + "epoch": 2.3839206207861596, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.757705688476562, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8926794528961182, + "num_tokens": 727733364.0, + "step": 18740 + }, + { + "epoch": 2.38404783106475, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.734090805053711, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9088754057884216, + "num_tokens": 727771727.0, + "step": 18741 + }, + { + "epoch": 2.3841750413433407, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.772289276123047, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.900682270526886, + "num_tokens": 727812636.0, + "step": 18742 + }, + { + "epoch": 2.384302251621931, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.693507194519043, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8998558521270752, + "num_tokens": 727854055.0, + "step": 18743 + }, + { + "epoch": 2.3844294619005217, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.752120971679688, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9073314666748047, + "num_tokens": 727891311.0, + "step": 18744 + }, + { + "epoch": 2.3845566721791123, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.693426132202148, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9111368656158447, + "num_tokens": 727937244.0, + "step": 18745 + }, + { + "epoch": 2.3846838824577024, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.731566429138184, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.906089723110199, + "num_tokens": 727976596.0, + "step": 18746 + }, + { + "epoch": 2.3848110927362933, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.841068267822266, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.9062414765357971, + "num_tokens": 728011109.0, + "step": 18747 + }, + { + "epoch": 2.3849383030148834, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.769964218139648, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8906296491622925, + "num_tokens": 728052085.0, + "step": 18748 + }, + { + "epoch": 2.385065513293474, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.700135231018066, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9111610651016235, + "num_tokens": 728088751.0, + "step": 18749 + }, + { + "epoch": 2.3851927235720645, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.717582702636719, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9100784063339233, + "num_tokens": 728127069.0, + "step": 18750 + }, + { + "epoch": 2.385319933850655, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.720364570617676, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.9050009250640869, + "num_tokens": 728169401.0, + "step": 18751 + }, + { + "epoch": 2.3854471441292455, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.801291465759277, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.9020805358886719, + "num_tokens": 728204433.0, + "step": 18752 + }, + { + "epoch": 2.385574354407836, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.719576835632324, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8974913358688354, + "num_tokens": 728244014.0, + "step": 18753 + }, + { + "epoch": 2.3857015646864266, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.837252616882324, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.9000084400177002, + "num_tokens": 728279449.0, + "step": 18754 + }, + { + "epoch": 2.385828774965017, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.753925323486328, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8906707763671875, + "num_tokens": 728315897.0, + "step": 18755 + }, + { + "epoch": 2.3859559852436076, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.808572769165039, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8812043070793152, + "num_tokens": 728356021.0, + "step": 18756 + }, + { + "epoch": 2.386083195522198, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.850135803222656, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8857864141464233, + "num_tokens": 728396923.0, + "step": 18757 + }, + { + "epoch": 2.3862104058007887, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.760523796081543, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8933817148208618, + "num_tokens": 728437698.0, + "step": 18758 + }, + { + "epoch": 2.386337616079379, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.746820449829102, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8870974779129028, + "num_tokens": 728478117.0, + "step": 18759 + }, + { + "epoch": 2.3864648263579697, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.873662948608398, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8902748823165894, + "num_tokens": 728515806.0, + "step": 18760 + }, + { + "epoch": 2.3865920366365603, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.786969184875488, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8988713026046753, + "num_tokens": 728548068.0, + "step": 18761 + }, + { + "epoch": 2.386719246915151, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.673758506774902, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.900278627872467, + "num_tokens": 728589094.0, + "step": 18762 + }, + { + "epoch": 2.3868464571937413, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.634526252746582, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9032386541366577, + "num_tokens": 728630933.0, + "step": 18763 + }, + { + "epoch": 2.386973667472332, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.78172779083252, + "learning_rate": 1e-06, + "loss": 0.2432, + "mean_token_accuracy": 0.9159867763519287, + "num_tokens": 728662734.0, + "step": 18764 + }, + { + "epoch": 2.3871008777509224, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.825884819030762, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8903346657752991, + "num_tokens": 728703456.0, + "step": 18765 + }, + { + "epoch": 2.387228088029513, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.692709922790527, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8928154110908508, + "num_tokens": 728746207.0, + "step": 18766 + }, + { + "epoch": 2.3873552983081034, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.728798866271973, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.896465539932251, + "num_tokens": 728786325.0, + "step": 18767 + }, + { + "epoch": 2.387482508586694, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.7566556930542, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9033327102661133, + "num_tokens": 728824345.0, + "step": 18768 + }, + { + "epoch": 2.3876097188652845, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.7465181350708, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8921456336975098, + "num_tokens": 728867644.0, + "step": 18769 + }, + { + "epoch": 2.387736929143875, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.763044357299805, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8985913395881653, + "num_tokens": 728902460.0, + "step": 18770 + }, + { + "epoch": 2.387864139422465, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.841473579406738, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.903964638710022, + "num_tokens": 728936008.0, + "step": 18771 + }, + { + "epoch": 2.387991349701056, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.804384231567383, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8992476463317871, + "num_tokens": 728972917.0, + "step": 18772 + }, + { + "epoch": 2.388118559979646, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.752466201782227, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8916869163513184, + "num_tokens": 729011081.0, + "step": 18773 + }, + { + "epoch": 2.3882457702582367, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.653691291809082, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9112286567687988, + "num_tokens": 729051841.0, + "step": 18774 + }, + { + "epoch": 2.3883729805368272, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.714241027832031, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8906661868095398, + "num_tokens": 729089957.0, + "step": 18775 + }, + { + "epoch": 2.3885001908154178, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.749183654785156, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.9010956883430481, + "num_tokens": 729129094.0, + "step": 18776 + }, + { + "epoch": 2.3886274010940083, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.694669723510742, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.9010962247848511, + "num_tokens": 729162574.0, + "step": 18777 + }, + { + "epoch": 2.388754611372599, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.839874267578125, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.897517204284668, + "num_tokens": 729199041.0, + "step": 18778 + }, + { + "epoch": 2.3888818216511893, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.79480266571045, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8933241367340088, + "num_tokens": 729235453.0, + "step": 18779 + }, + { + "epoch": 2.38900903192978, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.694591522216797, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8960233330726624, + "num_tokens": 729277847.0, + "step": 18780 + }, + { + "epoch": 2.3891362422083704, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.788411140441895, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8999621272087097, + "num_tokens": 729314498.0, + "step": 18781 + }, + { + "epoch": 2.389263452486961, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.780444145202637, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8902663588523865, + "num_tokens": 729356802.0, + "step": 18782 + }, + { + "epoch": 2.3893906627655515, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.732199668884277, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9061909914016724, + "num_tokens": 729399803.0, + "step": 18783 + }, + { + "epoch": 2.389517873044142, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.70865535736084, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.9014492034912109, + "num_tokens": 729442371.0, + "step": 18784 + }, + { + "epoch": 2.3896450833227325, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.910628318786621, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8781575560569763, + "num_tokens": 729483082.0, + "step": 18785 + }, + { + "epoch": 2.389772293601323, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.739302635192871, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.9033222794532776, + "num_tokens": 729517559.0, + "step": 18786 + }, + { + "epoch": 2.3898995038799136, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.837885856628418, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8986911177635193, + "num_tokens": 729555055.0, + "step": 18787 + }, + { + "epoch": 2.390026714158504, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.758382797241211, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.9014378190040588, + "num_tokens": 729588026.0, + "step": 18788 + }, + { + "epoch": 2.3901539244370946, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.7642822265625, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.9000119566917419, + "num_tokens": 729626750.0, + "step": 18789 + }, + { + "epoch": 2.390281134715685, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.772254943847656, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8981125354766846, + "num_tokens": 729666308.0, + "step": 18790 + }, + { + "epoch": 2.3904083449942757, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.824040412902832, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8876104950904846, + "num_tokens": 729705953.0, + "step": 18791 + }, + { + "epoch": 2.390535555272866, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.780842781066895, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8998330235481262, + "num_tokens": 729746061.0, + "step": 18792 + }, + { + "epoch": 2.3906627655514567, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.807804107666016, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8797818422317505, + "num_tokens": 729788212.0, + "step": 18793 + }, + { + "epoch": 2.390789975830047, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.729129791259766, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.9019383192062378, + "num_tokens": 729825050.0, + "step": 18794 + }, + { + "epoch": 2.390917186108638, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.711356163024902, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.9001467227935791, + "num_tokens": 729864454.0, + "step": 18795 + }, + { + "epoch": 2.391044396387228, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.825748443603516, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.887319803237915, + "num_tokens": 729904388.0, + "step": 18796 + }, + { + "epoch": 2.3911716066658184, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.717490196228027, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.911104679107666, + "num_tokens": 729938799.0, + "step": 18797 + }, + { + "epoch": 2.391298816944409, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.777932167053223, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8940128087997437, + "num_tokens": 729979020.0, + "step": 18798 + }, + { + "epoch": 2.3914260272229995, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.747369766235352, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.88872230052948, + "num_tokens": 730022118.0, + "step": 18799 + }, + { + "epoch": 2.39155323750159, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.734058380126953, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8912001848220825, + "num_tokens": 730064366.0, + "step": 18800 + }, + { + "epoch": 2.3916804477801805, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.648773193359375, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9120481014251709, + "num_tokens": 730104002.0, + "step": 18801 + }, + { + "epoch": 2.391807658058771, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.811657905578613, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8865970373153687, + "num_tokens": 730140392.0, + "step": 18802 + }, + { + "epoch": 2.3919348683373616, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.791496276855469, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8942160606384277, + "num_tokens": 730176918.0, + "step": 18803 + }, + { + "epoch": 2.392062078615952, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.688101768493652, + "learning_rate": 1e-06, + "loss": 0.2565, + "mean_token_accuracy": 0.9121989607810974, + "num_tokens": 730214457.0, + "step": 18804 + }, + { + "epoch": 2.3921892888945426, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.723609924316406, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8919591903686523, + "num_tokens": 730252257.0, + "step": 18805 + }, + { + "epoch": 2.392316499173133, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.709182739257812, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9096201658248901, + "num_tokens": 730287673.0, + "step": 18806 + }, + { + "epoch": 2.3924437094517237, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.774374008178711, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8962870836257935, + "num_tokens": 730322561.0, + "step": 18807 + }, + { + "epoch": 2.392570919730314, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.73568058013916, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8934414386749268, + "num_tokens": 730363336.0, + "step": 18808 + }, + { + "epoch": 2.3926981300089047, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.667062759399414, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9086282849311829, + "num_tokens": 730402872.0, + "step": 18809 + }, + { + "epoch": 2.3928253402874953, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.835286140441895, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.898047924041748, + "num_tokens": 730437472.0, + "step": 18810 + }, + { + "epoch": 2.392952550566086, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.670426368713379, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.913550078868866, + "num_tokens": 730473494.0, + "step": 18811 + }, + { + "epoch": 2.3930797608446763, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.812416076660156, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8984675407409668, + "num_tokens": 730510578.0, + "step": 18812 + }, + { + "epoch": 2.393206971123267, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.83769416809082, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8864012956619263, + "num_tokens": 730550088.0, + "step": 18813 + }, + { + "epoch": 2.3933341814018574, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.631243705749512, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9120782613754272, + "num_tokens": 730591472.0, + "step": 18814 + }, + { + "epoch": 2.393461391680448, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.857147216796875, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8942746520042419, + "num_tokens": 730627630.0, + "step": 18815 + }, + { + "epoch": 2.3935886019590384, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.803757667541504, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8933365941047668, + "num_tokens": 730664015.0, + "step": 18816 + }, + { + "epoch": 2.393715812237629, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.703863143920898, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.9005271196365356, + "num_tokens": 730704243.0, + "step": 18817 + }, + { + "epoch": 2.3938430225162195, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.684649467468262, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8980251550674438, + "num_tokens": 730746667.0, + "step": 18818 + }, + { + "epoch": 2.3939702327948096, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.69878101348877, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8982180953025818, + "num_tokens": 730791678.0, + "step": 18819 + }, + { + "epoch": 2.3940974430734006, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.817596435546875, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.9033390283584595, + "num_tokens": 730826102.0, + "step": 18820 + }, + { + "epoch": 2.3942246533519906, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.750617980957031, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8900691270828247, + "num_tokens": 730867613.0, + "step": 18821 + }, + { + "epoch": 2.394351863630581, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.708927154541016, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.9007354378700256, + "num_tokens": 730901787.0, + "step": 18822 + }, + { + "epoch": 2.3944790739091717, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.795027732849121, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8915517926216125, + "num_tokens": 730941749.0, + "step": 18823 + }, + { + "epoch": 2.3946062841877622, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.74110221862793, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8979461193084717, + "num_tokens": 730982733.0, + "step": 18824 + }, + { + "epoch": 2.3947334944663528, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.66189956665039, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.9043810367584229, + "num_tokens": 731025967.0, + "step": 18825 + }, + { + "epoch": 2.3948607047449433, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.740888595581055, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.9016662836074829, + "num_tokens": 731066916.0, + "step": 18826 + }, + { + "epoch": 2.394987915023534, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.645397186279297, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9085875749588013, + "num_tokens": 731111839.0, + "step": 18827 + }, + { + "epoch": 2.3951151253021243, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.739794731140137, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.9008650183677673, + "num_tokens": 731150567.0, + "step": 18828 + }, + { + "epoch": 2.395242335580715, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.882843971252441, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8803398013114929, + "num_tokens": 731190824.0, + "step": 18829 + }, + { + "epoch": 2.3953695458593054, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.721672058105469, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9054132699966431, + "num_tokens": 731227714.0, + "step": 18830 + }, + { + "epoch": 2.395496756137896, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.795662879943848, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8993121385574341, + "num_tokens": 731262929.0, + "step": 18831 + }, + { + "epoch": 2.3956239664164865, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.63813591003418, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8916773796081543, + "num_tokens": 731308594.0, + "step": 18832 + }, + { + "epoch": 2.395751176695077, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.727161407470703, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8945726752281189, + "num_tokens": 731349782.0, + "step": 18833 + }, + { + "epoch": 2.3958783869736675, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.755316734313965, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.9017429351806641, + "num_tokens": 731390810.0, + "step": 18834 + }, + { + "epoch": 2.396005597252258, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.849466323852539, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8919041156768799, + "num_tokens": 731422856.0, + "step": 18835 + }, + { + "epoch": 2.3961328075308486, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.7439546585083, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8985971212387085, + "num_tokens": 731458595.0, + "step": 18836 + }, + { + "epoch": 2.396260017809439, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.923192024230957, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8925890922546387, + "num_tokens": 731491172.0, + "step": 18837 + }, + { + "epoch": 2.3963872280880296, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.786235809326172, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.9048762321472168, + "num_tokens": 731526106.0, + "step": 18838 + }, + { + "epoch": 2.39651443836662, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.74499225616455, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9029511213302612, + "num_tokens": 731562875.0, + "step": 18839 + }, + { + "epoch": 2.3966416486452107, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.744623184204102, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.917202353477478, + "num_tokens": 731595151.0, + "step": 18840 + }, + { + "epoch": 2.396768858923801, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.701642036437988, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9079240560531616, + "num_tokens": 731634099.0, + "step": 18841 + }, + { + "epoch": 2.3968960692023917, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.768047332763672, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8901782631874084, + "num_tokens": 731676787.0, + "step": 18842 + }, + { + "epoch": 2.3970232794809823, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.594100952148438, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9122655391693115, + "num_tokens": 731715009.0, + "step": 18843 + }, + { + "epoch": 2.3971504897595723, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.845409393310547, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8959475755691528, + "num_tokens": 731749245.0, + "step": 18844 + }, + { + "epoch": 2.3972777000381633, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.731895446777344, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8999107480049133, + "num_tokens": 731791526.0, + "step": 18845 + }, + { + "epoch": 2.3974049103167534, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.721579551696777, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8927559852600098, + "num_tokens": 731834436.0, + "step": 18846 + }, + { + "epoch": 2.397532120595344, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.843111038208008, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8920199871063232, + "num_tokens": 731875939.0, + "step": 18847 + }, + { + "epoch": 2.3976593308739345, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.852030754089355, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8920455574989319, + "num_tokens": 731910239.0, + "step": 18848 + }, + { + "epoch": 2.397786541152525, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.720836639404297, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.912511944770813, + "num_tokens": 731946685.0, + "step": 18849 + }, + { + "epoch": 2.3979137514311155, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.748927116394043, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.9034769535064697, + "num_tokens": 731991215.0, + "step": 18850 + }, + { + "epoch": 2.398040961709706, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.725876808166504, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.9010871648788452, + "num_tokens": 732032045.0, + "step": 18851 + }, + { + "epoch": 2.3981681719882966, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.798121452331543, + "learning_rate": 1e-06, + "loss": 0.2445, + "mean_token_accuracy": 0.9149478673934937, + "num_tokens": 732064240.0, + "step": 18852 + }, + { + "epoch": 2.398295382266887, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.760066032409668, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8924194574356079, + "num_tokens": 732109046.0, + "step": 18853 + }, + { + "epoch": 2.3984225925454776, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.846710205078125, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8964102268218994, + "num_tokens": 732143379.0, + "step": 18854 + }, + { + "epoch": 2.398549802824068, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.660057067871094, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.908787727355957, + "num_tokens": 732187904.0, + "step": 18855 + }, + { + "epoch": 2.3986770131026587, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.741764068603516, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8878384232521057, + "num_tokens": 732227255.0, + "step": 18856 + }, + { + "epoch": 2.398804223381249, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.785228729248047, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8947446942329407, + "num_tokens": 732264594.0, + "step": 18857 + }, + { + "epoch": 2.3989314336598397, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.758523941040039, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9070762991905212, + "num_tokens": 732300299.0, + "step": 18858 + }, + { + "epoch": 2.3990586439384303, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.751304626464844, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8961502909660339, + "num_tokens": 732339016.0, + "step": 18859 + }, + { + "epoch": 2.399185854217021, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.736059188842773, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9086560010910034, + "num_tokens": 732378870.0, + "step": 18860 + }, + { + "epoch": 2.3993130644956113, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.670912742614746, + "learning_rate": 1e-06, + "loss": 0.2591, + "mean_token_accuracy": 0.9121901988983154, + "num_tokens": 732420304.0, + "step": 18861 + }, + { + "epoch": 2.399440274774202, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.841772079467773, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8973477482795715, + "num_tokens": 732456687.0, + "step": 18862 + }, + { + "epoch": 2.3995674850527924, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.761630058288574, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8924756050109863, + "num_tokens": 732495716.0, + "step": 18863 + }, + { + "epoch": 2.399694695331383, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.71618366241455, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.9001930952072144, + "num_tokens": 732537211.0, + "step": 18864 + }, + { + "epoch": 2.3998219056099734, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.84406852722168, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8937270641326904, + "num_tokens": 732573922.0, + "step": 18865 + }, + { + "epoch": 2.399949115888564, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.748395919799805, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.886009693145752, + "num_tokens": 732618780.0, + "step": 18866 + }, + { + "epoch": 2.4000763261671545, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.796226501464844, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8951448798179626, + "num_tokens": 732656069.0, + "step": 18867 + }, + { + "epoch": 2.400203536445745, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.699995040893555, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8957975506782532, + "num_tokens": 732701944.0, + "step": 18868 + }, + { + "epoch": 2.400330746724335, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.814661026000977, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8842312693595886, + "num_tokens": 732738382.0, + "step": 18869 + }, + { + "epoch": 2.400457957002926, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.724501609802246, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.9031172394752502, + "num_tokens": 732780283.0, + "step": 18870 + }, + { + "epoch": 2.400585167281516, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.824670791625977, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8985582590103149, + "num_tokens": 732813887.0, + "step": 18871 + }, + { + "epoch": 2.4007123775601067, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.882165908813477, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.890584409236908, + "num_tokens": 732850013.0, + "step": 18872 + }, + { + "epoch": 2.400839587838697, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.821438789367676, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8929629325866699, + "num_tokens": 732886279.0, + "step": 18873 + }, + { + "epoch": 2.4009667981172877, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.825772285461426, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8972867131233215, + "num_tokens": 732921347.0, + "step": 18874 + }, + { + "epoch": 2.4010940083958783, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.723336219787598, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.897416353225708, + "num_tokens": 732964695.0, + "step": 18875 + }, + { + "epoch": 2.401221218674469, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.722615242004395, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8953410983085632, + "num_tokens": 733011052.0, + "step": 18876 + }, + { + "epoch": 2.4013484289530593, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.842586517333984, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8984994888305664, + "num_tokens": 733045983.0, + "step": 18877 + }, + { + "epoch": 2.40147563923165, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.781861305236816, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8992830514907837, + "num_tokens": 733082157.0, + "step": 18878 + }, + { + "epoch": 2.4016028495102404, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 10.131135940551758, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8953943252563477, + "num_tokens": 733122235.0, + "step": 18879 + }, + { + "epoch": 2.401730059788831, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.83350944519043, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8964293599128723, + "num_tokens": 733163928.0, + "step": 18880 + }, + { + "epoch": 2.4018572700674214, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.597830772399902, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8998606204986572, + "num_tokens": 733202708.0, + "step": 18881 + }, + { + "epoch": 2.401984480346012, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.75497055053711, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8999273777008057, + "num_tokens": 733239774.0, + "step": 18882 + }, + { + "epoch": 2.4021116906246025, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.787693977355957, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.9002366662025452, + "num_tokens": 733279078.0, + "step": 18883 + }, + { + "epoch": 2.402238900903193, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.753235816955566, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8934836387634277, + "num_tokens": 733317634.0, + "step": 18884 + }, + { + "epoch": 2.4023661111817836, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.779765129089355, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8936947584152222, + "num_tokens": 733354228.0, + "step": 18885 + }, + { + "epoch": 2.402493321460374, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.738569259643555, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.9001318216323853, + "num_tokens": 733394168.0, + "step": 18886 + }, + { + "epoch": 2.4026205317389646, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.763572692871094, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.902342677116394, + "num_tokens": 733434015.0, + "step": 18887 + }, + { + "epoch": 2.402747742017555, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.838884353637695, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9018539190292358, + "num_tokens": 733465824.0, + "step": 18888 + }, + { + "epoch": 2.4028749522961457, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.890595436096191, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8964959383010864, + "num_tokens": 733497581.0, + "step": 18889 + }, + { + "epoch": 2.403002162574736, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.779154777526855, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8958126306533813, + "num_tokens": 733539641.0, + "step": 18890 + }, + { + "epoch": 2.4031293728533267, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.723038673400879, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8933209180831909, + "num_tokens": 733578274.0, + "step": 18891 + }, + { + "epoch": 2.403256583131917, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.723855972290039, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.897150993347168, + "num_tokens": 733620635.0, + "step": 18892 + }, + { + "epoch": 2.403383793410508, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.72077465057373, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.9019366502761841, + "num_tokens": 733662369.0, + "step": 18893 + }, + { + "epoch": 2.403511003689098, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.749368667602539, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8987718820571899, + "num_tokens": 733701403.0, + "step": 18894 + }, + { + "epoch": 2.4036382139676884, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.664163589477539, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.9079050421714783, + "num_tokens": 733743498.0, + "step": 18895 + }, + { + "epoch": 2.403765424246279, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.715032577514648, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8994807004928589, + "num_tokens": 733785148.0, + "step": 18896 + }, + { + "epoch": 2.4038926345248695, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.719564437866211, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9085754156112671, + "num_tokens": 733822663.0, + "step": 18897 + }, + { + "epoch": 2.40401984480346, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.06638240814209, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8784992694854736, + "num_tokens": 733855102.0, + "step": 18898 + }, + { + "epoch": 2.4041470550820505, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.762861251831055, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8903542160987854, + "num_tokens": 733893500.0, + "step": 18899 + }, + { + "epoch": 2.404274265360641, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.798495292663574, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8977311849594116, + "num_tokens": 733931583.0, + "step": 18900 + }, + { + "epoch": 2.4044014756392316, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.74909782409668, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8934499025344849, + "num_tokens": 733971178.0, + "step": 18901 + }, + { + "epoch": 2.404528685917822, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.80650520324707, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8966426849365234, + "num_tokens": 734007889.0, + "step": 18902 + }, + { + "epoch": 2.4046558961964126, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.7420015335083, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9034217596054077, + "num_tokens": 734044773.0, + "step": 18903 + }, + { + "epoch": 2.404783106475003, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.728011131286621, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9092460870742798, + "num_tokens": 734080587.0, + "step": 18904 + }, + { + "epoch": 2.4049103167535937, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.969849586486816, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9035367965698242, + "num_tokens": 734107321.0, + "step": 18905 + }, + { + "epoch": 2.405037527032184, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.677873611450195, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.9093458652496338, + "num_tokens": 734145208.0, + "step": 18906 + }, + { + "epoch": 2.4051647373107747, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.732295036315918, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8912379741668701, + "num_tokens": 734192824.0, + "step": 18907 + }, + { + "epoch": 2.4052919475893653, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.828157424926758, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8980065584182739, + "num_tokens": 734232287.0, + "step": 18908 + }, + { + "epoch": 2.405419157867956, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.73946762084961, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8961414098739624, + "num_tokens": 734276554.0, + "step": 18909 + }, + { + "epoch": 2.4055463681465463, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.829948425292969, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.9033994674682617, + "num_tokens": 734311043.0, + "step": 18910 + }, + { + "epoch": 2.405673578425137, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.765145301818848, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.900492787361145, + "num_tokens": 734348263.0, + "step": 18911 + }, + { + "epoch": 2.4058007887037274, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.815339088439941, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8961227536201477, + "num_tokens": 734386832.0, + "step": 18912 + }, + { + "epoch": 2.405927998982318, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.764262199401855, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9095438718795776, + "num_tokens": 734427558.0, + "step": 18913 + }, + { + "epoch": 2.4060552092609084, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.745408058166504, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9086073637008667, + "num_tokens": 734465995.0, + "step": 18914 + }, + { + "epoch": 2.406182419539499, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.795737266540527, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8975783586502075, + "num_tokens": 734503572.0, + "step": 18915 + }, + { + "epoch": 2.4063096298180895, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.828195571899414, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8961821794509888, + "num_tokens": 734542985.0, + "step": 18916 + }, + { + "epoch": 2.4064368400966796, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.791708946228027, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8871578574180603, + "num_tokens": 734585779.0, + "step": 18917 + }, + { + "epoch": 2.4065640503752705, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.721817016601562, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9103937149047852, + "num_tokens": 734622741.0, + "step": 18918 + }, + { + "epoch": 2.4066912606538606, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.83280086517334, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8839364051818848, + "num_tokens": 734661257.0, + "step": 18919 + }, + { + "epoch": 2.406818470932451, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.866132736206055, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.9008132219314575, + "num_tokens": 734696524.0, + "step": 18920 + }, + { + "epoch": 2.4069456812110417, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.744305610656738, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.9010720252990723, + "num_tokens": 734732997.0, + "step": 18921 + }, + { + "epoch": 2.407072891489632, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.772860527038574, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8997957110404968, + "num_tokens": 734772940.0, + "step": 18922 + }, + { + "epoch": 2.4072001017682227, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.737569808959961, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8975571393966675, + "num_tokens": 734816115.0, + "step": 18923 + }, + { + "epoch": 2.4073273120468133, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.840204238891602, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8955292701721191, + "num_tokens": 734854622.0, + "step": 18924 + }, + { + "epoch": 2.407454522325404, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.785635948181152, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8974714875221252, + "num_tokens": 734887618.0, + "step": 18925 + }, + { + "epoch": 2.4075817326039943, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.812716484069824, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.9052648544311523, + "num_tokens": 734922127.0, + "step": 18926 + }, + { + "epoch": 2.407708942882585, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.766402244567871, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8869860172271729, + "num_tokens": 734960529.0, + "step": 18927 + }, + { + "epoch": 2.4078361531611754, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.82800006866455, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.891817033290863, + "num_tokens": 734999087.0, + "step": 18928 + }, + { + "epoch": 2.407963363439766, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.720309257507324, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8934694528579712, + "num_tokens": 735041972.0, + "step": 18929 + }, + { + "epoch": 2.4080905737183564, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.82217025756836, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8925212621688843, + "num_tokens": 735079104.0, + "step": 18930 + }, + { + "epoch": 2.408217783996947, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.755022048950195, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8934682011604309, + "num_tokens": 735120086.0, + "step": 18931 + }, + { + "epoch": 2.4083449942755375, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.780577659606934, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8894471526145935, + "num_tokens": 735161295.0, + "step": 18932 + }, + { + "epoch": 2.408472204554128, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.762418746948242, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.890062689781189, + "num_tokens": 735200565.0, + "step": 18933 + }, + { + "epoch": 2.4085994148327186, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.770301818847656, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.894293487071991, + "num_tokens": 735235555.0, + "step": 18934 + }, + { + "epoch": 2.408726625111309, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.793534278869629, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9013091921806335, + "num_tokens": 735270735.0, + "step": 18935 + }, + { + "epoch": 2.4088538353898996, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.774742126464844, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9035214185714722, + "num_tokens": 735307719.0, + "step": 18936 + }, + { + "epoch": 2.40898104566849, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.869318008422852, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8976644277572632, + "num_tokens": 735339706.0, + "step": 18937 + }, + { + "epoch": 2.4091082559470807, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.704584121704102, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8959382772445679, + "num_tokens": 735382924.0, + "step": 18938 + }, + { + "epoch": 2.409235466225671, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.76780891418457, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8943426609039307, + "num_tokens": 735423189.0, + "step": 18939 + }, + { + "epoch": 2.4093626765042617, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.739832878112793, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9166671633720398, + "num_tokens": 735460218.0, + "step": 18940 + }, + { + "epoch": 2.4094898867828523, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.787944793701172, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8841987252235413, + "num_tokens": 735505723.0, + "step": 18941 + }, + { + "epoch": 2.4096170970614423, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.923829078674316, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8941266536712646, + "num_tokens": 735538362.0, + "step": 18942 + }, + { + "epoch": 2.4097443073400333, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.839910507202148, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.892897367477417, + "num_tokens": 735576927.0, + "step": 18943 + }, + { + "epoch": 2.4098715176186234, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.677270889282227, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9064086675643921, + "num_tokens": 735621533.0, + "step": 18944 + }, + { + "epoch": 2.409998727897214, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.7486572265625, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8959578275680542, + "num_tokens": 735666846.0, + "step": 18945 + }, + { + "epoch": 2.4101259381758044, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.783430099487305, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9096230268478394, + "num_tokens": 735706302.0, + "step": 18946 + }, + { + "epoch": 2.410253148454395, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.71176528930664, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8982537388801575, + "num_tokens": 735747509.0, + "step": 18947 + }, + { + "epoch": 2.4103803587329855, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.736295700073242, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.899369478225708, + "num_tokens": 735792526.0, + "step": 18948 + }, + { + "epoch": 2.410507569011576, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.770100593566895, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8957656621932983, + "num_tokens": 735834126.0, + "step": 18949 + }, + { + "epoch": 2.4106347792901666, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.79162311553955, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9093471169471741, + "num_tokens": 735865187.0, + "step": 18950 + }, + { + "epoch": 2.410761989568757, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.779149055480957, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8903011679649353, + "num_tokens": 735906157.0, + "step": 18951 + }, + { + "epoch": 2.4108891998473476, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.778816223144531, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8981101512908936, + "num_tokens": 735944339.0, + "step": 18952 + }, + { + "epoch": 2.411016410125938, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.749881744384766, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.9004213213920593, + "num_tokens": 735979369.0, + "step": 18953 + }, + { + "epoch": 2.4111436204045287, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.872208595275879, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.9004598259925842, + "num_tokens": 736010546.0, + "step": 18954 + }, + { + "epoch": 2.411270830683119, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.705965042114258, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9029272794723511, + "num_tokens": 736048468.0, + "step": 18955 + }, + { + "epoch": 2.4113980409617097, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.770963668823242, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8994107246398926, + "num_tokens": 736087147.0, + "step": 18956 + }, + { + "epoch": 2.4115252512403003, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.720402717590332, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9015307426452637, + "num_tokens": 736125116.0, + "step": 18957 + }, + { + "epoch": 2.411652461518891, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.789877891540527, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9081121683120728, + "num_tokens": 736159275.0, + "step": 18958 + }, + { + "epoch": 2.4117796717974813, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.718159675598145, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9107953310012817, + "num_tokens": 736200809.0, + "step": 18959 + }, + { + "epoch": 2.411906882076072, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.852242469787598, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.9037700891494751, + "num_tokens": 736235701.0, + "step": 18960 + }, + { + "epoch": 2.4120340923546624, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.722593307495117, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9024971723556519, + "num_tokens": 736271451.0, + "step": 18961 + }, + { + "epoch": 2.412161302633253, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.749914169311523, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8894646763801575, + "num_tokens": 736312383.0, + "step": 18962 + }, + { + "epoch": 2.4122885129118434, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.884981155395508, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8963258266448975, + "num_tokens": 736348496.0, + "step": 18963 + }, + { + "epoch": 2.412415723190434, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.66714859008789, + "learning_rate": 1e-06, + "loss": 0.2426, + "mean_token_accuracy": 0.9165668487548828, + "num_tokens": 736389145.0, + "step": 18964 + }, + { + "epoch": 2.4125429334690245, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.840018272399902, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8989280462265015, + "num_tokens": 736428517.0, + "step": 18965 + }, + { + "epoch": 2.412670143747615, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.795329093933105, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.9079729914665222, + "num_tokens": 736466050.0, + "step": 18966 + }, + { + "epoch": 2.412797354026205, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.870800971984863, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9030594825744629, + "num_tokens": 736499147.0, + "step": 18967 + }, + { + "epoch": 2.412924564304796, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.812198638916016, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8986582159996033, + "num_tokens": 736538020.0, + "step": 18968 + }, + { + "epoch": 2.413051774583386, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.843891143798828, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8954763412475586, + "num_tokens": 736571792.0, + "step": 18969 + }, + { + "epoch": 2.4131789848619767, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.770940780639648, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8941521048545837, + "num_tokens": 736614161.0, + "step": 18970 + }, + { + "epoch": 2.413306195140567, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.743815422058105, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8996143937110901, + "num_tokens": 736652870.0, + "step": 18971 + }, + { + "epoch": 2.4134334054191577, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.863340377807617, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8963714241981506, + "num_tokens": 736687711.0, + "step": 18972 + }, + { + "epoch": 2.4135606156977483, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.77794361114502, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8917484283447266, + "num_tokens": 736723995.0, + "step": 18973 + }, + { + "epoch": 2.413687825976339, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.742783546447754, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8981364965438843, + "num_tokens": 736760848.0, + "step": 18974 + }, + { + "epoch": 2.4138150362549293, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.88581657409668, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9041715860366821, + "num_tokens": 736799049.0, + "step": 18975 + }, + { + "epoch": 2.41394224653352, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.716398239135742, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.9003896713256836, + "num_tokens": 736838235.0, + "step": 18976 + }, + { + "epoch": 2.4140694568121104, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.821722984313965, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8979771137237549, + "num_tokens": 736875899.0, + "step": 18977 + }, + { + "epoch": 2.414196667090701, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.779845237731934, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8910378813743591, + "num_tokens": 736919568.0, + "step": 18978 + }, + { + "epoch": 2.4143238773692914, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.66420841217041, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8966426849365234, + "num_tokens": 736963592.0, + "step": 18979 + }, + { + "epoch": 2.414451087647882, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.769068717956543, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9047192335128784, + "num_tokens": 737001366.0, + "step": 18980 + }, + { + "epoch": 2.4145782979264725, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.801053047180176, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8914361000061035, + "num_tokens": 737035784.0, + "step": 18981 + }, + { + "epoch": 2.414705508205063, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.879826545715332, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8883817195892334, + "num_tokens": 737068511.0, + "step": 18982 + }, + { + "epoch": 2.4148327184836536, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.782418251037598, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9032844305038452, + "num_tokens": 737104895.0, + "step": 18983 + }, + { + "epoch": 2.414959928762244, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.649924278259277, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.9045639038085938, + "num_tokens": 737149743.0, + "step": 18984 + }, + { + "epoch": 2.4150871390408346, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.821342468261719, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8969289064407349, + "num_tokens": 737182678.0, + "step": 18985 + }, + { + "epoch": 2.415214349319425, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.69895076751709, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8953352570533752, + "num_tokens": 737223053.0, + "step": 18986 + }, + { + "epoch": 2.4153415595980157, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.645264625549316, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9074466228485107, + "num_tokens": 737265944.0, + "step": 18987 + }, + { + "epoch": 2.415468769876606, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.84774398803711, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8884392380714417, + "num_tokens": 737306085.0, + "step": 18988 + }, + { + "epoch": 2.4155959801551967, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.693949699401855, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.90334552526474, + "num_tokens": 737346057.0, + "step": 18989 + }, + { + "epoch": 2.415723190433787, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.699737548828125, + "learning_rate": 1e-06, + "loss": 0.25, + "mean_token_accuracy": 0.9156647324562073, + "num_tokens": 737379727.0, + "step": 18990 + }, + { + "epoch": 2.4158504007123778, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.813019752502441, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.910542905330658, + "num_tokens": 737417702.0, + "step": 18991 + }, + { + "epoch": 2.415977610990968, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.776214599609375, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.9001636505126953, + "num_tokens": 737455432.0, + "step": 18992 + }, + { + "epoch": 2.4161048212695584, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.753061294555664, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.896414041519165, + "num_tokens": 737497500.0, + "step": 18993 + }, + { + "epoch": 2.416232031548149, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.761798858642578, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8896753787994385, + "num_tokens": 737535890.0, + "step": 18994 + }, + { + "epoch": 2.4163592418267394, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.841509819030762, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8956136703491211, + "num_tokens": 737570515.0, + "step": 18995 + }, + { + "epoch": 2.41648645210533, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.764812469482422, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8817871809005737, + "num_tokens": 737615702.0, + "step": 18996 + }, + { + "epoch": 2.4166136623839205, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.75980281829834, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8921088576316833, + "num_tokens": 737658235.0, + "step": 18997 + }, + { + "epoch": 2.416740872662511, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.779207229614258, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9068105220794678, + "num_tokens": 737692687.0, + "step": 18998 + }, + { + "epoch": 2.4168680829411016, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.71689510345459, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9152885675430298, + "num_tokens": 737729853.0, + "step": 18999 + }, + { + "epoch": 2.416995293219692, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.69378662109375, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9137041568756104, + "num_tokens": 737772158.0, + "step": 19000 + }, + { + "epoch": 2.4171225034982826, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.913314819335938, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8930851221084595, + "num_tokens": 737811653.0, + "step": 19001 + }, + { + "epoch": 2.417249713776873, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.76744270324707, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.9060318470001221, + "num_tokens": 737852810.0, + "step": 19002 + }, + { + "epoch": 2.4173769240554637, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.747188568115234, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8958557844161987, + "num_tokens": 737897876.0, + "step": 19003 + }, + { + "epoch": 2.417504134334054, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.925915718078613, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8966600894927979, + "num_tokens": 737936887.0, + "step": 19004 + }, + { + "epoch": 2.4176313446126447, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.746783256530762, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9046765565872192, + "num_tokens": 737976600.0, + "step": 19005 + }, + { + "epoch": 2.4177585548912353, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.754749298095703, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8979604244232178, + "num_tokens": 738015834.0, + "step": 19006 + }, + { + "epoch": 2.417885765169826, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.757126808166504, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8985525369644165, + "num_tokens": 738053984.0, + "step": 19007 + }, + { + "epoch": 2.4180129754484163, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.985624313354492, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8857316970825195, + "num_tokens": 738086946.0, + "step": 19008 + }, + { + "epoch": 2.418140185727007, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.857864379882812, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8964864015579224, + "num_tokens": 738122313.0, + "step": 19009 + }, + { + "epoch": 2.4182673960055974, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.719925880432129, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.9029719233512878, + "num_tokens": 738163625.0, + "step": 19010 + }, + { + "epoch": 2.418394606284188, + "ewc_loss": 0.02001953125, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.785964012145996, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9013561010360718, + "num_tokens": 738202619.0, + "step": 19011 + }, + { + "epoch": 2.4185218165627784, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.885736465454102, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.893242597579956, + "num_tokens": 738241920.0, + "step": 19012 + }, + { + "epoch": 2.418649026841369, + "ewc_loss": 0.0201416015625, + "ewc_loss_diag": 8.487701416015625e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.694354057312012, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8949682712554932, + "num_tokens": 738282781.0, + "step": 19013 + }, + { + "epoch": 2.4187762371199595, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.863037109375, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8968701958656311, + "num_tokens": 738318777.0, + "step": 19014 + }, + { + "epoch": 2.4189034473985496, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.652946472167969, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8960756063461304, + "num_tokens": 738363139.0, + "step": 19015 + }, + { + "epoch": 2.4190306576771405, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.835948944091797, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.9012118577957153, + "num_tokens": 738400418.0, + "step": 19016 + }, + { + "epoch": 2.4191578679557306, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.819473266601562, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.9005048274993896, + "num_tokens": 738434025.0, + "step": 19017 + }, + { + "epoch": 2.419285078234321, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.751253128051758, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8976653814315796, + "num_tokens": 738473921.0, + "step": 19018 + }, + { + "epoch": 2.4194122885129117, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.813998222351074, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.898826003074646, + "num_tokens": 738508917.0, + "step": 19019 + }, + { + "epoch": 2.419539498791502, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.818625450134277, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8944165110588074, + "num_tokens": 738547165.0, + "step": 19020 + }, + { + "epoch": 2.4196667090700927, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.759967803955078, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8973242044448853, + "num_tokens": 738584114.0, + "step": 19021 + }, + { + "epoch": 2.4197939193486833, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.729615211486816, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9049252271652222, + "num_tokens": 738622180.0, + "step": 19022 + }, + { + "epoch": 2.419921129627274, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.58306884765625e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.795510292053223, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8954048752784729, + "num_tokens": 738661484.0, + "step": 19023 + }, + { + "epoch": 2.4200483399058643, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.842504501342773, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8914684653282166, + "num_tokens": 738700232.0, + "step": 19024 + }, + { + "epoch": 2.420175550184455, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.797656059265137, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8907051086425781, + "num_tokens": 738742247.0, + "step": 19025 + }, + { + "epoch": 2.4203027604630454, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.754742622375488, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.902875542640686, + "num_tokens": 738782776.0, + "step": 19026 + }, + { + "epoch": 2.420429970741636, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.535385131835938e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.859896659851074, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8928502202033997, + "num_tokens": 738822311.0, + "step": 19027 + }, + { + "epoch": 2.4205571810202264, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.728952407836914, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8935941457748413, + "num_tokens": 738864062.0, + "step": 19028 + }, + { + "epoch": 2.420684391298817, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.782114028930664, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8929544687271118, + "num_tokens": 738906834.0, + "step": 19029 + }, + { + "epoch": 2.4208116015774075, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.535385131835938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.79949951171875, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9112708568572998, + "num_tokens": 738940568.0, + "step": 19030 + }, + { + "epoch": 2.420938811855998, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.804189682006836, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8980926275253296, + "num_tokens": 738976463.0, + "step": 19031 + }, + { + "epoch": 2.4210660221345885, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.809326171875, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.894899845123291, + "num_tokens": 739014257.0, + "step": 19032 + }, + { + "epoch": 2.421193232413179, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.535385131835938e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.672333717346191, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.9024393558502197, + "num_tokens": 739057335.0, + "step": 19033 + }, + { + "epoch": 2.4213204426917696, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.693564414978027, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9008320569992065, + "num_tokens": 739102179.0, + "step": 19034 + }, + { + "epoch": 2.42144765297036, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.727967262268066, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9052845239639282, + "num_tokens": 739141770.0, + "step": 19035 + }, + { + "epoch": 2.4215748632489507, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 10.136605262756348, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8986957669258118, + "num_tokens": 739179554.0, + "step": 19036 + }, + { + "epoch": 2.421702073527541, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.79772663116455, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.9017949104309082, + "num_tokens": 739221265.0, + "step": 19037 + }, + { + "epoch": 2.4218292838061317, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.733190536499023, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.886187732219696, + "num_tokens": 739260361.0, + "step": 19038 + }, + { + "epoch": 2.4219564940847222, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.758580207824707, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.9090810418128967, + "num_tokens": 739296427.0, + "step": 19039 + }, + { + "epoch": 2.4220837043633123, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.767502784729004, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8896323442459106, + "num_tokens": 739337840.0, + "step": 19040 + }, + { + "epoch": 2.4222109146419033, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.727523803710938, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8907631635665894, + "num_tokens": 739379535.0, + "step": 19041 + }, + { + "epoch": 2.4223381249204934, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.885498046875, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8938627243041992, + "num_tokens": 739413018.0, + "step": 19042 + }, + { + "epoch": 2.422465335199084, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.761631965637207, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8877438306808472, + "num_tokens": 739452711.0, + "step": 19043 + }, + { + "epoch": 2.4225925454776744, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.81290054321289, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8952603340148926, + "num_tokens": 739491656.0, + "step": 19044 + }, + { + "epoch": 2.422719755756265, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.726966857910156, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8896428942680359, + "num_tokens": 739535407.0, + "step": 19045 + }, + { + "epoch": 2.4228469660348555, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.80199909210205, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.880975604057312, + "num_tokens": 739570795.0, + "step": 19046 + }, + { + "epoch": 2.422974176313446, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.768406867980957, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9100513458251953, + "num_tokens": 739608290.0, + "step": 19047 + }, + { + "epoch": 2.4231013865920366, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.659673690795898, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8949345946311951, + "num_tokens": 739648460.0, + "step": 19048 + }, + { + "epoch": 2.423228596870627, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.797178268432617, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.9052829146385193, + "num_tokens": 739683884.0, + "step": 19049 + }, + { + "epoch": 2.4233558071492176, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.844178199768066, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.9021510481834412, + "num_tokens": 739716609.0, + "step": 19050 + }, + { + "epoch": 2.423483017427808, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.730623245239258, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.9038563370704651, + "num_tokens": 739754100.0, + "step": 19051 + }, + { + "epoch": 2.4236102277063987, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.66801643371582, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9008876085281372, + "num_tokens": 739799614.0, + "step": 19052 + }, + { + "epoch": 2.423737437984989, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.863515853881836, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9090201258659363, + "num_tokens": 739833261.0, + "step": 19053 + }, + { + "epoch": 2.4238646482635797, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.757328033447266, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8990722894668579, + "num_tokens": 739874732.0, + "step": 19054 + }, + { + "epoch": 2.4239918585421703, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.838241577148438, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8932421803474426, + "num_tokens": 739909778.0, + "step": 19055 + }, + { + "epoch": 2.424119068820761, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.840738296508789, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9074169397354126, + "num_tokens": 739943541.0, + "step": 19056 + }, + { + "epoch": 2.4242462790993513, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.725614547729492, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.906330943107605, + "num_tokens": 739979600.0, + "step": 19057 + }, + { + "epoch": 2.424373489377942, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.826513290405273, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9008674621582031, + "num_tokens": 740017278.0, + "step": 19058 + }, + { + "epoch": 2.4245006996565324, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.72678279876709, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9030590057373047, + "num_tokens": 740057910.0, + "step": 19059 + }, + { + "epoch": 2.424627909935123, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.861767768859863, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8886827230453491, + "num_tokens": 740092955.0, + "step": 19060 + }, + { + "epoch": 2.4247551202137134, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.630752563476562e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.76526927947998, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.9030600786209106, + "num_tokens": 740138041.0, + "step": 19061 + }, + { + "epoch": 2.424882330492304, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.694191932678223, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8927879333496094, + "num_tokens": 740183037.0, + "step": 19062 + }, + { + "epoch": 2.4250095407708945, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.795439720153809, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.9044677019119263, + "num_tokens": 740219786.0, + "step": 19063 + }, + { + "epoch": 2.425136751049485, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.748870849609375, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.9035438299179077, + "num_tokens": 740263159.0, + "step": 19064 + }, + { + "epoch": 2.425263961328075, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.709250450134277, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9037238359451294, + "num_tokens": 740304898.0, + "step": 19065 + }, + { + "epoch": 2.425391171606666, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.740351676940918, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8966705799102783, + "num_tokens": 740348666.0, + "step": 19066 + }, + { + "epoch": 2.425518381885256, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.75693416595459, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9036871194839478, + "num_tokens": 740386492.0, + "step": 19067 + }, + { + "epoch": 2.4256455921638467, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.792911529541016, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9081395864486694, + "num_tokens": 740418706.0, + "step": 19068 + }, + { + "epoch": 2.425772802442437, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.827659606933594, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.9080089330673218, + "num_tokens": 740451070.0, + "step": 19069 + }, + { + "epoch": 2.4259000127210277, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.753639221191406, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8873664140701294, + "num_tokens": 740488775.0, + "step": 19070 + }, + { + "epoch": 2.4260272229996183, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.805177688598633, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8946107625961304, + "num_tokens": 740525817.0, + "step": 19071 + }, + { + "epoch": 2.426154433278209, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.659525871276855, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9102000594139099, + "num_tokens": 740566722.0, + "step": 19072 + }, + { + "epoch": 2.4262816435567993, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.759825706481934, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8976161479949951, + "num_tokens": 740603422.0, + "step": 19073 + }, + { + "epoch": 2.42640885383539, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.700959205627441, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9064635038375854, + "num_tokens": 740645589.0, + "step": 19074 + }, + { + "epoch": 2.4265360641139804, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.81395149230957, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8990322351455688, + "num_tokens": 740683717.0, + "step": 19075 + }, + { + "epoch": 2.426663274392571, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.735422134399414, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9060724973678589, + "num_tokens": 740723031.0, + "step": 19076 + }, + { + "epoch": 2.4267904846711614, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.860798835754395, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8896883726119995, + "num_tokens": 740760743.0, + "step": 19077 + }, + { + "epoch": 2.426917694949752, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.818580627441406, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8977645039558411, + "num_tokens": 740795901.0, + "step": 19078 + }, + { + "epoch": 2.4270449052283425, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.835979461669922, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8877691030502319, + "num_tokens": 740833828.0, + "step": 19079 + }, + { + "epoch": 2.427172115506933, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.749979972839355, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9059574604034424, + "num_tokens": 740871908.0, + "step": 19080 + }, + { + "epoch": 2.4272993257855235, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.750691413879395, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8988577127456665, + "num_tokens": 740911082.0, + "step": 19081 + }, + { + "epoch": 2.427426536064114, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.718194007873535, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9096867442131042, + "num_tokens": 740948251.0, + "step": 19082 + }, + { + "epoch": 2.4275537463427046, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.806497573852539, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8883563280105591, + "num_tokens": 740987051.0, + "step": 19083 + }, + { + "epoch": 2.427680956621295, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.919843673706055, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8877580761909485, + "num_tokens": 741020190.0, + "step": 19084 + }, + { + "epoch": 2.4278081668998857, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.682066917419434, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.9068623781204224, + "num_tokens": 741062262.0, + "step": 19085 + }, + { + "epoch": 2.427935377178476, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.851587295532227, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8902403712272644, + "num_tokens": 741101883.0, + "step": 19086 + }, + { + "epoch": 2.4280625874570667, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 10.597902297973633, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9087821245193481, + "num_tokens": 741130648.0, + "step": 19087 + }, + { + "epoch": 2.428189797735657, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.737166404724121, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8912225961685181, + "num_tokens": 741169881.0, + "step": 19088 + }, + { + "epoch": 2.4283170080142478, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.757741928100586, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9034470915794373, + "num_tokens": 741206004.0, + "step": 19089 + }, + { + "epoch": 2.428444218292838, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.78018569946289, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.9008833169937134, + "num_tokens": 741241712.0, + "step": 19090 + }, + { + "epoch": 2.4285714285714284, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.656567573547363, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.9052594900131226, + "num_tokens": 741287426.0, + "step": 19091 + }, + { + "epoch": 2.428698638850019, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.831583976745605, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8884841203689575, + "num_tokens": 741325755.0, + "step": 19092 + }, + { + "epoch": 2.4288258491286094, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.749444007873535, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9023820757865906, + "num_tokens": 741364893.0, + "step": 19093 + }, + { + "epoch": 2.4289530594072, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.73324203491211, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8794655799865723, + "num_tokens": 741409500.0, + "step": 19094 + }, + { + "epoch": 2.4290802696857905, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.852560997009277, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.9029676914215088, + "num_tokens": 741446952.0, + "step": 19095 + }, + { + "epoch": 2.429207479964381, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.685884475708008, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.9041050672531128, + "num_tokens": 741491886.0, + "step": 19096 + }, + { + "epoch": 2.4293346902429716, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.839536666870117, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.9008569121360779, + "num_tokens": 741529021.0, + "step": 19097 + }, + { + "epoch": 2.429461900521562, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.838754653930664, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.9022468328475952, + "num_tokens": 741565228.0, + "step": 19098 + }, + { + "epoch": 2.4295891108001526, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.873598098754883, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8926376104354858, + "num_tokens": 741602761.0, + "step": 19099 + }, + { + "epoch": 2.429716321078743, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.80892276763916, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8861085176467896, + "num_tokens": 741643267.0, + "step": 19100 + }, + { + "epoch": 2.4298435313573337, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.765064239501953, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.9047147035598755, + "num_tokens": 741678236.0, + "step": 19101 + }, + { + "epoch": 2.429970741635924, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.784124374389648, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.9016938805580139, + "num_tokens": 741720878.0, + "step": 19102 + }, + { + "epoch": 2.4300979519145147, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.735703468322754, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9045604467391968, + "num_tokens": 741759293.0, + "step": 19103 + }, + { + "epoch": 2.4302251621931052, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.845415115356445, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8882951736450195, + "num_tokens": 741800535.0, + "step": 19104 + }, + { + "epoch": 2.4303523724716958, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.711664199829102, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9038334488868713, + "num_tokens": 741839325.0, + "step": 19105 + }, + { + "epoch": 2.4304795827502863, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.784274101257324, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8973677754402161, + "num_tokens": 741878921.0, + "step": 19106 + }, + { + "epoch": 2.430606793028877, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.77966594696045, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.9069377183914185, + "num_tokens": 741916536.0, + "step": 19107 + }, + { + "epoch": 2.4307340033074674, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.851275444030762, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.9084396362304688, + "num_tokens": 741948570.0, + "step": 19108 + }, + { + "epoch": 2.430861213586058, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.807555198669434, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.9001660346984863, + "num_tokens": 741985124.0, + "step": 19109 + }, + { + "epoch": 2.4309884238646484, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.677691459655762, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8996987342834473, + "num_tokens": 742028047.0, + "step": 19110 + }, + { + "epoch": 2.431115634143239, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.785627365112305, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8892019987106323, + "num_tokens": 742070675.0, + "step": 19111 + }, + { + "epoch": 2.4312428444218295, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.79599380493164, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.9029637575149536, + "num_tokens": 742110841.0, + "step": 19112 + }, + { + "epoch": 2.4313700547004196, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.777593612670898, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.886013388633728, + "num_tokens": 742149692.0, + "step": 19113 + }, + { + "epoch": 2.4314972649790105, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.761237144470215, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.9004344940185547, + "num_tokens": 742192342.0, + "step": 19114 + }, + { + "epoch": 2.4316244752576006, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.715313911437988, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.9012517929077148, + "num_tokens": 742231653.0, + "step": 19115 + }, + { + "epoch": 2.431751685536191, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.757170677185059, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.9022485613822937, + "num_tokens": 742275595.0, + "step": 19116 + }, + { + "epoch": 2.4318788958147817, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.785689353942871, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9110337495803833, + "num_tokens": 742309274.0, + "step": 19117 + }, + { + "epoch": 2.432006106093372, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.863188743591309, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8942203521728516, + "num_tokens": 742346388.0, + "step": 19118 + }, + { + "epoch": 2.4321333163719627, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.747211456298828, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9039183855056763, + "num_tokens": 742389835.0, + "step": 19119 + }, + { + "epoch": 2.4322605266505533, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.789813041687012, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8963632583618164, + "num_tokens": 742428031.0, + "step": 19120 + }, + { + "epoch": 2.432387736929144, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.838269233703613, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.895675778388977, + "num_tokens": 742469131.0, + "step": 19121 + }, + { + "epoch": 2.4325149472077343, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.771493911743164, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.9036140441894531, + "num_tokens": 742508775.0, + "step": 19122 + }, + { + "epoch": 2.432642157486325, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.84969425201416, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8986704349517822, + "num_tokens": 742546356.0, + "step": 19123 + }, + { + "epoch": 2.4327693677649154, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.728511810302734, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.9049752950668335, + "num_tokens": 742592233.0, + "step": 19124 + }, + { + "epoch": 2.432896578043506, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.823797225952148, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8942041397094727, + "num_tokens": 742627850.0, + "step": 19125 + }, + { + "epoch": 2.4330237883220964, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.756216049194336, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8992511034011841, + "num_tokens": 742667876.0, + "step": 19126 + }, + { + "epoch": 2.433150998600687, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.768610954284668, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.9025329947471619, + "num_tokens": 742709308.0, + "step": 19127 + }, + { + "epoch": 2.4332782088792775, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.859975814819336, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8917188048362732, + "num_tokens": 742744008.0, + "step": 19128 + }, + { + "epoch": 2.433405419157868, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.733097076416016, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9105324745178223, + "num_tokens": 742784357.0, + "step": 19129 + }, + { + "epoch": 2.4335326294364585, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.778534889221191, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9095752239227295, + "num_tokens": 742820873.0, + "step": 19130 + }, + { + "epoch": 2.433659839715049, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.824328422546387, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8987847566604614, + "num_tokens": 742861366.0, + "step": 19131 + }, + { + "epoch": 2.4337870499936396, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.76431655883789, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8952518701553345, + "num_tokens": 742906180.0, + "step": 19132 + }, + { + "epoch": 2.43391426027223, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.873565673828125, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8898418545722961, + "num_tokens": 742945719.0, + "step": 19133 + }, + { + "epoch": 2.4340414705508207, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.82521915435791, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8947470784187317, + "num_tokens": 742982561.0, + "step": 19134 + }, + { + "epoch": 2.434168680829411, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.715874671936035, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8964849710464478, + "num_tokens": 743026733.0, + "step": 19135 + }, + { + "epoch": 2.4342958911080017, + "ewc_loss": 0.020263671875, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.811447143554688, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.9001852869987488, + "num_tokens": 743064561.0, + "step": 19136 + }, + { + "epoch": 2.4344231013865922, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.792137145996094, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8944182395935059, + "num_tokens": 743111781.0, + "step": 19137 + }, + { + "epoch": 2.4345503116651823, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.665409088134766, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.9053435325622559, + "num_tokens": 743155322.0, + "step": 19138 + }, + { + "epoch": 2.4346775219437733, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.737783432006836, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.9018029570579529, + "num_tokens": 743198619.0, + "step": 19139 + }, + { + "epoch": 2.4348047322223634, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.7860689163208, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8992362022399902, + "num_tokens": 743237539.0, + "step": 19140 + }, + { + "epoch": 2.434931942500954, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.787318229675293, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8877023458480835, + "num_tokens": 743281235.0, + "step": 19141 + }, + { + "epoch": 2.4350591527795444, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.668112754821777, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9083747863769531, + "num_tokens": 743324380.0, + "step": 19142 + }, + { + "epoch": 2.435186363058135, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.789984703063965, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8975815773010254, + "num_tokens": 743360328.0, + "step": 19143 + }, + { + "epoch": 2.4353135733367255, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.80562973022461, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.8993788957595825, + "num_tokens": 743395898.0, + "step": 19144 + }, + { + "epoch": 2.435440783615316, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.88475227355957, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9007166028022766, + "num_tokens": 743424814.0, + "step": 19145 + }, + { + "epoch": 2.4355679938939065, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.726104736328125, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8979189991950989, + "num_tokens": 743465654.0, + "step": 19146 + }, + { + "epoch": 2.435695204172497, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.835367202758789, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9059114456176758, + "num_tokens": 743504382.0, + "step": 19147 + }, + { + "epoch": 2.4358224144510876, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.8287992477417, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8866485953330994, + "num_tokens": 743543613.0, + "step": 19148 + }, + { + "epoch": 2.435949624729678, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.722092628479004, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8923566341400146, + "num_tokens": 743584057.0, + "step": 19149 + }, + { + "epoch": 2.4360768350082687, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.755345344543457, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8970894813537598, + "num_tokens": 743627047.0, + "step": 19150 + }, + { + "epoch": 2.436204045286859, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.838791847229004, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8982782363891602, + "num_tokens": 743663834.0, + "step": 19151 + }, + { + "epoch": 2.4363312555654497, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.754143714904785, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9022729396820068, + "num_tokens": 743700350.0, + "step": 19152 + }, + { + "epoch": 2.4364584658440402, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.797309875488281, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8952353596687317, + "num_tokens": 743740143.0, + "step": 19153 + }, + { + "epoch": 2.4365856761226308, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.822958946228027, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8825052380561829, + "num_tokens": 743775977.0, + "step": 19154 + }, + { + "epoch": 2.4367128864012213, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.778471946716309, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8935103416442871, + "num_tokens": 743818576.0, + "step": 19155 + }, + { + "epoch": 2.436840096679812, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.889892578125, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.903437614440918, + "num_tokens": 743848330.0, + "step": 19156 + }, + { + "epoch": 2.4369673069584024, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.79386043548584, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8972416520118713, + "num_tokens": 743888029.0, + "step": 19157 + }, + { + "epoch": 2.437094517236993, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.673616409301758, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9117738008499146, + "num_tokens": 743923696.0, + "step": 19158 + }, + { + "epoch": 2.4372217275155834, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.896882057189941, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.874982476234436, + "num_tokens": 743962410.0, + "step": 19159 + }, + { + "epoch": 2.437348937794174, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.682239532470703, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9103372693061829, + "num_tokens": 744006741.0, + "step": 19160 + }, + { + "epoch": 2.4374761480727645, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.776992797851562, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8897992968559265, + "num_tokens": 744050382.0, + "step": 19161 + }, + { + "epoch": 2.437603358351355, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.75171947479248, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9019114971160889, + "num_tokens": 744092683.0, + "step": 19162 + }, + { + "epoch": 2.437730568629945, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.816553115844727, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8900280594825745, + "num_tokens": 744135879.0, + "step": 19163 + }, + { + "epoch": 2.437857778908536, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.741576194763184, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.9039039015769958, + "num_tokens": 744179788.0, + "step": 19164 + }, + { + "epoch": 2.437984989187126, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.696490287780762, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.9074312448501587, + "num_tokens": 744226434.0, + "step": 19165 + }, + { + "epoch": 2.4381121994657167, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.814841270446777, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.9012408256530762, + "num_tokens": 744263073.0, + "step": 19166 + }, + { + "epoch": 2.438239409744307, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.844524383544922, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8992593288421631, + "num_tokens": 744300703.0, + "step": 19167 + }, + { + "epoch": 2.4383666200228977, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.84383773803711, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8985628485679626, + "num_tokens": 744336309.0, + "step": 19168 + }, + { + "epoch": 2.4384938303014883, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.853408813476562, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.9051352739334106, + "num_tokens": 744364648.0, + "step": 19169 + }, + { + "epoch": 2.438621040580079, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.777522087097168, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9086538553237915, + "num_tokens": 744401744.0, + "step": 19170 + }, + { + "epoch": 2.4387482508586693, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.781100273132324, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.905640184879303, + "num_tokens": 744438109.0, + "step": 19171 + }, + { + "epoch": 2.43887546113726, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.76586627960205, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.894572377204895, + "num_tokens": 744479583.0, + "step": 19172 + }, + { + "epoch": 2.4390026714158504, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.88920783996582, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9019172191619873, + "num_tokens": 744514293.0, + "step": 19173 + }, + { + "epoch": 2.439129881694441, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.732354164123535, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8967746496200562, + "num_tokens": 744553552.0, + "step": 19174 + }, + { + "epoch": 2.4392570919730314, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.831575393676758, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8960309028625488, + "num_tokens": 744590953.0, + "step": 19175 + }, + { + "epoch": 2.439384302251622, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.860419273376465, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8881413340568542, + "num_tokens": 744632577.0, + "step": 19176 + }, + { + "epoch": 2.4395115125302125, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.732531547546387, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8970392942428589, + "num_tokens": 744670401.0, + "step": 19177 + }, + { + "epoch": 2.439638722808803, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.767606735229492, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8999308943748474, + "num_tokens": 744706733.0, + "step": 19178 + }, + { + "epoch": 2.4397659330873935, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.783577919006348, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.9008851647377014, + "num_tokens": 744743960.0, + "step": 19179 + }, + { + "epoch": 2.439893143365984, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.694256782531738, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9017462134361267, + "num_tokens": 744788292.0, + "step": 19180 + }, + { + "epoch": 2.4400203536445746, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.788966178894043, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8974759578704834, + "num_tokens": 744826538.0, + "step": 19181 + }, + { + "epoch": 2.440147563923165, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.741141319274902, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9000272750854492, + "num_tokens": 744864673.0, + "step": 19182 + }, + { + "epoch": 2.4402747742017556, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.674827575683594, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8919026255607605, + "num_tokens": 744909758.0, + "step": 19183 + }, + { + "epoch": 2.440401984480346, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.865571975708008, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8976677060127258, + "num_tokens": 744943186.0, + "step": 19184 + }, + { + "epoch": 2.4405291947589367, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.750238418579102, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8956426978111267, + "num_tokens": 744982579.0, + "step": 19185 + }, + { + "epoch": 2.440656405037527, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.767844200134277, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.900973379611969, + "num_tokens": 745023057.0, + "step": 19186 + }, + { + "epoch": 2.4407836153161178, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.697259902954102, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8998599052429199, + "num_tokens": 745064376.0, + "step": 19187 + }, + { + "epoch": 2.440910825594708, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.697796821594238, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.9073482155799866, + "num_tokens": 745103182.0, + "step": 19188 + }, + { + "epoch": 2.4410380358732984, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.94136905670166, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8880931735038757, + "num_tokens": 745145502.0, + "step": 19189 + }, + { + "epoch": 2.441165246151889, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.773058891296387, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.890041708946228, + "num_tokens": 745184682.0, + "step": 19190 + }, + { + "epoch": 2.4412924564304794, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.739885330200195, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8872183561325073, + "num_tokens": 745234265.0, + "step": 19191 + }, + { + "epoch": 2.44141966670907, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.822700500488281, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8903344869613647, + "num_tokens": 745278343.0, + "step": 19192 + }, + { + "epoch": 2.4415468769876605, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.726119995117188e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.742904663085938, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9099366664886475, + "num_tokens": 745315290.0, + "step": 19193 + }, + { + "epoch": 2.441674087266251, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.784307479858398, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8775540590286255, + "num_tokens": 745357904.0, + "step": 19194 + }, + { + "epoch": 2.4418012975448415, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.788413047790527, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8910831212997437, + "num_tokens": 745401645.0, + "step": 19195 + }, + { + "epoch": 2.441928507823432, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.75389575958252, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.906062662601471, + "num_tokens": 745442602.0, + "step": 19196 + }, + { + "epoch": 2.4420557181020226, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.831619262695312, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.9003881216049194, + "num_tokens": 745480670.0, + "step": 19197 + }, + { + "epoch": 2.442182928380613, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.89931583404541, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8906629085540771, + "num_tokens": 745521597.0, + "step": 19198 + }, + { + "epoch": 2.4423101386592037, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.700406074523926, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8960005044937134, + "num_tokens": 745565542.0, + "step": 19199 + }, + { + "epoch": 2.442437348937794, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.792900085449219, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.886502206325531, + "num_tokens": 745609066.0, + "step": 19200 + }, + { + "epoch": 2.4425645592163847, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.834577560424805, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8853679895401001, + "num_tokens": 745653514.0, + "step": 19201 + }, + { + "epoch": 2.4426917694949752, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.79073715209961, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9013084173202515, + "num_tokens": 745691339.0, + "step": 19202 + }, + { + "epoch": 2.4428189797735658, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.837108612060547, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8963462114334106, + "num_tokens": 745728313.0, + "step": 19203 + }, + { + "epoch": 2.4429461900521563, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.889496803283691, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.887582004070282, + "num_tokens": 745764259.0, + "step": 19204 + }, + { + "epoch": 2.443073400330747, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.781817436218262, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.9030112624168396, + "num_tokens": 745803797.0, + "step": 19205 + }, + { + "epoch": 2.4432006106093374, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.784656524658203, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8971216678619385, + "num_tokens": 745840593.0, + "step": 19206 + }, + { + "epoch": 2.443327820887928, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.837803840637207, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8966796398162842, + "num_tokens": 745879755.0, + "step": 19207 + }, + { + "epoch": 2.4434550311665184, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.734777450561523, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9155052900314331, + "num_tokens": 745912905.0, + "step": 19208 + }, + { + "epoch": 2.443582241445109, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.904109954833984, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8916282653808594, + "num_tokens": 745948051.0, + "step": 19209 + }, + { + "epoch": 2.4437094517236995, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.80435562133789, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.9016181230545044, + "num_tokens": 745983088.0, + "step": 19210 + }, + { + "epoch": 2.4438366620022896, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.773530006408691, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8957960605621338, + "num_tokens": 746023546.0, + "step": 19211 + }, + { + "epoch": 2.4439638722808805, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.758867263793945, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8923200368881226, + "num_tokens": 746063539.0, + "step": 19212 + }, + { + "epoch": 2.4440910825594706, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.718632698059082, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9111638069152832, + "num_tokens": 746096213.0, + "step": 19213 + }, + { + "epoch": 2.444218292838061, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.754279136657715, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8987213373184204, + "num_tokens": 746135307.0, + "step": 19214 + }, + { + "epoch": 2.4443455031166517, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.767022132873535, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8877658843994141, + "num_tokens": 746171730.0, + "step": 19215 + }, + { + "epoch": 2.444472713395242, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.756669044494629, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.9039484858512878, + "num_tokens": 746209834.0, + "step": 19216 + }, + { + "epoch": 2.4445999236738327, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.754583358764648, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8988440632820129, + "num_tokens": 746250613.0, + "step": 19217 + }, + { + "epoch": 2.4447271339524232, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.835134506225586, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8838580846786499, + "num_tokens": 746292105.0, + "step": 19218 + }, + { + "epoch": 2.4448543442310138, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.73531723022461, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.9002861976623535, + "num_tokens": 746334447.0, + "step": 19219 + }, + { + "epoch": 2.4449815545096043, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.807397842407227, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8973116874694824, + "num_tokens": 746371919.0, + "step": 19220 + }, + { + "epoch": 2.445108764788195, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.818918228149414, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8861982822418213, + "num_tokens": 746410398.0, + "step": 19221 + }, + { + "epoch": 2.4452359750667854, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.766593933105469, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9064950942993164, + "num_tokens": 746450847.0, + "step": 19222 + }, + { + "epoch": 2.445363185345376, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.6781005859375, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8982591032981873, + "num_tokens": 746495651.0, + "step": 19223 + }, + { + "epoch": 2.4454903956239664, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.797807693481445, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8968065977096558, + "num_tokens": 746535954.0, + "step": 19224 + }, + { + "epoch": 2.445617605902557, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.779040336608887, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.896632969379425, + "num_tokens": 746577357.0, + "step": 19225 + }, + { + "epoch": 2.4457448161811475, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.781037330627441, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.908195972442627, + "num_tokens": 746611421.0, + "step": 19226 + }, + { + "epoch": 2.445872026459738, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.811056137084961, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8958353996276855, + "num_tokens": 746646850.0, + "step": 19227 + }, + { + "epoch": 2.4459992367383285, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.686410903930664, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9014908671379089, + "num_tokens": 746685382.0, + "step": 19228 + }, + { + "epoch": 2.446126447016919, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.695699691772461, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9158320426940918, + "num_tokens": 746725257.0, + "step": 19229 + }, + { + "epoch": 2.4462536572955096, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.717331886291504, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8925483226776123, + "num_tokens": 746767715.0, + "step": 19230 + }, + { + "epoch": 2.4463808675741, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.77920150756836, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8972921371459961, + "num_tokens": 746800897.0, + "step": 19231 + }, + { + "epoch": 2.4465080778526906, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.790644645690918, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.900335967540741, + "num_tokens": 746840033.0, + "step": 19232 + }, + { + "epoch": 2.446635288131281, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.708009719848633, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9106043577194214, + "num_tokens": 746878564.0, + "step": 19233 + }, + { + "epoch": 2.4467624984098717, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.750448226928711, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8966460227966309, + "num_tokens": 746919861.0, + "step": 19234 + }, + { + "epoch": 2.4468897086884622, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.870246887207031, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8934080004692078, + "num_tokens": 746957902.0, + "step": 19235 + }, + { + "epoch": 2.4470169189670523, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.665140151977539, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9118525981903076, + "num_tokens": 746995815.0, + "step": 19236 + }, + { + "epoch": 2.4471441292456433, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.867191314697266, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.9149974584579468, + "num_tokens": 747032245.0, + "step": 19237 + }, + { + "epoch": 2.4472713395242334, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.657864570617676, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9127417802810669, + "num_tokens": 747071976.0, + "step": 19238 + }, + { + "epoch": 2.447398549802824, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.92851734161377, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8860528469085693, + "num_tokens": 747108363.0, + "step": 19239 + }, + { + "epoch": 2.4475257600814144, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.737971305847168, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.899810791015625, + "num_tokens": 747152012.0, + "step": 19240 + }, + { + "epoch": 2.447652970360005, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.857426643371582, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8968711495399475, + "num_tokens": 747183118.0, + "step": 19241 + }, + { + "epoch": 2.4477801806385955, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.788825988769531, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.901064395904541, + "num_tokens": 747220336.0, + "step": 19242 + }, + { + "epoch": 2.447907390917186, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.694289207458496, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.902944803237915, + "num_tokens": 747263948.0, + "step": 19243 + }, + { + "epoch": 2.4480346011957765, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.700142860412598, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.9040976762771606, + "num_tokens": 747302378.0, + "step": 19244 + }, + { + "epoch": 2.448161811474367, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.730534553527832, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.9139602184295654, + "num_tokens": 747339542.0, + "step": 19245 + }, + { + "epoch": 2.4482890217529576, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.700180053710938, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8961807489395142, + "num_tokens": 747383459.0, + "step": 19246 + }, + { + "epoch": 2.448416232031548, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.828939437866211, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.885897159576416, + "num_tokens": 747421341.0, + "step": 19247 + }, + { + "epoch": 2.4485434423101387, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.732611656188965, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8886898756027222, + "num_tokens": 747463385.0, + "step": 19248 + }, + { + "epoch": 2.448670652588729, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.669471740722656, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.908279538154602, + "num_tokens": 747503518.0, + "step": 19249 + }, + { + "epoch": 2.4487978628673197, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.78693962097168, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8958972096443176, + "num_tokens": 747540132.0, + "step": 19250 + }, + { + "epoch": 2.4489250731459102, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.743525505065918, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.9024937748908997, + "num_tokens": 747582035.0, + "step": 19251 + }, + { + "epoch": 2.4490522834245008, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.79952621459961, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8942012786865234, + "num_tokens": 747621295.0, + "step": 19252 + }, + { + "epoch": 2.4491794937030913, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.9888277053833, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.9015733599662781, + "num_tokens": 747662068.0, + "step": 19253 + }, + { + "epoch": 2.449306703981682, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.694510459899902, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8998073935508728, + "num_tokens": 747701517.0, + "step": 19254 + }, + { + "epoch": 2.4494339142602723, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.78406810760498, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8966972231864929, + "num_tokens": 747741620.0, + "step": 19255 + }, + { + "epoch": 2.449561124538863, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.796278953552246, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8913273811340332, + "num_tokens": 747782867.0, + "step": 19256 + }, + { + "epoch": 2.4496883348174534, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.894062042236328, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8939014673233032, + "num_tokens": 747819917.0, + "step": 19257 + }, + { + "epoch": 2.449815545096044, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.691948890686035, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9111298322677612, + "num_tokens": 747860070.0, + "step": 19258 + }, + { + "epoch": 2.4499427553746345, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.75205135345459, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8964886665344238, + "num_tokens": 747900355.0, + "step": 19259 + }, + { + "epoch": 2.450069965653225, + "ewc_loss": 0.0203857421875, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1622905731201172e-05, + "grad_norm": 9.704846382141113, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.89288330078125, + "num_tokens": 747941340.0, + "step": 19260 + }, + { + "epoch": 2.450197175931815, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.791581153869629, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8942294716835022, + "num_tokens": 747982477.0, + "step": 19261 + }, + { + "epoch": 2.450324386210406, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.71156120300293, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9081764221191406, + "num_tokens": 748025564.0, + "step": 19262 + }, + { + "epoch": 2.450451596488996, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.926209449768066, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9041200876235962, + "num_tokens": 748056356.0, + "step": 19263 + }, + { + "epoch": 2.4505788067675867, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.74728775024414, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8957459926605225, + "num_tokens": 748099751.0, + "step": 19264 + }, + { + "epoch": 2.450706017046177, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.801581382751465, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8900868892669678, + "num_tokens": 748140509.0, + "step": 19265 + }, + { + "epoch": 2.4508332273247677, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.672531127929688, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9077186584472656, + "num_tokens": 748182942.0, + "step": 19266 + }, + { + "epoch": 2.4509604376033582, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.817604064941406, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.903190016746521, + "num_tokens": 748219947.0, + "step": 19267 + }, + { + "epoch": 2.4510876478819488, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.814547538757324, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8946892023086548, + "num_tokens": 748259866.0, + "step": 19268 + }, + { + "epoch": 2.4512148581605393, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.826904296875, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9055935144424438, + "num_tokens": 748292813.0, + "step": 19269 + }, + { + "epoch": 2.45134206843913, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1682510375976562e-05, + "grad_norm": 9.782696723937988, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9127032160758972, + "num_tokens": 748329455.0, + "step": 19270 + }, + { + "epoch": 2.4514692787177204, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.72928237915039, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9063013792037964, + "num_tokens": 748366197.0, + "step": 19271 + }, + { + "epoch": 2.451596488996311, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.758976936340332, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.9038331508636475, + "num_tokens": 748409454.0, + "step": 19272 + }, + { + "epoch": 2.4517236992749014, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.788167953491211, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8989135026931763, + "num_tokens": 748447408.0, + "step": 19273 + }, + { + "epoch": 2.451850909553492, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.733646392822266, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.906126081943512, + "num_tokens": 748487532.0, + "step": 19274 + }, + { + "epoch": 2.4519781198320825, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.748781204223633, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8992741107940674, + "num_tokens": 748527825.0, + "step": 19275 + }, + { + "epoch": 2.452105330110673, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.795440673828125, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.9013264179229736, + "num_tokens": 748566664.0, + "step": 19276 + }, + { + "epoch": 2.4522325403892635, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.855219841003418, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8788672685623169, + "num_tokens": 748605370.0, + "step": 19277 + }, + { + "epoch": 2.452359750667854, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.839578628540039, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9043725728988647, + "num_tokens": 748636858.0, + "step": 19278 + }, + { + "epoch": 2.4524869609464446, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.697894096374512, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.9022442102432251, + "num_tokens": 748683744.0, + "step": 19279 + }, + { + "epoch": 2.452614171225035, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.850378036499023, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8887470364570618, + "num_tokens": 748718791.0, + "step": 19280 + }, + { + "epoch": 2.4527413815036256, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.737176895141602, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8969522714614868, + "num_tokens": 748759099.0, + "step": 19281 + }, + { + "epoch": 2.452868591782216, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.850570678710938, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8838528394699097, + "num_tokens": 748802365.0, + "step": 19282 + }, + { + "epoch": 2.4529958020608067, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.78403377532959, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8914921283721924, + "num_tokens": 748843262.0, + "step": 19283 + }, + { + "epoch": 2.453123012339397, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.722888946533203, + "learning_rate": 1e-06, + "loss": 0.2315, + "mean_token_accuracy": 0.9208970665931702, + "num_tokens": 748877668.0, + "step": 19284 + }, + { + "epoch": 2.4532502226179878, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.75515079498291, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9052044749259949, + "num_tokens": 748917050.0, + "step": 19285 + }, + { + "epoch": 2.453377432896578, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.707207679748535, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8969452381134033, + "num_tokens": 748965403.0, + "step": 19286 + }, + { + "epoch": 2.4535046431751684, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.691252708435059, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.9011428356170654, + "num_tokens": 749006986.0, + "step": 19287 + }, + { + "epoch": 2.453631853453759, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.805089950561523, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8928619027137756, + "num_tokens": 749052362.0, + "step": 19288 + }, + { + "epoch": 2.4537590637323494, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.794940948486328, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9131033420562744, + "num_tokens": 749087768.0, + "step": 19289 + }, + { + "epoch": 2.45388627401094, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.714872360229492, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.9108225107192993, + "num_tokens": 749129271.0, + "step": 19290 + }, + { + "epoch": 2.4540134842895305, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.780394554138184, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9067379236221313, + "num_tokens": 749170295.0, + "step": 19291 + }, + { + "epoch": 2.454140694568121, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.733612060546875, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.9008821249008179, + "num_tokens": 749209850.0, + "step": 19292 + }, + { + "epoch": 2.4542679048467115, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.916658401489258, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8967928886413574, + "num_tokens": 749245083.0, + "step": 19293 + }, + { + "epoch": 2.454395115125302, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.935590744018555, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.890221357345581, + "num_tokens": 749278287.0, + "step": 19294 + }, + { + "epoch": 2.4545223254038926, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.831514358520508, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8939696550369263, + "num_tokens": 749319559.0, + "step": 19295 + }, + { + "epoch": 2.454649535682483, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.867740631103516, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9013167023658752, + "num_tokens": 749353115.0, + "step": 19296 + }, + { + "epoch": 2.4547767459610736, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.820056915283203, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8861976861953735, + "num_tokens": 749396237.0, + "step": 19297 + }, + { + "epoch": 2.454903956239664, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.89123821258545, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.9043692350387573, + "num_tokens": 749427019.0, + "step": 19298 + }, + { + "epoch": 2.4550311665182547, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.81663990020752, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8994697332382202, + "num_tokens": 749461361.0, + "step": 19299 + }, + { + "epoch": 2.4551583767968452, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.79224681854248, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.904352605342865, + "num_tokens": 749496529.0, + "step": 19300 + }, + { + "epoch": 2.4552855870754358, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.798238754272461, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8996603488922119, + "num_tokens": 749533025.0, + "step": 19301 + }, + { + "epoch": 2.4554127973540263, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.795747756958008, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8949590921401978, + "num_tokens": 749573137.0, + "step": 19302 + }, + { + "epoch": 2.455540007632617, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.796863555908203, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8887742161750793, + "num_tokens": 749613734.0, + "step": 19303 + }, + { + "epoch": 2.4556672179112073, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.814031600952148, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9069296717643738, + "num_tokens": 749651698.0, + "step": 19304 + }, + { + "epoch": 2.455794428189798, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.832555770874023, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8979530930519104, + "num_tokens": 749687336.0, + "step": 19305 + }, + { + "epoch": 2.4559216384683884, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.763433456420898, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8922576904296875, + "num_tokens": 749729269.0, + "step": 19306 + }, + { + "epoch": 2.456048848746979, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.815674781799316, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8880943059921265, + "num_tokens": 749770021.0, + "step": 19307 + }, + { + "epoch": 2.4561760590255695, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.799337387084961, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8948581218719482, + "num_tokens": 749805852.0, + "step": 19308 + }, + { + "epoch": 2.4563032693041595, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.787235260009766, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8999382853507996, + "num_tokens": 749841472.0, + "step": 19309 + }, + { + "epoch": 2.4564304795827505, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.786026954650879, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.915346086025238, + "num_tokens": 749874987.0, + "step": 19310 + }, + { + "epoch": 2.4565576898613406, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.745672225952148, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9082027673721313, + "num_tokens": 749913978.0, + "step": 19311 + }, + { + "epoch": 2.456684900139931, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.831435203552246, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9071223735809326, + "num_tokens": 749955995.0, + "step": 19312 + }, + { + "epoch": 2.4568121104185217, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.813612937927246, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.902090847492218, + "num_tokens": 749988327.0, + "step": 19313 + }, + { + "epoch": 2.456939320697112, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.806695938110352, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.896451473236084, + "num_tokens": 750026936.0, + "step": 19314 + }, + { + "epoch": 2.4570665309757027, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.91259765625, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.889892041683197, + "num_tokens": 750059704.0, + "step": 19315 + }, + { + "epoch": 2.4571937412542932, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.7831392288208, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9029240012168884, + "num_tokens": 750092880.0, + "step": 19316 + }, + { + "epoch": 2.4573209515328838, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.820710182189941, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.905921220779419, + "num_tokens": 750129336.0, + "step": 19317 + }, + { + "epoch": 2.4574481618114743, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.868544578552246, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.895180881023407, + "num_tokens": 750163131.0, + "step": 19318 + }, + { + "epoch": 2.457575372090065, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.798258781433105, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8919816613197327, + "num_tokens": 750204850.0, + "step": 19319 + }, + { + "epoch": 2.4577025823686554, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.735400199890137, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8952440023422241, + "num_tokens": 750247695.0, + "step": 19320 + }, + { + "epoch": 2.457829792647246, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.804023742675781, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8945925831794739, + "num_tokens": 750289582.0, + "step": 19321 + }, + { + "epoch": 2.4579570029258364, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.78327465057373, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.896803617477417, + "num_tokens": 750324457.0, + "step": 19322 + }, + { + "epoch": 2.458084213204427, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.728492736816406, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8993308544158936, + "num_tokens": 750367104.0, + "step": 19323 + }, + { + "epoch": 2.4582114234830175, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.767345428466797, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8914670944213867, + "num_tokens": 750409880.0, + "step": 19324 + }, + { + "epoch": 2.458338633761608, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.85154914855957, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.895400881767273, + "num_tokens": 750446720.0, + "step": 19325 + }, + { + "epoch": 2.4584658440401985, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.636871337890625, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9106276035308838, + "num_tokens": 750490505.0, + "step": 19326 + }, + { + "epoch": 2.458593054318789, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.784978866577148, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.9077849984169006, + "num_tokens": 750533634.0, + "step": 19327 + }, + { + "epoch": 2.4587202645973796, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.819721221923828, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8977437019348145, + "num_tokens": 750574030.0, + "step": 19328 + }, + { + "epoch": 2.45884747487597, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.817206382751465, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8927210569381714, + "num_tokens": 750615111.0, + "step": 19329 + }, + { + "epoch": 2.4589746851545606, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.68819522857666, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.9020117521286011, + "num_tokens": 750654503.0, + "step": 19330 + }, + { + "epoch": 2.459101895433151, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.84524154663086, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.9023194909095764, + "num_tokens": 750687767.0, + "step": 19331 + }, + { + "epoch": 2.4592291057117417, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.74299144744873, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8980597853660583, + "num_tokens": 750729567.0, + "step": 19332 + }, + { + "epoch": 2.459356315990332, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.92862319946289, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8933067321777344, + "num_tokens": 750763199.0, + "step": 19333 + }, + { + "epoch": 2.4594835262689223, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.787132263183594, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8996032476425171, + "num_tokens": 750804323.0, + "step": 19334 + }, + { + "epoch": 2.4596107365475133, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.643377304077148, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9101884365081787, + "num_tokens": 750846740.0, + "step": 19335 + }, + { + "epoch": 2.4597379468261034, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.886688232421875, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8842741250991821, + "num_tokens": 750883158.0, + "step": 19336 + }, + { + "epoch": 2.459865157104694, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.780349731445312, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.89036625623703, + "num_tokens": 750924957.0, + "step": 19337 + }, + { + "epoch": 2.4599923673832844, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.829376220703125, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.9029765725135803, + "num_tokens": 750958947.0, + "step": 19338 + }, + { + "epoch": 2.460119577661875, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.789040565490723, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.898809015750885, + "num_tokens": 751002915.0, + "step": 19339 + }, + { + "epoch": 2.4602467879404655, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.777386665344238, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8919427394866943, + "num_tokens": 751043694.0, + "step": 19340 + }, + { + "epoch": 2.460373998219056, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.835100173950195, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8977465629577637, + "num_tokens": 751083263.0, + "step": 19341 + }, + { + "epoch": 2.4605012084976465, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.831365585327148, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.903977632522583, + "num_tokens": 751119689.0, + "step": 19342 + }, + { + "epoch": 2.460628418776237, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.735255241394043, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.9166610240936279, + "num_tokens": 751158545.0, + "step": 19343 + }, + { + "epoch": 2.4607556290548276, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.97441577911377, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.9011367559432983, + "num_tokens": 751190417.0, + "step": 19344 + }, + { + "epoch": 2.460882839333418, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.835638046264648, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.9034016728401184, + "num_tokens": 751221966.0, + "step": 19345 + }, + { + "epoch": 2.4610100496120086, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.76688003540039, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.906694769859314, + "num_tokens": 751259476.0, + "step": 19346 + }, + { + "epoch": 2.461137259890599, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.814359664916992, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8954789042472839, + "num_tokens": 751297468.0, + "step": 19347 + }, + { + "epoch": 2.4612644701691897, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.73105525970459, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9094139337539673, + "num_tokens": 751339189.0, + "step": 19348 + }, + { + "epoch": 2.4613916804477802, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.72083568572998, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.908003568649292, + "num_tokens": 751382150.0, + "step": 19349 + }, + { + "epoch": 2.4615188907263708, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.831856727600098, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8913699388504028, + "num_tokens": 751421031.0, + "step": 19350 + }, + { + "epoch": 2.4616461010049613, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.890490531921387, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8921259045600891, + "num_tokens": 751457769.0, + "step": 19351 + }, + { + "epoch": 2.461773311283552, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.839993476867676, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.8833101987838745, + "num_tokens": 751491742.0, + "step": 19352 + }, + { + "epoch": 2.4619005215621423, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.738687515258789, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.884202241897583, + "num_tokens": 751536150.0, + "step": 19353 + }, + { + "epoch": 2.462027731840733, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.779515266418457, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8873769640922546, + "num_tokens": 751578928.0, + "step": 19354 + }, + { + "epoch": 2.4621549421193234, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.807374000549316, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8970456123352051, + "num_tokens": 751618566.0, + "step": 19355 + }, + { + "epoch": 2.462282152397914, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.753044128417969, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9091063737869263, + "num_tokens": 751656252.0, + "step": 19356 + }, + { + "epoch": 2.4624093626765045, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.739461898803711, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8830958604812622, + "num_tokens": 751697826.0, + "step": 19357 + }, + { + "epoch": 2.462536572955095, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.732253074645996, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8914356827735901, + "num_tokens": 751741622.0, + "step": 19358 + }, + { + "epoch": 2.462663783233685, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 10.027885437011719, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8764036893844604, + "num_tokens": 751774236.0, + "step": 19359 + }, + { + "epoch": 2.462790993512276, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.777032852172852, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.9047666788101196, + "num_tokens": 751809430.0, + "step": 19360 + }, + { + "epoch": 2.462918203790866, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.862262725830078, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8922370076179504, + "num_tokens": 751845438.0, + "step": 19361 + }, + { + "epoch": 2.4630454140694567, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.758488655090332, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8853234648704529, + "num_tokens": 751885777.0, + "step": 19362 + }, + { + "epoch": 2.463172624348047, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.782022476196289, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9054461717605591, + "num_tokens": 751924219.0, + "step": 19363 + }, + { + "epoch": 2.4632998346266377, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.730381965637207, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8927055597305298, + "num_tokens": 751967490.0, + "step": 19364 + }, + { + "epoch": 2.4634270449052282, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.870423316955566, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.909118115901947, + "num_tokens": 752006967.0, + "step": 19365 + }, + { + "epoch": 2.4635542551838188, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.838993072509766, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9078898429870605, + "num_tokens": 752038551.0, + "step": 19366 + }, + { + "epoch": 2.4636814654624093, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.6909761428833, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.9040514230728149, + "num_tokens": 752080812.0, + "step": 19367 + }, + { + "epoch": 2.463808675741, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.754682540893555, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8937872648239136, + "num_tokens": 752124785.0, + "step": 19368 + }, + { + "epoch": 2.4639358860195903, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.778899192810059, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8855804204940796, + "num_tokens": 752171141.0, + "step": 19369 + }, + { + "epoch": 2.464063096298181, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.801993370056152, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8967611789703369, + "num_tokens": 752211750.0, + "step": 19370 + }, + { + "epoch": 2.4641903065767714, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.698989868164062, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9173868298530579, + "num_tokens": 752251046.0, + "step": 19371 + }, + { + "epoch": 2.464317516855362, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.859183311462402, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8988973498344421, + "num_tokens": 752288994.0, + "step": 19372 + }, + { + "epoch": 2.4644447271339525, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.845856666564941, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.9004860520362854, + "num_tokens": 752327110.0, + "step": 19373 + }, + { + "epoch": 2.464571937412543, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.814864158630371, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8911961913108826, + "num_tokens": 752365656.0, + "step": 19374 + }, + { + "epoch": 2.4646991476911335, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.811113357543945, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8969816565513611, + "num_tokens": 752407291.0, + "step": 19375 + }, + { + "epoch": 2.464826357969724, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.727167129516602, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9131698608398438, + "num_tokens": 752447360.0, + "step": 19376 + }, + { + "epoch": 2.4649535682483146, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.75469970703125, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9085114002227783, + "num_tokens": 752482055.0, + "step": 19377 + }, + { + "epoch": 2.465080778526905, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.805006980895996, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.9005527496337891, + "num_tokens": 752521943.0, + "step": 19378 + }, + { + "epoch": 2.4652079888054956, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.782354354858398, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8933805227279663, + "num_tokens": 752568822.0, + "step": 19379 + }, + { + "epoch": 2.465335199084086, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.828407287597656, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9101331830024719, + "num_tokens": 752604202.0, + "step": 19380 + }, + { + "epoch": 2.4654624093626767, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.81174373626709, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.9008998870849609, + "num_tokens": 752641844.0, + "step": 19381 + }, + { + "epoch": 2.4655896196412668, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.756671905517578, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8875353336334229, + "num_tokens": 752684859.0, + "step": 19382 + }, + { + "epoch": 2.4657168299198577, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.736726760864258, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9023688435554504, + "num_tokens": 752721631.0, + "step": 19383 + }, + { + "epoch": 2.465844040198448, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.769875526428223, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8948484063148499, + "num_tokens": 752761581.0, + "step": 19384 + }, + { + "epoch": 2.4659712504770384, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.76451587677002, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8994876146316528, + "num_tokens": 752799620.0, + "step": 19385 + }, + { + "epoch": 2.466098460755629, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.880531311035156, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8874284625053406, + "num_tokens": 752835703.0, + "step": 19386 + }, + { + "epoch": 2.4662256710342194, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.844367980957031, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9057180285453796, + "num_tokens": 752866410.0, + "step": 19387 + }, + { + "epoch": 2.46635288131281, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.763504028320312, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.897660493850708, + "num_tokens": 752908472.0, + "step": 19388 + }, + { + "epoch": 2.4664800915914005, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.743971824645996, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9011447429656982, + "num_tokens": 752948686.0, + "step": 19389 + }, + { + "epoch": 2.466607301869991, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.766324996948242, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9031702280044556, + "num_tokens": 752990572.0, + "step": 19390 + }, + { + "epoch": 2.4667345121485815, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.7210693359375, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.9018977880477905, + "num_tokens": 753034376.0, + "step": 19391 + }, + { + "epoch": 2.466861722427172, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.715045928955078, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9137445092201233, + "num_tokens": 753077659.0, + "step": 19392 + }, + { + "epoch": 2.4669889327057626, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.811426162719727, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8937874436378479, + "num_tokens": 753116238.0, + "step": 19393 + }, + { + "epoch": 2.467116142984353, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.657472610473633, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.9064593315124512, + "num_tokens": 753162412.0, + "step": 19394 + }, + { + "epoch": 2.4672433532629436, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.831997871398926, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8868968486785889, + "num_tokens": 753202144.0, + "step": 19395 + }, + { + "epoch": 2.467370563541534, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.862300872802734, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8923072814941406, + "num_tokens": 753240589.0, + "step": 19396 + }, + { + "epoch": 2.4674977738201247, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.835415840148926, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.881321132183075, + "num_tokens": 753281177.0, + "step": 19397 + }, + { + "epoch": 2.4676249840987152, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.7257719039917, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9005416631698608, + "num_tokens": 753320273.0, + "step": 19398 + }, + { + "epoch": 2.4677521943773058, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.855067253112793, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8775811195373535, + "num_tokens": 753355089.0, + "step": 19399 + }, + { + "epoch": 2.4678794046558963, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.819721221923828, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.899929940700531, + "num_tokens": 753390175.0, + "step": 19400 + }, + { + "epoch": 2.468006614934487, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.731056213378906, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8962380290031433, + "num_tokens": 753431225.0, + "step": 19401 + }, + { + "epoch": 2.4681338252130773, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.668126106262207, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9137677550315857, + "num_tokens": 753473438.0, + "step": 19402 + }, + { + "epoch": 2.468261035491668, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.79743766784668, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8925148248672485, + "num_tokens": 753514254.0, + "step": 19403 + }, + { + "epoch": 2.4683882457702584, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.758511543273926, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.905731201171875, + "num_tokens": 753553489.0, + "step": 19404 + }, + { + "epoch": 2.468515456048849, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.756998062133789, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.9047552347183228, + "num_tokens": 753591125.0, + "step": 19405 + }, + { + "epoch": 2.4686426663274394, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.741676330566406, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8851255178451538, + "num_tokens": 753632937.0, + "step": 19406 + }, + { + "epoch": 2.4687698766060295, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.712775230407715, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.9060537815093994, + "num_tokens": 753674469.0, + "step": 19407 + }, + { + "epoch": 2.4688970868846205, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.783841133117676, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.9038352370262146, + "num_tokens": 753714157.0, + "step": 19408 + }, + { + "epoch": 2.4690242971632106, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.834890365600586, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8940931558609009, + "num_tokens": 753750899.0, + "step": 19409 + }, + { + "epoch": 2.469151507441801, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.888622283935547, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8882286548614502, + "num_tokens": 753785896.0, + "step": 19410 + }, + { + "epoch": 2.4692787177203916, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.756760597229004, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8993148803710938, + "num_tokens": 753827267.0, + "step": 19411 + }, + { + "epoch": 2.469405927998982, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.954813003540039, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.891412615776062, + "num_tokens": 753858671.0, + "step": 19412 + }, + { + "epoch": 2.4695331382775727, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.777106285095215, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.912092387676239, + "num_tokens": 753897024.0, + "step": 19413 + }, + { + "epoch": 2.4696603485561632, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.830066680908203, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8868141174316406, + "num_tokens": 753935658.0, + "step": 19414 + }, + { + "epoch": 2.4697875588347538, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.835533142089844, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9037925004959106, + "num_tokens": 753971874.0, + "step": 19415 + }, + { + "epoch": 2.4699147691133443, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.711160659790039, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.9024277925491333, + "num_tokens": 754019639.0, + "step": 19416 + }, + { + "epoch": 2.470041979391935, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.789429664611816, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.9043893814086914, + "num_tokens": 754062457.0, + "step": 19417 + }, + { + "epoch": 2.4701691896705253, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.829882621765137, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9054838418960571, + "num_tokens": 754096610.0, + "step": 19418 + }, + { + "epoch": 2.470296399949116, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.783988952636719, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.9079914093017578, + "num_tokens": 754131897.0, + "step": 19419 + }, + { + "epoch": 2.4704236102277064, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.80384635925293, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.9071915149688721, + "num_tokens": 754171302.0, + "step": 19420 + }, + { + "epoch": 2.470550820506297, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.82655143737793, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8886297941207886, + "num_tokens": 754213653.0, + "step": 19421 + }, + { + "epoch": 2.4706780307848875, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.846709251403809, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8845407366752625, + "num_tokens": 754252349.0, + "step": 19422 + }, + { + "epoch": 2.470805241063478, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.746002197265625, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.9129303693771362, + "num_tokens": 754290208.0, + "step": 19423 + }, + { + "epoch": 2.4709324513420685, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.82648754119873, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.9036177396774292, + "num_tokens": 754329833.0, + "step": 19424 + }, + { + "epoch": 2.471059661620659, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.801717758178711, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.895901083946228, + "num_tokens": 754374504.0, + "step": 19425 + }, + { + "epoch": 2.4711868718992496, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.760635375976562, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.9020488262176514, + "num_tokens": 754415688.0, + "step": 19426 + }, + { + "epoch": 2.47131408217784, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.935612678527832, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8936495780944824, + "num_tokens": 754449951.0, + "step": 19427 + }, + { + "epoch": 2.4714412924564306, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.76326847076416, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9063340425491333, + "num_tokens": 754484255.0, + "step": 19428 + }, + { + "epoch": 2.471568502735021, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.869959831237793, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8983016014099121, + "num_tokens": 754522406.0, + "step": 19429 + }, + { + "epoch": 2.4716957130136117, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.78906536102295, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.9025864601135254, + "num_tokens": 754557743.0, + "step": 19430 + }, + { + "epoch": 2.471822923292202, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.79041576385498, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.9017879962921143, + "num_tokens": 754592437.0, + "step": 19431 + }, + { + "epoch": 2.4719501335707923, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.73497486114502, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.9005070328712463, + "num_tokens": 754637212.0, + "step": 19432 + }, + { + "epoch": 2.4720773438493833, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.7567720413208, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9043892621994019, + "num_tokens": 754675807.0, + "step": 19433 + }, + { + "epoch": 2.4722045541279734, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.84691047668457, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8929716944694519, + "num_tokens": 754711525.0, + "step": 19434 + }, + { + "epoch": 2.472331764406564, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.702171325683594, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8944970965385437, + "num_tokens": 754755687.0, + "step": 19435 + }, + { + "epoch": 2.4724589746851544, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.878995895385742, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8974023461341858, + "num_tokens": 754793525.0, + "step": 19436 + }, + { + "epoch": 2.472586184963745, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.952136993408203, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8909279108047485, + "num_tokens": 754827539.0, + "step": 19437 + }, + { + "epoch": 2.4727133952423355, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.75282096862793, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8979817628860474, + "num_tokens": 754863561.0, + "step": 19438 + }, + { + "epoch": 2.472840605520926, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.68491268157959, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9056937098503113, + "num_tokens": 754905655.0, + "step": 19439 + }, + { + "epoch": 2.4729678157995165, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.719525337219238, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9125046730041504, + "num_tokens": 754946331.0, + "step": 19440 + }, + { + "epoch": 2.473095026078107, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.812671661376953, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8930323123931885, + "num_tokens": 754985494.0, + "step": 19441 + }, + { + "epoch": 2.4732222363566976, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.926684379577637, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8819295763969421, + "num_tokens": 755023072.0, + "step": 19442 + }, + { + "epoch": 2.473349446635288, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.846879005432129, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8942196369171143, + "num_tokens": 755058851.0, + "step": 19443 + }, + { + "epoch": 2.4734766569138786, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.76939868927002, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8957587480545044, + "num_tokens": 755100821.0, + "step": 19444 + }, + { + "epoch": 2.473603867192469, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.876408576965332, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8904964327812195, + "num_tokens": 755137094.0, + "step": 19445 + }, + { + "epoch": 2.4737310774710597, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.673738479614258, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.9065855741500854, + "num_tokens": 755185226.0, + "step": 19446 + }, + { + "epoch": 2.47385828774965, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.846901893615723, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.905643105506897, + "num_tokens": 755220416.0, + "step": 19447 + }, + { + "epoch": 2.4739854980282407, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.755396842956543, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8978263139724731, + "num_tokens": 755262356.0, + "step": 19448 + }, + { + "epoch": 2.4741127083068313, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.863258361816406, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.9001083374023438, + "num_tokens": 755297060.0, + "step": 19449 + }, + { + "epoch": 2.474239918585422, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.770257949829102, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8923150300979614, + "num_tokens": 755337376.0, + "step": 19450 + }, + { + "epoch": 2.4743671288640123, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.774863243103027, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8948930501937866, + "num_tokens": 755375554.0, + "step": 19451 + }, + { + "epoch": 2.474494339142603, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.8355712890625, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9031237959861755, + "num_tokens": 755408627.0, + "step": 19452 + }, + { + "epoch": 2.4746215494211934, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.715986251831055, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9042627811431885, + "num_tokens": 755452206.0, + "step": 19453 + }, + { + "epoch": 2.474748759699784, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.771687507629395, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8967766165733337, + "num_tokens": 755496617.0, + "step": 19454 + }, + { + "epoch": 2.4748759699783744, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.872267723083496, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8878204822540283, + "num_tokens": 755534965.0, + "step": 19455 + }, + { + "epoch": 2.475003180256965, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.850692749023438, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8941735625267029, + "num_tokens": 755569473.0, + "step": 19456 + }, + { + "epoch": 2.475130390535555, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.816680908203125, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8994009494781494, + "num_tokens": 755605729.0, + "step": 19457 + }, + { + "epoch": 2.475257600814146, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.7738037109375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.838768005371094, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9073096513748169, + "num_tokens": 755636860.0, + "step": 19458 + }, + { + "epoch": 2.475384811092736, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.762560844421387, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9096076488494873, + "num_tokens": 755678671.0, + "step": 19459 + }, + { + "epoch": 2.4755120213713266, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.805057525634766, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8955743312835693, + "num_tokens": 755717267.0, + "step": 19460 + }, + { + "epoch": 2.475639231649917, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.872291564941406, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.898439884185791, + "num_tokens": 755754607.0, + "step": 19461 + }, + { + "epoch": 2.4757664419285077, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.903051376342773, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8901984691619873, + "num_tokens": 755792631.0, + "step": 19462 + }, + { + "epoch": 2.4758936522070982, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.804216384887695, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8858273029327393, + "num_tokens": 755837763.0, + "step": 19463 + }, + { + "epoch": 2.4760208624856888, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.846600532531738, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8955060243606567, + "num_tokens": 755876170.0, + "step": 19464 + }, + { + "epoch": 2.4761480727642793, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.841795921325684, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.9003757834434509, + "num_tokens": 755915054.0, + "step": 19465 + }, + { + "epoch": 2.47627528304287, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.820154190063477, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.898402214050293, + "num_tokens": 755956047.0, + "step": 19466 + }, + { + "epoch": 2.4764024933214603, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.793517112731934, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9066914319992065, + "num_tokens": 755995044.0, + "step": 19467 + }, + { + "epoch": 2.476529703600051, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.806848526000977, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.9057156443595886, + "num_tokens": 756035509.0, + "step": 19468 + }, + { + "epoch": 2.4766569138786414, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.80258560180664, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.9074971079826355, + "num_tokens": 756069580.0, + "step": 19469 + }, + { + "epoch": 2.476784124157232, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.805933952331543, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.895248532295227, + "num_tokens": 756117699.0, + "step": 19470 + }, + { + "epoch": 2.4769113344358225, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.807209968566895, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8901089429855347, + "num_tokens": 756157556.0, + "step": 19471 + }, + { + "epoch": 2.477038544714413, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.821487426757812e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.76609992980957, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.9003095030784607, + "num_tokens": 756202274.0, + "step": 19472 + }, + { + "epoch": 2.4771657549930035, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.784433364868164, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9108239412307739, + "num_tokens": 756235822.0, + "step": 19473 + }, + { + "epoch": 2.477292965271594, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.891552925109863, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.9049654006958008, + "num_tokens": 756268554.0, + "step": 19474 + }, + { + "epoch": 2.4774201755501846, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.729154586791992, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.9002812504768372, + "num_tokens": 756311153.0, + "step": 19475 + }, + { + "epoch": 2.477547385828775, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.854349136352539, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.9024742245674133, + "num_tokens": 756349886.0, + "step": 19476 + }, + { + "epoch": 2.4776745961073656, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.838388442993164, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8962849378585815, + "num_tokens": 756387440.0, + "step": 19477 + }, + { + "epoch": 2.477801806385956, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.76697063446045, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9137595295906067, + "num_tokens": 756426826.0, + "step": 19478 + }, + { + "epoch": 2.4779290166645467, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.795368194580078, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8994737267494202, + "num_tokens": 756465118.0, + "step": 19479 + }, + { + "epoch": 2.4780562269431368, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.894235610961914, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8936849236488342, + "num_tokens": 756501791.0, + "step": 19480 + }, + { + "epoch": 2.4781834372217277, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.793149948120117, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8994401097297668, + "num_tokens": 756542737.0, + "step": 19481 + }, + { + "epoch": 2.478310647500318, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.762852668762207, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.9044452905654907, + "num_tokens": 756582704.0, + "step": 19482 + }, + { + "epoch": 2.4784378577789083, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.819392204284668, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8976448774337769, + "num_tokens": 756620832.0, + "step": 19483 + }, + { + "epoch": 2.478565068057499, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.899835586547852, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.9028955698013306, + "num_tokens": 756650989.0, + "step": 19484 + }, + { + "epoch": 2.4786922783360894, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.857068061828613, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8831367492675781, + "num_tokens": 756692604.0, + "step": 19485 + }, + { + "epoch": 2.47881948861468, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.799540519714355, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9047181606292725, + "num_tokens": 756727404.0, + "step": 19486 + }, + { + "epoch": 2.4789466988932705, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.888710975646973, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8936893939971924, + "num_tokens": 756764498.0, + "step": 19487 + }, + { + "epoch": 2.479073909171861, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.739893913269043, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9104467630386353, + "num_tokens": 756801398.0, + "step": 19488 + }, + { + "epoch": 2.4792011194504515, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.821988105773926, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8997398614883423, + "num_tokens": 756840805.0, + "step": 19489 + }, + { + "epoch": 2.479328329729042, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.83837890625, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8990024328231812, + "num_tokens": 756876271.0, + "step": 19490 + }, + { + "epoch": 2.4794555400076326, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.75012493133545, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.899491012096405, + "num_tokens": 756913943.0, + "step": 19491 + }, + { + "epoch": 2.479582750286223, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.849586486816406, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8931379914283752, + "num_tokens": 756953676.0, + "step": 19492 + }, + { + "epoch": 2.4797099605648136, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.818099975585938, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8974456787109375, + "num_tokens": 756988657.0, + "step": 19493 + }, + { + "epoch": 2.479837170843404, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.690135955810547, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9012621641159058, + "num_tokens": 757036546.0, + "step": 19494 + }, + { + "epoch": 2.4799643811219947, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.766707420349121, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.9093842506408691, + "num_tokens": 757077429.0, + "step": 19495 + }, + { + "epoch": 2.480091591400585, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.878795623779297, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9059673547744751, + "num_tokens": 757111363.0, + "step": 19496 + }, + { + "epoch": 2.4802188016791757, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.739461898803711, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8885557651519775, + "num_tokens": 757157487.0, + "step": 19497 + }, + { + "epoch": 2.4803460119577663, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.839905738830566, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8857824802398682, + "num_tokens": 757197247.0, + "step": 19498 + }, + { + "epoch": 2.480473222236357, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.781229019165039, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8861086964607239, + "num_tokens": 757245340.0, + "step": 19499 + }, + { + "epoch": 2.4806004325149473, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.780141830444336, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8926841020584106, + "num_tokens": 757288799.0, + "step": 19500 + }, + { + "epoch": 2.480727642793538, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.846200942993164, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8975507020950317, + "num_tokens": 757325478.0, + "step": 19501 + }, + { + "epoch": 2.4808548530721284, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.819522857666016, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8854197263717651, + "num_tokens": 757370527.0, + "step": 19502 + }, + { + "epoch": 2.480982063350719, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.775039672851562, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.9059575796127319, + "num_tokens": 757406688.0, + "step": 19503 + }, + { + "epoch": 2.4811092736293094, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.839863777160645, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9056078791618347, + "num_tokens": 757442344.0, + "step": 19504 + }, + { + "epoch": 2.4812364839078995, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.800601959228516, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8873200416564941, + "num_tokens": 757479116.0, + "step": 19505 + }, + { + "epoch": 2.4813636941864905, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.796875, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.9024145603179932, + "num_tokens": 757519667.0, + "step": 19506 + }, + { + "epoch": 2.4814909044650806, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.893872261047363, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8971478343009949, + "num_tokens": 757557538.0, + "step": 19507 + }, + { + "epoch": 2.481618114743671, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.821653366088867, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8959275484085083, + "num_tokens": 757595846.0, + "step": 19508 + }, + { + "epoch": 2.4817453250222616, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.79939079284668, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.9020165205001831, + "num_tokens": 757631185.0, + "step": 19509 + }, + { + "epoch": 2.481872535300852, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.885692596435547, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8951183557510376, + "num_tokens": 757664836.0, + "step": 19510 + }, + { + "epoch": 2.4819997455794427, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.78997802734375, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8981738090515137, + "num_tokens": 757703474.0, + "step": 19511 + }, + { + "epoch": 2.4821269558580332, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.74365520477295, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.910334050655365, + "num_tokens": 757737359.0, + "step": 19512 + }, + { + "epoch": 2.4822541661366238, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.872617721557617, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8980472087860107, + "num_tokens": 757773781.0, + "step": 19513 + }, + { + "epoch": 2.4823813764152143, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.750168800354004, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8997417092323303, + "num_tokens": 757809795.0, + "step": 19514 + }, + { + "epoch": 2.482508586693805, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.891307830810547, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8983472585678101, + "num_tokens": 757845335.0, + "step": 19515 + }, + { + "epoch": 2.4826357969723953, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.855953216552734, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.9003804922103882, + "num_tokens": 757879126.0, + "step": 19516 + }, + { + "epoch": 2.482763007250986, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.790289878845215, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.9081559181213379, + "num_tokens": 757916321.0, + "step": 19517 + }, + { + "epoch": 2.4828902175295764, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.85894775390625, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8984361290931702, + "num_tokens": 757951740.0, + "step": 19518 + }, + { + "epoch": 2.483017427808167, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.876378059387207, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8924343585968018, + "num_tokens": 757984883.0, + "step": 19519 + }, + { + "epoch": 2.4831446380867574, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.74786376953125, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9095379114151001, + "num_tokens": 758024510.0, + "step": 19520 + }, + { + "epoch": 2.483271848365348, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.779687881469727, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.9059609770774841, + "num_tokens": 758063615.0, + "step": 19521 + }, + { + "epoch": 2.4833990586439385, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.84879207611084, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8912090063095093, + "num_tokens": 758101398.0, + "step": 19522 + }, + { + "epoch": 2.483526268922529, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.857246398925781, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8935133814811707, + "num_tokens": 758138184.0, + "step": 19523 + }, + { + "epoch": 2.4836534792011196, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.842693328857422, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8874943256378174, + "num_tokens": 758174374.0, + "step": 19524 + }, + { + "epoch": 2.48378068947971, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.865937232971191, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8966633677482605, + "num_tokens": 758209823.0, + "step": 19525 + }, + { + "epoch": 2.4839078997583006, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.689369201660156, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.9032903909683228, + "num_tokens": 758255854.0, + "step": 19526 + }, + { + "epoch": 2.484035110036891, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.7079496383667, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9116193056106567, + "num_tokens": 758300777.0, + "step": 19527 + }, + { + "epoch": 2.4841623203154817, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.784711837768555, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9066470265388489, + "num_tokens": 758343892.0, + "step": 19528 + }, + { + "epoch": 2.484289530594072, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.829442024230957, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8984438180923462, + "num_tokens": 758381946.0, + "step": 19529 + }, + { + "epoch": 2.4844167408726623, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.834271430969238, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.900336742401123, + "num_tokens": 758417942.0, + "step": 19530 + }, + { + "epoch": 2.4845439511512533, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.714010238647461, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.896264910697937, + "num_tokens": 758460132.0, + "step": 19531 + }, + { + "epoch": 2.4846711614298433, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.83591365814209, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8905266523361206, + "num_tokens": 758502788.0, + "step": 19532 + }, + { + "epoch": 2.484798371708434, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.826825141906738, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8945528268814087, + "num_tokens": 758542086.0, + "step": 19533 + }, + { + "epoch": 2.4849255819870244, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.805435180664062, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8873900175094604, + "num_tokens": 758583817.0, + "step": 19534 + }, + { + "epoch": 2.485052792265615, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.832629203796387, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8904342651367188, + "num_tokens": 758621042.0, + "step": 19535 + }, + { + "epoch": 2.4851800025442055, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.844202041625977, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8888227939605713, + "num_tokens": 758657960.0, + "step": 19536 + }, + { + "epoch": 2.485307212822796, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.83524227142334, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8858289122581482, + "num_tokens": 758695061.0, + "step": 19537 + }, + { + "epoch": 2.4854344231013865, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.934000968933105, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8977956771850586, + "num_tokens": 758728378.0, + "step": 19538 + }, + { + "epoch": 2.485561633379977, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.786865234375, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.9004791975021362, + "num_tokens": 758766893.0, + "step": 19539 + }, + { + "epoch": 2.4856888436585676, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.896509170532227, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8881832361221313, + "num_tokens": 758801729.0, + "step": 19540 + }, + { + "epoch": 2.485816053937158, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.816313743591309, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9020683765411377, + "num_tokens": 758839869.0, + "step": 19541 + }, + { + "epoch": 2.4859432642157486, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.852952003479004, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8980751037597656, + "num_tokens": 758874707.0, + "step": 19542 + }, + { + "epoch": 2.486070474494339, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.74052906036377, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9101231098175049, + "num_tokens": 758921725.0, + "step": 19543 + }, + { + "epoch": 2.4861976847729297, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.755043029785156, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.9120059013366699, + "num_tokens": 758967461.0, + "step": 19544 + }, + { + "epoch": 2.48632489505152, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.920039176940918, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.9028222560882568, + "num_tokens": 759002284.0, + "step": 19545 + }, + { + "epoch": 2.4864521053301107, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.729570388793945, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.91164231300354, + "num_tokens": 759043505.0, + "step": 19546 + }, + { + "epoch": 2.4865793156087013, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.761474609375, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8994396924972534, + "num_tokens": 759086798.0, + "step": 19547 + }, + { + "epoch": 2.486706525887292, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.823219299316406, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8885197639465332, + "num_tokens": 759128923.0, + "step": 19548 + }, + { + "epoch": 2.4868337361658823, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.878811836242676, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8931496143341064, + "num_tokens": 759166305.0, + "step": 19549 + }, + { + "epoch": 2.486960946444473, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.763592720031738, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.9080457091331482, + "num_tokens": 759205249.0, + "step": 19550 + }, + { + "epoch": 2.4870881567230634, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.720338821411133, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.9040425419807434, + "num_tokens": 759250588.0, + "step": 19551 + }, + { + "epoch": 2.487215367001654, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.927882194519043, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8896898627281189, + "num_tokens": 759285213.0, + "step": 19552 + }, + { + "epoch": 2.4873425772802444, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.844001770019531, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8951635956764221, + "num_tokens": 759319615.0, + "step": 19553 + }, + { + "epoch": 2.487469787558835, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.794371604919434, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9044979810714722, + "num_tokens": 759357880.0, + "step": 19554 + }, + { + "epoch": 2.487596997837425, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.794445037841797, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8962862491607666, + "num_tokens": 759399737.0, + "step": 19555 + }, + { + "epoch": 2.487724208116016, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.730791091918945, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.9021697640419006, + "num_tokens": 759440524.0, + "step": 19556 + }, + { + "epoch": 2.487851418394606, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.988566398620605, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8925085067749023, + "num_tokens": 759474267.0, + "step": 19557 + }, + { + "epoch": 2.4879786286731966, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.805476188659668, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8940417766571045, + "num_tokens": 759512031.0, + "step": 19558 + }, + { + "epoch": 2.488105838951787, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.848116874694824, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9024361371994019, + "num_tokens": 759551836.0, + "step": 19559 + }, + { + "epoch": 2.4882330492303777, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.831664085388184, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8995428085327148, + "num_tokens": 759589950.0, + "step": 19560 + }, + { + "epoch": 2.488360259508968, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.715593338012695, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9070495367050171, + "num_tokens": 759632681.0, + "step": 19561 + }, + { + "epoch": 2.4884874697875587, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.857321739196777, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.896341860294342, + "num_tokens": 759671230.0, + "step": 19562 + }, + { + "epoch": 2.4886146800661493, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.821447372436523, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8873206377029419, + "num_tokens": 759711458.0, + "step": 19563 + }, + { + "epoch": 2.48874189034474, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.890503883361816, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9055071473121643, + "num_tokens": 759743813.0, + "step": 19564 + }, + { + "epoch": 2.4888691006233303, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.796866416931152, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8953016400337219, + "num_tokens": 759780225.0, + "step": 19565 + }, + { + "epoch": 2.488996310901921, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.827413558959961, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8943226933479309, + "num_tokens": 759816676.0, + "step": 19566 + }, + { + "epoch": 2.4891235211805114, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.80321979522705, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8890442848205566, + "num_tokens": 759861362.0, + "step": 19567 + }, + { + "epoch": 2.489250731459102, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.771330833435059, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.9097041487693787, + "num_tokens": 759905193.0, + "step": 19568 + }, + { + "epoch": 2.4893779417376924, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.836275100708008, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8985708355903625, + "num_tokens": 759942175.0, + "step": 19569 + }, + { + "epoch": 2.489505152016283, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.82026481628418, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.9049199223518372, + "num_tokens": 759976548.0, + "step": 19570 + }, + { + "epoch": 2.4896323622948735, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.806349754333496, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.9059292078018188, + "num_tokens": 760013852.0, + "step": 19571 + }, + { + "epoch": 2.489759572573464, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.815202713012695, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8920676112174988, + "num_tokens": 760056205.0, + "step": 19572 + }, + { + "epoch": 2.4898867828520546, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.817010879516602, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8948243856430054, + "num_tokens": 760097428.0, + "step": 19573 + }, + { + "epoch": 2.490013993130645, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.727415084838867, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9057032465934753, + "num_tokens": 760137064.0, + "step": 19574 + }, + { + "epoch": 2.4901412034092356, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.867836952209473, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8953039646148682, + "num_tokens": 760173668.0, + "step": 19575 + }, + { + "epoch": 2.490268413687826, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.814790725708008, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8959551453590393, + "num_tokens": 760209537.0, + "step": 19576 + }, + { + "epoch": 2.4903956239664167, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.843962669372559, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8829488754272461, + "num_tokens": 760247339.0, + "step": 19577 + }, + { + "epoch": 2.4905228342450068, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.794840812683105, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.9038032293319702, + "num_tokens": 760281575.0, + "step": 19578 + }, + { + "epoch": 2.4906500445235977, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.834766387939453, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8940268158912659, + "num_tokens": 760315582.0, + "step": 19579 + }, + { + "epoch": 2.490777254802188, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.820639610290527, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8786872029304504, + "num_tokens": 760357592.0, + "step": 19580 + }, + { + "epoch": 2.4909044650807783, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.782767295837402, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.9004195928573608, + "num_tokens": 760399535.0, + "step": 19581 + }, + { + "epoch": 2.491031675359369, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.76618480682373, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8972612619400024, + "num_tokens": 760443441.0, + "step": 19582 + }, + { + "epoch": 2.4911588856379594, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.785723686218262, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9115263819694519, + "num_tokens": 760478953.0, + "step": 19583 + }, + { + "epoch": 2.49128609591655, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.750364303588867, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9049680233001709, + "num_tokens": 760519627.0, + "step": 19584 + }, + { + "epoch": 2.4914133061951405, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.729870796203613, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8882863521575928, + "num_tokens": 760565091.0, + "step": 19585 + }, + { + "epoch": 2.491540516473731, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.833271980285645, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.9007290601730347, + "num_tokens": 760603079.0, + "step": 19586 + }, + { + "epoch": 2.4916677267523215, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.740362167358398, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.9010205864906311, + "num_tokens": 760647046.0, + "step": 19587 + }, + { + "epoch": 2.491794937030912, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.820371627807617, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8990138173103333, + "num_tokens": 760689164.0, + "step": 19588 + }, + { + "epoch": 2.4919221473095026, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.921208381652832, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8835725784301758, + "num_tokens": 760728810.0, + "step": 19589 + }, + { + "epoch": 2.492049357588093, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.86914348602295, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8940888047218323, + "num_tokens": 760765171.0, + "step": 19590 + }, + { + "epoch": 2.4921765678666836, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.768421173095703, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.895520806312561, + "num_tokens": 760809265.0, + "step": 19591 + }, + { + "epoch": 2.492303778145274, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.810352325439453, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.887598991394043, + "num_tokens": 760851979.0, + "step": 19592 + }, + { + "epoch": 2.4924309884238647, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.8970947265625, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8992483615875244, + "num_tokens": 760884417.0, + "step": 19593 + }, + { + "epoch": 2.492558198702455, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.850163459777832, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.879624605178833, + "num_tokens": 760924172.0, + "step": 19594 + }, + { + "epoch": 2.4926854089810457, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.810538291931152, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8998256921768188, + "num_tokens": 760959299.0, + "step": 19595 + }, + { + "epoch": 2.4928126192596363, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.745786666870117, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8933185935020447, + "num_tokens": 760999081.0, + "step": 19596 + }, + { + "epoch": 2.492939829538227, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.798259735107422, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8946391344070435, + "num_tokens": 761039937.0, + "step": 19597 + }, + { + "epoch": 2.4930670398168173, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.802470207214355, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.9019900560379028, + "num_tokens": 761083037.0, + "step": 19598 + }, + { + "epoch": 2.493194250095408, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.865140914916992, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8923080563545227, + "num_tokens": 761116048.0, + "step": 19599 + }, + { + "epoch": 2.4933214603739984, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.85062026977539, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8927781581878662, + "num_tokens": 761152737.0, + "step": 19600 + }, + { + "epoch": 2.493448670652589, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.757241249084473, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.8952071070671082, + "num_tokens": 761188908.0, + "step": 19601 + }, + { + "epoch": 2.4935758809311794, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.048187255859375, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8954324126243591, + "num_tokens": 761220201.0, + "step": 19602 + }, + { + "epoch": 2.4937030912097695, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.852689743041992, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9067923426628113, + "num_tokens": 761255580.0, + "step": 19603 + }, + { + "epoch": 2.4938303014883605, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.712018013000488, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8983428478240967, + "num_tokens": 761296495.0, + "step": 19604 + }, + { + "epoch": 2.4939575117669506, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.9956636428833, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8927192687988281, + "num_tokens": 761332210.0, + "step": 19605 + }, + { + "epoch": 2.494084722045541, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.836009979248047, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8921990394592285, + "num_tokens": 761368050.0, + "step": 19606 + }, + { + "epoch": 2.4942119323241316, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.757840156555176, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8913842439651489, + "num_tokens": 761410108.0, + "step": 19607 + }, + { + "epoch": 2.494339142602722, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.806275367736816, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9055453538894653, + "num_tokens": 761448040.0, + "step": 19608 + }, + { + "epoch": 2.4944663528813127, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.786466598510742, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9023879766464233, + "num_tokens": 761487284.0, + "step": 19609 + }, + { + "epoch": 2.494593563159903, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.944618225097656, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8990447521209717, + "num_tokens": 761524229.0, + "step": 19610 + }, + { + "epoch": 2.4947207734384937, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.773627281188965, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.900443971157074, + "num_tokens": 761563588.0, + "step": 19611 + }, + { + "epoch": 2.4948479837170843, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.72878360748291, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.915951132774353, + "num_tokens": 761595687.0, + "step": 19612 + }, + { + "epoch": 2.494975193995675, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.849987030029297, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.9011924862861633, + "num_tokens": 761640489.0, + "step": 19613 + }, + { + "epoch": 2.4951024042742653, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.804774284362793, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8799530863761902, + "num_tokens": 761681796.0, + "step": 19614 + }, + { + "epoch": 2.495229614552856, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.929759979248047, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8947620987892151, + "num_tokens": 761714926.0, + "step": 19615 + }, + { + "epoch": 2.4953568248314464, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.848637580871582, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8911346197128296, + "num_tokens": 761754601.0, + "step": 19616 + }, + { + "epoch": 2.495484035110037, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.875903129577637, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.888845682144165, + "num_tokens": 761795125.0, + "step": 19617 + }, + { + "epoch": 2.4956112453886274, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.82606029510498, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8959459066390991, + "num_tokens": 761835393.0, + "step": 19618 + }, + { + "epoch": 2.495738455667218, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.759242057800293, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8898581266403198, + "num_tokens": 761878536.0, + "step": 19619 + }, + { + "epoch": 2.4958656659458085, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.80864429473877, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8975919485092163, + "num_tokens": 761920439.0, + "step": 19620 + }, + { + "epoch": 2.495992876224399, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.887897491455078, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8943186402320862, + "num_tokens": 761956928.0, + "step": 19621 + }, + { + "epoch": 2.4961200865029896, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.915279388427734, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8900375962257385, + "num_tokens": 761991044.0, + "step": 19622 + }, + { + "epoch": 2.49624729678158, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.812161445617676, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8945226669311523, + "num_tokens": 762031980.0, + "step": 19623 + }, + { + "epoch": 2.4963745070601706, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.79290771484375, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8973568677902222, + "num_tokens": 762071050.0, + "step": 19624 + }, + { + "epoch": 2.496501717338761, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.868184089660645, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8949165344238281, + "num_tokens": 762110122.0, + "step": 19625 + }, + { + "epoch": 2.4966289276173517, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.91173267364502, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8817164897918701, + "num_tokens": 762149504.0, + "step": 19626 + }, + { + "epoch": 2.496756137895942, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.75328254699707, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.9016482830047607, + "num_tokens": 762188722.0, + "step": 19627 + }, + { + "epoch": 2.4968833481745323, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.91278076171875, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.897072970867157, + "num_tokens": 762222356.0, + "step": 19628 + }, + { + "epoch": 2.4970105584531233, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.934093475341797, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.900223433971405, + "num_tokens": 762253534.0, + "step": 19629 + }, + { + "epoch": 2.4971377687317133, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.872739791870117, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8811615109443665, + "num_tokens": 762291712.0, + "step": 19630 + }, + { + "epoch": 2.497264979010304, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.7801513671875, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8937170505523682, + "num_tokens": 762332749.0, + "step": 19631 + }, + { + "epoch": 2.4973921892888944, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.838420867919922, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8967908620834351, + "num_tokens": 762367650.0, + "step": 19632 + }, + { + "epoch": 2.497519399567485, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.84932804107666, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.890981912612915, + "num_tokens": 762410651.0, + "step": 19633 + }, + { + "epoch": 2.4976466098460754, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.782418251037598, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.888002336025238, + "num_tokens": 762454397.0, + "step": 19634 + }, + { + "epoch": 2.497773820124666, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.9088134765625, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8952898383140564, + "num_tokens": 762491542.0, + "step": 19635 + }, + { + "epoch": 2.4979010304032565, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.849180221557617, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8843488097190857, + "num_tokens": 762526998.0, + "step": 19636 + }, + { + "epoch": 2.498028240681847, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.854598045349121, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9043688774108887, + "num_tokens": 762563866.0, + "step": 19637 + }, + { + "epoch": 2.4981554509604376, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.862013816833496, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.9064955115318298, + "num_tokens": 762597592.0, + "step": 19638 + }, + { + "epoch": 2.498282661239028, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.834717750549316, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8981068730354309, + "num_tokens": 762634959.0, + "step": 19639 + }, + { + "epoch": 2.4984098715176186, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.90221118927002, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.889170229434967, + "num_tokens": 762671775.0, + "step": 19640 + }, + { + "epoch": 2.498537081796209, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.682735443115234, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.9001466631889343, + "num_tokens": 762715794.0, + "step": 19641 + }, + { + "epoch": 2.4986642920747997, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.799540519714355, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9068058729171753, + "num_tokens": 762750026.0, + "step": 19642 + }, + { + "epoch": 2.49879150235339, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.775235176086426, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8983802199363708, + "num_tokens": 762790283.0, + "step": 19643 + }, + { + "epoch": 2.4989187126319807, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.833460807800293, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8988943099975586, + "num_tokens": 762824594.0, + "step": 19644 + }, + { + "epoch": 2.4990459229105713, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.846738815307617, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.9049948453903198, + "num_tokens": 762859472.0, + "step": 19645 + }, + { + "epoch": 2.499173133189162, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.778236389160156, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8992946743965149, + "num_tokens": 762898039.0, + "step": 19646 + }, + { + "epoch": 2.4993003434677523, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.846924781799316, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8997769355773926, + "num_tokens": 762937390.0, + "step": 19647 + }, + { + "epoch": 2.499427553746343, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.746091842651367, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8989351987838745, + "num_tokens": 762974305.0, + "step": 19648 + }, + { + "epoch": 2.4995547640249334, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.823711395263672, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.898753821849823, + "num_tokens": 763017012.0, + "step": 19649 + }, + { + "epoch": 2.499681974303524, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.943201065063477, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8977192640304565, + "num_tokens": 763048709.0, + "step": 19650 + }, + { + "epoch": 2.499809184582114, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.826960563659668, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.9044545888900757, + "num_tokens": 763083910.0, + "step": 19651 + }, + { + "epoch": 2.499936394860705, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.827716827392578, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8944713473320007, + "num_tokens": 763126523.0, + "step": 19652 + }, + { + "epoch": 2.500063605139295, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.773321151733398, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9086563587188721, + "num_tokens": 763172252.0, + "step": 19653 + }, + { + "epoch": 2.500190815417886, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.788658142089844, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8953944444656372, + "num_tokens": 763212066.0, + "step": 19654 + }, + { + "epoch": 2.500318025696476, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.857390403747559, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9028500318527222, + "num_tokens": 763256686.0, + "step": 19655 + }, + { + "epoch": 2.5004452359750666, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.880337715148926, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8900590538978577, + "num_tokens": 763298301.0, + "step": 19656 + }, + { + "epoch": 2.500572446253657, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.7227201461792, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.9046825170516968, + "num_tokens": 763339696.0, + "step": 19657 + }, + { + "epoch": 2.5006996565322477, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.969761848449707, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8862913846969604, + "num_tokens": 763379625.0, + "step": 19658 + }, + { + "epoch": 2.500826866810838, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.80362319946289, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.9088293313980103, + "num_tokens": 763415012.0, + "step": 19659 + }, + { + "epoch": 2.5009540770894287, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.877531051635742, + "learning_rate": 1e-06, + "loss": 0.3908, + "mean_token_accuracy": 0.872353196144104, + "num_tokens": 763460521.0, + "step": 19660 + }, + { + "epoch": 2.5010812873680193, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.836504936218262, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9032044410705566, + "num_tokens": 763504273.0, + "step": 19661 + }, + { + "epoch": 2.50120849764661, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.827094078063965, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.9002156257629395, + "num_tokens": 763538699.0, + "step": 19662 + }, + { + "epoch": 2.5013357079252003, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.786347389221191, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.9029234647750854, + "num_tokens": 763579328.0, + "step": 19663 + }, + { + "epoch": 2.501462918203791, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.845043182373047, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8910418748855591, + "num_tokens": 763621595.0, + "step": 19664 + }, + { + "epoch": 2.5015901284823814, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.877315521240234, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8907362222671509, + "num_tokens": 763660454.0, + "step": 19665 + }, + { + "epoch": 2.501717338760972, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.789711952209473, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9012051820755005, + "num_tokens": 763695558.0, + "step": 19666 + }, + { + "epoch": 2.5018445490395624, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.849678039550781, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.883526623249054, + "num_tokens": 763736524.0, + "step": 19667 + }, + { + "epoch": 2.501971759318153, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.791071891784668, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8991209268569946, + "num_tokens": 763775647.0, + "step": 19668 + }, + { + "epoch": 2.5020989695967435, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.680230140686035, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8991925716400146, + "num_tokens": 763820169.0, + "step": 19669 + }, + { + "epoch": 2.502226179875334, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.750362396240234, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.910110354423523, + "num_tokens": 763862993.0, + "step": 19670 + }, + { + "epoch": 2.5023533901539246, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.795900344848633, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9075409173965454, + "num_tokens": 763899549.0, + "step": 19671 + }, + { + "epoch": 2.502480600432515, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.876208305358887, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8945343494415283, + "num_tokens": 763937043.0, + "step": 19672 + }, + { + "epoch": 2.5026078107111056, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.815582275390625, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9062405824661255, + "num_tokens": 763970276.0, + "step": 19673 + }, + { + "epoch": 2.5027350209896957, + "ewc_loss": 0.0205078125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.901813507080078, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8906509876251221, + "num_tokens": 764005700.0, + "step": 19674 + }, + { + "epoch": 2.5028622312682867, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.900575637817383, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8950458765029907, + "num_tokens": 764040579.0, + "step": 19675 + }, + { + "epoch": 2.5029894415468767, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.747345924377441, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.902618408203125, + "num_tokens": 764077800.0, + "step": 19676 + }, + { + "epoch": 2.5031166518254677, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.869171142578125e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.833940505981445, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8989271521568298, + "num_tokens": 764117720.0, + "step": 19677 + }, + { + "epoch": 2.503243862104058, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.83535099029541, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8977841138839722, + "num_tokens": 764158580.0, + "step": 19678 + }, + { + "epoch": 2.5033710723826488, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.778508186340332, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8956699967384338, + "num_tokens": 764205553.0, + "step": 19679 + }, + { + "epoch": 2.503498282661239, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.791253089904785, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9041010141372681, + "num_tokens": 764244413.0, + "step": 19680 + }, + { + "epoch": 2.5036254929398294, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.92509651184082, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8956432938575745, + "num_tokens": 764284070.0, + "step": 19681 + }, + { + "epoch": 2.50375270321842, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.84600830078125, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8998428583145142, + "num_tokens": 764322763.0, + "step": 19682 + }, + { + "epoch": 2.5038799134970104, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.868243217468262, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8957822322845459, + "num_tokens": 764358926.0, + "step": 19683 + }, + { + "epoch": 2.504007123775601, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.786918640136719, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9113807082176208, + "num_tokens": 764399033.0, + "step": 19684 + }, + { + "epoch": 2.5041343340541915, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.886919975280762, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.893073558807373, + "num_tokens": 764440718.0, + "step": 19685 + }, + { + "epoch": 2.504261544332782, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.806068420410156, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.9006205201148987, + "num_tokens": 764482428.0, + "step": 19686 + }, + { + "epoch": 2.5043887546113726, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.97939395904541, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8938523530960083, + "num_tokens": 764523048.0, + "step": 19687 + }, + { + "epoch": 2.504515964889963, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.803205490112305, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8939205408096313, + "num_tokens": 764565872.0, + "step": 19688 + }, + { + "epoch": 2.5046431751685536, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.856154441833496, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.9010710120201111, + "num_tokens": 764603884.0, + "step": 19689 + }, + { + "epoch": 2.504770385447144, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.969993591308594, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8890873193740845, + "num_tokens": 764641117.0, + "step": 19690 + }, + { + "epoch": 2.5048975957257347, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.8216552734375, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8993080854415894, + "num_tokens": 764676543.0, + "step": 19691 + }, + { + "epoch": 2.505024806004325, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.826815605163574, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9040082097053528, + "num_tokens": 764716364.0, + "step": 19692 + }, + { + "epoch": 2.5051520162829157, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.8128662109375, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.9004330635070801, + "num_tokens": 764757623.0, + "step": 19693 + }, + { + "epoch": 2.5052792265615063, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.870031356811523, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8955600261688232, + "num_tokens": 764792872.0, + "step": 19694 + }, + { + "epoch": 2.505406436840097, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.835078239440918, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8982492089271545, + "num_tokens": 764827493.0, + "step": 19695 + }, + { + "epoch": 2.5055336471186873, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.952142715454102, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8823288679122925, + "num_tokens": 764863037.0, + "step": 19696 + }, + { + "epoch": 2.505660857397278, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.790315628051758, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.9041743278503418, + "num_tokens": 764901162.0, + "step": 19697 + }, + { + "epoch": 2.5057880676758684, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.960598945617676, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8944332003593445, + "num_tokens": 764932634.0, + "step": 19698 + }, + { + "epoch": 2.5059152779544585, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.818710327148438, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.9018529057502747, + "num_tokens": 764969938.0, + "step": 19699 + }, + { + "epoch": 2.5060424882330494, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.864474296569824, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8828282356262207, + "num_tokens": 765011169.0, + "step": 19700 + }, + { + "epoch": 2.5061696985116395, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.80677604675293, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9072788953781128, + "num_tokens": 765047108.0, + "step": 19701 + }, + { + "epoch": 2.5062969087902305, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.831626892089844, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9048941135406494, + "num_tokens": 765085553.0, + "step": 19702 + }, + { + "epoch": 2.5064241190688206, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.955153465270996, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8998859524726868, + "num_tokens": 765118448.0, + "step": 19703 + }, + { + "epoch": 2.5065513293474115, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.822604179382324, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.889184296131134, + "num_tokens": 765160335.0, + "step": 19704 + }, + { + "epoch": 2.5066785396260016, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.811737060546875, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8929445743560791, + "num_tokens": 765199372.0, + "step": 19705 + }, + { + "epoch": 2.506805749904592, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.818490028381348, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8985626697540283, + "num_tokens": 765243886.0, + "step": 19706 + }, + { + "epoch": 2.5069329601831827, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.783092498779297, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8936840295791626, + "num_tokens": 765285110.0, + "step": 19707 + }, + { + "epoch": 2.507060170461773, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.833168983459473, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.9040737152099609, + "num_tokens": 765327345.0, + "step": 19708 + }, + { + "epoch": 2.5071873807403637, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.798524856567383, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8827516436576843, + "num_tokens": 765372400.0, + "step": 19709 + }, + { + "epoch": 2.5073145910189543, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.80762004852295, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.9005246758460999, + "num_tokens": 765414518.0, + "step": 19710 + }, + { + "epoch": 2.507441801297545, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.901039123535156, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8979970216751099, + "num_tokens": 765453577.0, + "step": 19711 + }, + { + "epoch": 2.5075690115761353, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.845513343811035, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8989344239234924, + "num_tokens": 765494998.0, + "step": 19712 + }, + { + "epoch": 2.507696221854726, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.882323265075684, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8801256418228149, + "num_tokens": 765531120.0, + "step": 19713 + }, + { + "epoch": 2.5078234321333164, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.945625305175781, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8886959552764893, + "num_tokens": 765567456.0, + "step": 19714 + }, + { + "epoch": 2.507950642411907, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.812944412231445, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.9010352492332458, + "num_tokens": 765605184.0, + "step": 19715 + }, + { + "epoch": 2.5080778526904974, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.898029327392578, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8914662599563599, + "num_tokens": 765645895.0, + "step": 19716 + }, + { + "epoch": 2.508205062969088, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.760209083557129, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8955080509185791, + "num_tokens": 765687588.0, + "step": 19717 + }, + { + "epoch": 2.5083322732476785, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.8098726272583, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8927438259124756, + "num_tokens": 765727756.0, + "step": 19718 + }, + { + "epoch": 2.508459483526269, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.877604484558105, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8883716464042664, + "num_tokens": 765764310.0, + "step": 19719 + }, + { + "epoch": 2.5085866938048595, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.770362854003906, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8963563442230225, + "num_tokens": 765803039.0, + "step": 19720 + }, + { + "epoch": 2.50871390408345, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.97903823852539, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8856042623519897, + "num_tokens": 765838649.0, + "step": 19721 + }, + { + "epoch": 2.5088411143620406, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.763705253601074, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.905820369720459, + "num_tokens": 765873328.0, + "step": 19722 + }, + { + "epoch": 2.508968324640631, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.87138557434082, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8981065154075623, + "num_tokens": 765905761.0, + "step": 19723 + }, + { + "epoch": 2.509095534919221, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.80051326751709, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.9034550189971924, + "num_tokens": 765944489.0, + "step": 19724 + }, + { + "epoch": 2.509222745197812, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.827590942382812, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8926494121551514, + "num_tokens": 765984573.0, + "step": 19725 + }, + { + "epoch": 2.5093499554764023, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.825332641601562, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9091726541519165, + "num_tokens": 766018986.0, + "step": 19726 + }, + { + "epoch": 2.5094771657549932, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.783157348632812, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.9032849073410034, + "num_tokens": 766059024.0, + "step": 19727 + }, + { + "epoch": 2.5096043760335833, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.75061321258545, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.9080471992492676, + "num_tokens": 766096457.0, + "step": 19728 + }, + { + "epoch": 2.5097315863121743, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.874789237976074, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.887496829032898, + "num_tokens": 766137538.0, + "step": 19729 + }, + { + "epoch": 2.5098587965907644, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.840011596679688, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.907724142074585, + "num_tokens": 766174308.0, + "step": 19730 + }, + { + "epoch": 2.509986006869355, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 10.015461921691895, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8980633020401001, + "num_tokens": 766203027.0, + "step": 19731 + }, + { + "epoch": 2.5101132171479454, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.818625450134277, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.887646496295929, + "num_tokens": 766243945.0, + "step": 19732 + }, + { + "epoch": 2.510240427426536, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.776562690734863, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8970992565155029, + "num_tokens": 766288379.0, + "step": 19733 + }, + { + "epoch": 2.5103676377051265, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.996077537536621, + "learning_rate": 1e-06, + "loss": 0.3597, + "mean_token_accuracy": 0.8784752488136292, + "num_tokens": 766324291.0, + "step": 19734 + }, + { + "epoch": 2.510494847983717, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.846633911132812, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.900616466999054, + "num_tokens": 766365199.0, + "step": 19735 + }, + { + "epoch": 2.5106220582623076, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.928350448608398, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.9025253057479858, + "num_tokens": 766400472.0, + "step": 19736 + }, + { + "epoch": 2.510749268540898, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.929591178894043, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.9008399248123169, + "num_tokens": 766436696.0, + "step": 19737 + }, + { + "epoch": 2.5108764788194886, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.836511611938477, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9077266454696655, + "num_tokens": 766474485.0, + "step": 19738 + }, + { + "epoch": 2.511003689098079, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.868610382080078, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8999009132385254, + "num_tokens": 766513747.0, + "step": 19739 + }, + { + "epoch": 2.5111308993766697, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.846464157104492, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9085897207260132, + "num_tokens": 766551901.0, + "step": 19740 + }, + { + "epoch": 2.51125810965526, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.864871978759766, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8926924467086792, + "num_tokens": 766592708.0, + "step": 19741 + }, + { + "epoch": 2.5113853199338507, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.973047256469727, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8869546055793762, + "num_tokens": 766627598.0, + "step": 19742 + }, + { + "epoch": 2.5115125302124413, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.798076629638672, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9148260354995728, + "num_tokens": 766664044.0, + "step": 19743 + }, + { + "epoch": 2.511639740491032, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.74396800994873, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9059701561927795, + "num_tokens": 766705894.0, + "step": 19744 + }, + { + "epoch": 2.5117669507696223, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.836337089538574, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.88987135887146, + "num_tokens": 766750240.0, + "step": 19745 + }, + { + "epoch": 2.511894161048213, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.833806991577148, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.894429087638855, + "num_tokens": 766793280.0, + "step": 19746 + }, + { + "epoch": 2.5120213713268034, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.787927627563477, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9043101668357849, + "num_tokens": 766835299.0, + "step": 19747 + }, + { + "epoch": 2.512148581605394, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.904335021972656, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8989233374595642, + "num_tokens": 766871111.0, + "step": 19748 + }, + { + "epoch": 2.512275791883984, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.880680084228516, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8858436346054077, + "num_tokens": 766913732.0, + "step": 19749 + }, + { + "epoch": 2.512403002162575, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.811334609985352, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8999482989311218, + "num_tokens": 766952399.0, + "step": 19750 + }, + { + "epoch": 2.512530212441165, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.911580085754395, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8843474388122559, + "num_tokens": 766992935.0, + "step": 19751 + }, + { + "epoch": 2.512657422719756, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.789349555969238, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8829600811004639, + "num_tokens": 767037658.0, + "step": 19752 + }, + { + "epoch": 2.512784632998346, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.832805633544922, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8862699866294861, + "num_tokens": 767080460.0, + "step": 19753 + }, + { + "epoch": 2.5129118432769366, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.965607643127441, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.895494282245636, + "num_tokens": 767113725.0, + "step": 19754 + }, + { + "epoch": 2.513039053555527, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.816110610961914, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9115884304046631, + "num_tokens": 767147585.0, + "step": 19755 + }, + { + "epoch": 2.5131662638341177, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.820904731750488, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8960130214691162, + "num_tokens": 767193189.0, + "step": 19756 + }, + { + "epoch": 2.513293474112708, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.850159645080566, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8934635519981384, + "num_tokens": 767235376.0, + "step": 19757 + }, + { + "epoch": 2.5134206843912987, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.88971996307373, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8897935748100281, + "num_tokens": 767270916.0, + "step": 19758 + }, + { + "epoch": 2.5135478946698893, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.848745346069336, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.9020768404006958, + "num_tokens": 767309932.0, + "step": 19759 + }, + { + "epoch": 2.51367510494848, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.823016166687012, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8990294933319092, + "num_tokens": 767347248.0, + "step": 19760 + }, + { + "epoch": 2.5138023152270703, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.913766860961914, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8838084936141968, + "num_tokens": 767387892.0, + "step": 19761 + }, + { + "epoch": 2.513929525505661, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.803179740905762, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9010090827941895, + "num_tokens": 767427203.0, + "step": 19762 + }, + { + "epoch": 2.5140567357842514, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.755472183227539, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9066358804702759, + "num_tokens": 767467906.0, + "step": 19763 + }, + { + "epoch": 2.514183946062842, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.892461776733398, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.895731508731842, + "num_tokens": 767505669.0, + "step": 19764 + }, + { + "epoch": 2.5143111563414324, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.848440170288086, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8969920873641968, + "num_tokens": 767546130.0, + "step": 19765 + }, + { + "epoch": 2.514438366620023, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.760664939880371, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.9001094102859497, + "num_tokens": 767587627.0, + "step": 19766 + }, + { + "epoch": 2.5145655768986135, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.789834022521973, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8955360651016235, + "num_tokens": 767629762.0, + "step": 19767 + }, + { + "epoch": 2.514692787177204, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.906774520874023, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.897190511226654, + "num_tokens": 767668857.0, + "step": 19768 + }, + { + "epoch": 2.5148199974557945, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.816900253295898, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9114236831665039, + "num_tokens": 767704034.0, + "step": 19769 + }, + { + "epoch": 2.514947207734385, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.872004508972168, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8891722559928894, + "num_tokens": 767747148.0, + "step": 19770 + }, + { + "epoch": 2.5150744180129756, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.82386302947998, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8858020305633545, + "num_tokens": 767790647.0, + "step": 19771 + }, + { + "epoch": 2.5152016282915657, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.96558666229248, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8915192484855652, + "num_tokens": 767828463.0, + "step": 19772 + }, + { + "epoch": 2.5153288385701567, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.836580276489258, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.901748776435852, + "num_tokens": 767867973.0, + "step": 19773 + }, + { + "epoch": 2.5154560488487467, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.816275596618652, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9073176383972168, + "num_tokens": 767906675.0, + "step": 19774 + }, + { + "epoch": 2.5155832591273377, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.945441246032715, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.895622968673706, + "num_tokens": 767945950.0, + "step": 19775 + }, + { + "epoch": 2.515710469405928, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.78433895111084, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.89774489402771, + "num_tokens": 767988855.0, + "step": 19776 + }, + { + "epoch": 2.5158376796845188, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.958075523376465, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8898413181304932, + "num_tokens": 768026422.0, + "step": 19777 + }, + { + "epoch": 2.515964889963109, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.839362144470215, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.906826376914978, + "num_tokens": 768064226.0, + "step": 19778 + }, + { + "epoch": 2.5160921002416994, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.765582084655762, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9087852239608765, + "num_tokens": 768106693.0, + "step": 19779 + }, + { + "epoch": 2.51621931052029, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.861001968383789, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8987916111946106, + "num_tokens": 768145630.0, + "step": 19780 + }, + { + "epoch": 2.5163465207988804, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.797595977783203, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.898526132106781, + "num_tokens": 768189224.0, + "step": 19781 + }, + { + "epoch": 2.516473731077471, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.81601619720459, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8965293169021606, + "num_tokens": 768230960.0, + "step": 19782 + }, + { + "epoch": 2.5166009413560615, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.951395034790039, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9036379456520081, + "num_tokens": 768260838.0, + "step": 19783 + }, + { + "epoch": 2.516728151634652, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.833833694458008, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.9022057056427002, + "num_tokens": 768300719.0, + "step": 19784 + }, + { + "epoch": 2.5168553619132426, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.75691032409668, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.9010175466537476, + "num_tokens": 768344463.0, + "step": 19785 + }, + { + "epoch": 2.516982572191833, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 10.010013580322266, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8919072151184082, + "num_tokens": 768374185.0, + "step": 19786 + }, + { + "epoch": 2.5171097824704236, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.824553489685059, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8822895288467407, + "num_tokens": 768422598.0, + "step": 19787 + }, + { + "epoch": 2.517236992749014, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.78695011138916, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8973591327667236, + "num_tokens": 768468331.0, + "step": 19788 + }, + { + "epoch": 2.5173642030276047, + "ewc_loss": 0.0206298828125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1742115020751953e-05, + "grad_norm": 9.813227653503418, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8988634943962097, + "num_tokens": 768507017.0, + "step": 19789 + }, + { + "epoch": 2.517491413306195, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.916854858398438e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.81325626373291, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9104833006858826, + "num_tokens": 768541870.0, + "step": 19790 + }, + { + "epoch": 2.5176186235847857, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.858163833618164, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.9003846645355225, + "num_tokens": 768580319.0, + "step": 19791 + }, + { + "epoch": 2.5177458338633762, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.854750633239746, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9040979743003845, + "num_tokens": 768615744.0, + "step": 19792 + }, + { + "epoch": 2.5178730441419668, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.797721862792969, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.9034309983253479, + "num_tokens": 768656857.0, + "step": 19793 + }, + { + "epoch": 2.5180002544205573, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.861698150634766, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9132465720176697, + "num_tokens": 768690167.0, + "step": 19794 + }, + { + "epoch": 2.518127464699148, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.806201934814453, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8997340202331543, + "num_tokens": 768732571.0, + "step": 19795 + }, + { + "epoch": 2.5182546749777384, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.781174659729004, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8993068337440491, + "num_tokens": 768775499.0, + "step": 19796 + }, + { + "epoch": 2.5183818852563284, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.82954216003418, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.9004702568054199, + "num_tokens": 768817294.0, + "step": 19797 + }, + { + "epoch": 2.5185090955349194, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.828121185302734, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8769518136978149, + "num_tokens": 768858992.0, + "step": 19798 + }, + { + "epoch": 2.5186363058135095, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.038888931274414, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8845466375350952, + "num_tokens": 768890660.0, + "step": 19799 + }, + { + "epoch": 2.5187635160921005, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.766331672668457, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9076383113861084, + "num_tokens": 768936117.0, + "step": 19800 + }, + { + "epoch": 2.5188907263706906, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.815042495727539, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8964425921440125, + "num_tokens": 768973708.0, + "step": 19801 + }, + { + "epoch": 2.5190179366492815, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.91419792175293, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8891875147819519, + "num_tokens": 769012582.0, + "step": 19802 + }, + { + "epoch": 2.5191451469278716, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.786179542541504, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.910129964351654, + "num_tokens": 769048817.0, + "step": 19803 + }, + { + "epoch": 2.519272357206462, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.819868087768555, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.9052009582519531, + "num_tokens": 769085979.0, + "step": 19804 + }, + { + "epoch": 2.5193995674850527, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.87062931060791, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.905173659324646, + "num_tokens": 769118315.0, + "step": 19805 + }, + { + "epoch": 2.519526777763643, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.945695877075195, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8880205154418945, + "num_tokens": 769153777.0, + "step": 19806 + }, + { + "epoch": 2.5196539880422337, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.769291877746582, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8991533517837524, + "num_tokens": 769191554.0, + "step": 19807 + }, + { + "epoch": 2.5197811983208243, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.859123229980469, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.9054486751556396, + "num_tokens": 769231538.0, + "step": 19808 + }, + { + "epoch": 2.519908408599415, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.773195266723633, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.904723048210144, + "num_tokens": 769272492.0, + "step": 19809 + }, + { + "epoch": 2.5200356188780053, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.820737838745117, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8896958827972412, + "num_tokens": 769317805.0, + "step": 19810 + }, + { + "epoch": 2.520162829156596, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.826373100280762, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.908196210861206, + "num_tokens": 769354193.0, + "step": 19811 + }, + { + "epoch": 2.5202900394351864, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.876179695129395, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8983801603317261, + "num_tokens": 769390384.0, + "step": 19812 + }, + { + "epoch": 2.520417249713777, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.844904899597168, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8880465030670166, + "num_tokens": 769433727.0, + "step": 19813 + }, + { + "epoch": 2.5205444599923674, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 10.050480842590332, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8864607810974121, + "num_tokens": 769469288.0, + "step": 19814 + }, + { + "epoch": 2.520671670270958, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.769972801208496, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8909287452697754, + "num_tokens": 769513483.0, + "step": 19815 + }, + { + "epoch": 2.5207988805495485, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.777494430541992, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.9049148559570312, + "num_tokens": 769555962.0, + "step": 19816 + }, + { + "epoch": 2.520926090828139, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.772127151489258, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.9021194577217102, + "num_tokens": 769597342.0, + "step": 19817 + }, + { + "epoch": 2.5210533011067295, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.910676956176758, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8833578824996948, + "num_tokens": 769638106.0, + "step": 19818 + }, + { + "epoch": 2.52118051138532, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.810542106628418, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8913796544075012, + "num_tokens": 769681146.0, + "step": 19819 + }, + { + "epoch": 2.5213077216639106, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 10.048221588134766, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8939225077629089, + "num_tokens": 769713474.0, + "step": 19820 + }, + { + "epoch": 2.521434931942501, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.85391616821289, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.9043894410133362, + "num_tokens": 769751905.0, + "step": 19821 + }, + { + "epoch": 2.521562142221091, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.83831787109375, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8953028917312622, + "num_tokens": 769790593.0, + "step": 19822 + }, + { + "epoch": 2.521689352499682, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.837319374084473, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9100694060325623, + "num_tokens": 769823794.0, + "step": 19823 + }, + { + "epoch": 2.5218165627782723, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.89079761505127, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.9015682935714722, + "num_tokens": 769862807.0, + "step": 19824 + }, + { + "epoch": 2.5219437730568632, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.77701187133789, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9078704118728638, + "num_tokens": 769903179.0, + "step": 19825 + }, + { + "epoch": 2.5220709833354533, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.859776496887207, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.9035661816596985, + "num_tokens": 769938982.0, + "step": 19826 + }, + { + "epoch": 2.522198193614044, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.848435401916504, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8823808431625366, + "num_tokens": 769985056.0, + "step": 19827 + }, + { + "epoch": 2.5223254038926344, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.84697437286377, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8900374174118042, + "num_tokens": 770027477.0, + "step": 19828 + }, + { + "epoch": 2.522452614171225, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.919846534729004, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8932849168777466, + "num_tokens": 770067745.0, + "step": 19829 + }, + { + "epoch": 2.5225798244498154, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.892635345458984, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.9015017151832581, + "num_tokens": 770101114.0, + "step": 19830 + }, + { + "epoch": 2.522707034728406, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.943477630615234, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8909969329833984, + "num_tokens": 770135207.0, + "step": 19831 + }, + { + "epoch": 2.5228342450069965, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.79723834991455, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8966276049613953, + "num_tokens": 770174004.0, + "step": 19832 + }, + { + "epoch": 2.522961455285587, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.806736946105957, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8887848854064941, + "num_tokens": 770218996.0, + "step": 19833 + }, + { + "epoch": 2.5230886655641775, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.79211139678955, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.902190625667572, + "num_tokens": 770258558.0, + "step": 19834 + }, + { + "epoch": 2.523215875842768, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.909333229064941, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8903629779815674, + "num_tokens": 770291961.0, + "step": 19835 + }, + { + "epoch": 2.5233430861213586, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.857139587402344, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8898272514343262, + "num_tokens": 770330916.0, + "step": 19836 + }, + { + "epoch": 2.523470296399949, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.796475410461426, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9029813408851624, + "num_tokens": 770371461.0, + "step": 19837 + }, + { + "epoch": 2.5235975066785397, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.979021072387695, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8802776336669922, + "num_tokens": 770405548.0, + "step": 19838 + }, + { + "epoch": 2.52372471695713, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.890007972717285, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8921380043029785, + "num_tokens": 770440476.0, + "step": 19839 + }, + { + "epoch": 2.5238519272357207, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.863079071044922, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8983727097511292, + "num_tokens": 770476297.0, + "step": 19840 + }, + { + "epoch": 2.5239791375143112, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.855995178222656, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8999372720718384, + "num_tokens": 770517467.0, + "step": 19841 + }, + { + "epoch": 2.5241063477929018, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.862387657165527, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8773971796035767, + "num_tokens": 770564094.0, + "step": 19842 + }, + { + "epoch": 2.5242335580714923, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.784821510314941, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.903924822807312, + "num_tokens": 770602555.0, + "step": 19843 + }, + { + "epoch": 2.524360768350083, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.836474418640137, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8895878195762634, + "num_tokens": 770642657.0, + "step": 19844 + }, + { + "epoch": 2.5244879786286734, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.863422393798828, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.9018889665603638, + "num_tokens": 770679447.0, + "step": 19845 + }, + { + "epoch": 2.524615188907264, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.8214693069458, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8893914222717285, + "num_tokens": 770724801.0, + "step": 19846 + }, + { + "epoch": 2.524742399185854, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.914287567138672, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8897450566291809, + "num_tokens": 770763435.0, + "step": 19847 + }, + { + "epoch": 2.524869609464445, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.825968742370605, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.9032856225967407, + "num_tokens": 770807248.0, + "step": 19848 + }, + { + "epoch": 2.524996819743035, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.995279312133789, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8961670398712158, + "num_tokens": 770839501.0, + "step": 19849 + }, + { + "epoch": 2.525124030021626, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.753520965576172, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8997273445129395, + "num_tokens": 770881252.0, + "step": 19850 + }, + { + "epoch": 2.525251240300216, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.858589172363281, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.911699652671814, + "num_tokens": 770919076.0, + "step": 19851 + }, + { + "epoch": 2.5253784505788066, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.775371551513672, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8922073841094971, + "num_tokens": 770957080.0, + "step": 19852 + }, + { + "epoch": 2.525505660857397, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.807464599609375, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9058393239974976, + "num_tokens": 770994887.0, + "step": 19853 + }, + { + "epoch": 2.5256328711359877, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.966869354248047, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9010419845581055, + "num_tokens": 771028499.0, + "step": 19854 + }, + { + "epoch": 2.525760081414578, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.917591094970703, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.881915271282196, + "num_tokens": 771072080.0, + "step": 19855 + }, + { + "epoch": 2.5258872916931687, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.871573448181152, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9081765413284302, + "num_tokens": 771112043.0, + "step": 19856 + }, + { + "epoch": 2.5260145019717593, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.833972930908203, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8944883346557617, + "num_tokens": 771147389.0, + "step": 19857 + }, + { + "epoch": 2.52614171225035, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.858875274658203, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8987398147583008, + "num_tokens": 771187626.0, + "step": 19858 + }, + { + "epoch": 2.5262689225289403, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.907852172851562, + "learning_rate": 1e-06, + "loss": 0.3475, + "mean_token_accuracy": 0.8848839998245239, + "num_tokens": 771226114.0, + "step": 19859 + }, + { + "epoch": 2.526396132807531, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.795771598815918, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.896183967590332, + "num_tokens": 771265316.0, + "step": 19860 + }, + { + "epoch": 2.5265233430861214, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.9100923538208, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.902277410030365, + "num_tokens": 771307083.0, + "step": 19861 + }, + { + "epoch": 2.526650553364712, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.896690368652344, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8935789465904236, + "num_tokens": 771346022.0, + "step": 19862 + }, + { + "epoch": 2.5267777636433024, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.857813835144043, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8890563249588013, + "num_tokens": 771386829.0, + "step": 19863 + }, + { + "epoch": 2.526904973921893, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.894980430603027, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8954858183860779, + "num_tokens": 771423487.0, + "step": 19864 + }, + { + "epoch": 2.5270321842004835, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.745826721191406, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.9002536535263062, + "num_tokens": 771464051.0, + "step": 19865 + }, + { + "epoch": 2.527159394479074, + "ewc_loss": 0.020751953125, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1801719665527344e-05, + "grad_norm": 9.945428848266602, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8871306777000427, + "num_tokens": 771498821.0, + "step": 19866 + }, + { + "epoch": 2.5272866047576645, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.79216480255127, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9061277508735657, + "num_tokens": 771540755.0, + "step": 19867 + }, + { + "epoch": 2.527413815036255, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.833928108215332, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8993820548057556, + "num_tokens": 771581995.0, + "step": 19868 + }, + { + "epoch": 2.5275410253148456, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.932921409606934, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8998194932937622, + "num_tokens": 771619241.0, + "step": 19869 + }, + { + "epoch": 2.5276682355934357, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.823063850402832, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.9047650694847107, + "num_tokens": 771660597.0, + "step": 19870 + }, + { + "epoch": 2.5277954458720266, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.882479667663574, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.9004313349723816, + "num_tokens": 771697811.0, + "step": 19871 + }, + { + "epoch": 2.5279226561506167, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.953167915344238, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8959317207336426, + "num_tokens": 771731888.0, + "step": 19872 + }, + { + "epoch": 2.5280498664292077, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.844320297241211, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8849158883094788, + "num_tokens": 771777095.0, + "step": 19873 + }, + { + "epoch": 2.528177076707798, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.85015869140625, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8955883383750916, + "num_tokens": 771814784.0, + "step": 19874 + }, + { + "epoch": 2.5283042869863888, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.86932373046875, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8997728824615479, + "num_tokens": 771856822.0, + "step": 19875 + }, + { + "epoch": 2.528431497264979, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.830595970153809, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9076548218727112, + "num_tokens": 771894955.0, + "step": 19876 + }, + { + "epoch": 2.5285587075435694, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.921621322631836, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8959746956825256, + "num_tokens": 771932040.0, + "step": 19877 + }, + { + "epoch": 2.52868591782216, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.892821311950684, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.9003449082374573, + "num_tokens": 771972574.0, + "step": 19878 + }, + { + "epoch": 2.5288131281007504, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.781900405883789, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9056249260902405, + "num_tokens": 772010206.0, + "step": 19879 + }, + { + "epoch": 2.528940338379341, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.840336799621582, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9101529121398926, + "num_tokens": 772051005.0, + "step": 19880 + }, + { + "epoch": 2.5290675486579315, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.909844398498535, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8994728326797485, + "num_tokens": 772084015.0, + "step": 19881 + }, + { + "epoch": 2.529194758936522, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.882742881774902, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.9009231925010681, + "num_tokens": 772122891.0, + "step": 19882 + }, + { + "epoch": 2.5293219692151125, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 10.075896263122559, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8816008567810059, + "num_tokens": 772157929.0, + "step": 19883 + }, + { + "epoch": 2.529449179493703, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.985711097717285, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8886924386024475, + "num_tokens": 772191319.0, + "step": 19884 + }, + { + "epoch": 2.5295763897722936, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.825024604797363, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9084047079086304, + "num_tokens": 772227529.0, + "step": 19885 + }, + { + "epoch": 2.529703600050884, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.857741355895996, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8963654041290283, + "num_tokens": 772267774.0, + "step": 19886 + }, + { + "epoch": 2.5298308103294747, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.822877883911133, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8964912295341492, + "num_tokens": 772310689.0, + "step": 19887 + }, + { + "epoch": 2.529958020608065, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.788844108581543, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.9045975208282471, + "num_tokens": 772352611.0, + "step": 19888 + }, + { + "epoch": 2.5300852308866557, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.875945091247559, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8887258768081665, + "num_tokens": 772393977.0, + "step": 19889 + }, + { + "epoch": 2.5302124411652462, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.836498260498047, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.9013767242431641, + "num_tokens": 772431699.0, + "step": 19890 + }, + { + "epoch": 2.5303396514438368, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.900713920593262, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8847134113311768, + "num_tokens": 772472034.0, + "step": 19891 + }, + { + "epoch": 2.5304668617224273, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.785852432250977, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9094384908676147, + "num_tokens": 772510702.0, + "step": 19892 + }, + { + "epoch": 2.530594072001018, + "ewc_loss": 0.0208740234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.878223419189453, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.9011530876159668, + "num_tokens": 772545770.0, + "step": 19893 + }, + { + "epoch": 2.5307212822796084, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.899559020996094, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9079567193984985, + "num_tokens": 772580936.0, + "step": 19894 + }, + { + "epoch": 2.5308484925581984, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.929914474487305, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.894780695438385, + "num_tokens": 772612066.0, + "step": 19895 + }, + { + "epoch": 2.5309757028367894, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.87295150756836, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9062900543212891, + "num_tokens": 772652889.0, + "step": 19896 + }, + { + "epoch": 2.5311029131153795, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.811695098876953, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8904845118522644, + "num_tokens": 772693805.0, + "step": 19897 + }, + { + "epoch": 2.5312301233939705, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.887659072875977, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.9050485491752625, + "num_tokens": 772727671.0, + "step": 19898 + }, + { + "epoch": 2.5313573336725606, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.024418830871582, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.892075777053833, + "num_tokens": 772760643.0, + "step": 19899 + }, + { + "epoch": 2.5314845439511515, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.855178833007812, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.900999903678894, + "num_tokens": 772796492.0, + "step": 19900 + }, + { + "epoch": 2.5316117542297416, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.767871856689453, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8931151628494263, + "num_tokens": 772841147.0, + "step": 19901 + }, + { + "epoch": 2.531738964508332, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.975162506103516, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8942177295684814, + "num_tokens": 772879845.0, + "step": 19902 + }, + { + "epoch": 2.5318661747869227, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.797700881958008, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8839219808578491, + "num_tokens": 772920313.0, + "step": 19903 + }, + { + "epoch": 2.531993385065513, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.900776863098145, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8982001543045044, + "num_tokens": 772959133.0, + "step": 19904 + }, + { + "epoch": 2.5321205953441037, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.786229133605957, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.9050681591033936, + "num_tokens": 772996735.0, + "step": 19905 + }, + { + "epoch": 2.5322478056226942, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.847661972045898, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9039105176925659, + "num_tokens": 773028238.0, + "step": 19906 + }, + { + "epoch": 2.5323750159012848, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 8.96453857421875e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.81778335571289, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8945921659469604, + "num_tokens": 773068932.0, + "step": 19907 + }, + { + "epoch": 2.5325022261798753, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.811994552612305, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.9076205492019653, + "num_tokens": 773104892.0, + "step": 19908 + }, + { + "epoch": 2.532629436458466, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.877118110656738, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8910385370254517, + "num_tokens": 773145759.0, + "step": 19909 + }, + { + "epoch": 2.5327566467370564, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.820091247558594, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8985767364501953, + "num_tokens": 773188122.0, + "step": 19910 + }, + { + "epoch": 2.532883857015647, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.817963600158691, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8999711275100708, + "num_tokens": 773230237.0, + "step": 19911 + }, + { + "epoch": 2.5330110672942374, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.929993629455566, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8970509767532349, + "num_tokens": 773264862.0, + "step": 19912 + }, + { + "epoch": 2.533138277572828, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.75616455078125, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.9098730683326721, + "num_tokens": 773307640.0, + "step": 19913 + }, + { + "epoch": 2.5332654878514185, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.951964378356934, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9047399759292603, + "num_tokens": 773337249.0, + "step": 19914 + }, + { + "epoch": 2.533392698130009, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.918356895446777, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8993999361991882, + "num_tokens": 773373033.0, + "step": 19915 + }, + { + "epoch": 2.5335199084085995, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 10.014852523803711, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8862776160240173, + "num_tokens": 773409668.0, + "step": 19916 + }, + { + "epoch": 2.53364711868719, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.795156478881836, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8978976011276245, + "num_tokens": 773454258.0, + "step": 19917 + }, + { + "epoch": 2.5337743289657806, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.90666389465332, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8907039165496826, + "num_tokens": 773497011.0, + "step": 19918 + }, + { + "epoch": 2.533901539244371, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.95446491241455, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8977795243263245, + "num_tokens": 773534263.0, + "step": 19919 + }, + { + "epoch": 2.534028749522961, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.838531494140625, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8848395347595215, + "num_tokens": 773577438.0, + "step": 19920 + }, + { + "epoch": 2.534155959801552, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.766196250915527, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9099459052085876, + "num_tokens": 773621287.0, + "step": 19921 + }, + { + "epoch": 2.5342831700801423, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.934677124023438, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8859236240386963, + "num_tokens": 773658518.0, + "step": 19922 + }, + { + "epoch": 2.5344103803587332, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.97201919555664, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8907092809677124, + "num_tokens": 773695918.0, + "step": 19923 + }, + { + "epoch": 2.5345375906373233, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.756804466247559, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9093321561813354, + "num_tokens": 773732769.0, + "step": 19924 + }, + { + "epoch": 2.534664800915914, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.931097984313965, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9067741632461548, + "num_tokens": 773765828.0, + "step": 19925 + }, + { + "epoch": 2.5347920111945044, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.941500663757324, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8869527578353882, + "num_tokens": 773802718.0, + "step": 19926 + }, + { + "epoch": 2.534919221473095, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.789737701416016, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8980883359909058, + "num_tokens": 773846786.0, + "step": 19927 + }, + { + "epoch": 2.5350464317516854, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.858367919921875, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8934458494186401, + "num_tokens": 773888466.0, + "step": 19928 + }, + { + "epoch": 2.535173642030276, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.913504600524902, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.900178849697113, + "num_tokens": 773924211.0, + "step": 19929 + }, + { + "epoch": 2.5353008523088665, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.842817306518555, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.9087109565734863, + "num_tokens": 773963391.0, + "step": 19930 + }, + { + "epoch": 2.535428062587457, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.835722923278809, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.912400484085083, + "num_tokens": 774004115.0, + "step": 19931 + }, + { + "epoch": 2.5355552728660475, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.904598236083984, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.898501992225647, + "num_tokens": 774039459.0, + "step": 19932 + }, + { + "epoch": 2.535682483144638, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.963068962097168, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.9042621850967407, + "num_tokens": 774073641.0, + "step": 19933 + }, + { + "epoch": 2.5358096934232286, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.878314971923828, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9123307466506958, + "num_tokens": 774107066.0, + "step": 19934 + }, + { + "epoch": 2.535936903701819, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.910529136657715, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8989108204841614, + "num_tokens": 774145094.0, + "step": 19935 + }, + { + "epoch": 2.5360641139804097, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.885749816894531, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.897747278213501, + "num_tokens": 774184868.0, + "step": 19936 + }, + { + "epoch": 2.536191324259, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.879598617553711, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.9055948853492737, + "num_tokens": 774220744.0, + "step": 19937 + }, + { + "epoch": 2.5363185345375907, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.818903923034668, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8940865993499756, + "num_tokens": 774261586.0, + "step": 19938 + }, + { + "epoch": 2.5364457448161812, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.961665153503418, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.896502673625946, + "num_tokens": 774297643.0, + "step": 19939 + }, + { + "epoch": 2.5365729550947718, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.81966781616211, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8821359872817993, + "num_tokens": 774339516.0, + "step": 19940 + }, + { + "epoch": 2.5367001653733623, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.926692962646484, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8934768438339233, + "num_tokens": 774376327.0, + "step": 19941 + }, + { + "epoch": 2.536827375651953, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.908254623413086, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.896242082118988, + "num_tokens": 774412393.0, + "step": 19942 + }, + { + "epoch": 2.5369545859305433, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.784735679626465, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8966427445411682, + "num_tokens": 774455478.0, + "step": 19943 + }, + { + "epoch": 2.537081796209134, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.873520851135254, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8913226127624512, + "num_tokens": 774492870.0, + "step": 19944 + }, + { + "epoch": 2.537209006487724, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.806157112121582, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9127469062805176, + "num_tokens": 774528838.0, + "step": 19945 + }, + { + "epoch": 2.537336216766315, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.001585960388184, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8869295716285706, + "num_tokens": 774562945.0, + "step": 19946 + }, + { + "epoch": 2.537463427044905, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.824745178222656, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.9061148166656494, + "num_tokens": 774603494.0, + "step": 19947 + }, + { + "epoch": 2.537590637323496, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.7943754196167, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8968393802642822, + "num_tokens": 774644681.0, + "step": 19948 + }, + { + "epoch": 2.537717847602086, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.846322059631348, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.9009495973587036, + "num_tokens": 774688140.0, + "step": 19949 + }, + { + "epoch": 2.5378450578806766, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.931915283203125, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8865109086036682, + "num_tokens": 774723759.0, + "step": 19950 + }, + { + "epoch": 2.537972268159267, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.812443733215332, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8964147567749023, + "num_tokens": 774762602.0, + "step": 19951 + }, + { + "epoch": 2.5380994784378577, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.897958755493164, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.9016225337982178, + "num_tokens": 774799920.0, + "step": 19952 + }, + { + "epoch": 2.538226688716448, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.930625915527344, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8907174468040466, + "num_tokens": 774840402.0, + "step": 19953 + }, + { + "epoch": 2.5383538989950387, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.844720840454102, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8963125944137573, + "num_tokens": 774880366.0, + "step": 19954 + }, + { + "epoch": 2.5384811092736292, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.978569030761719, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8872086405754089, + "num_tokens": 774911632.0, + "step": 19955 + }, + { + "epoch": 2.5386083195522198, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.862460136413574, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8913382291793823, + "num_tokens": 774948707.0, + "step": 19956 + }, + { + "epoch": 2.5387355298308103, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.730661392211914, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.9009935259819031, + "num_tokens": 774995940.0, + "step": 19957 + }, + { + "epoch": 2.538862740109401, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.95140266418457, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8990578055381775, + "num_tokens": 775032408.0, + "step": 19958 + }, + { + "epoch": 2.5389899503879914, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.883742332458496, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8892989158630371, + "num_tokens": 775067336.0, + "step": 19959 + }, + { + "epoch": 2.539117160666582, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.862789154052734, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9071611166000366, + "num_tokens": 775104359.0, + "step": 19960 + }, + { + "epoch": 2.5392443709451724, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.822720527648926, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9087001085281372, + "num_tokens": 775139631.0, + "step": 19961 + }, + { + "epoch": 2.539371581223763, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.853982925415039, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.9022111296653748, + "num_tokens": 775177094.0, + "step": 19962 + }, + { + "epoch": 2.5394987915023535, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.895134925842285, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.9001508355140686, + "num_tokens": 775214139.0, + "step": 19963 + }, + { + "epoch": 2.539626001780944, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.825397491455078, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.9047234058380127, + "num_tokens": 775253911.0, + "step": 19964 + }, + { + "epoch": 2.5397532120595345, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.784553527832031, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8935115337371826, + "num_tokens": 775300861.0, + "step": 19965 + }, + { + "epoch": 2.539880422338125, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.914443969726562, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.895280659198761, + "num_tokens": 775343288.0, + "step": 19966 + }, + { + "epoch": 2.5400076326167156, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.778305053710938, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9140974283218384, + "num_tokens": 775381533.0, + "step": 19967 + }, + { + "epoch": 2.5401348428953057, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.014854431152344, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8885543346405029, + "num_tokens": 775412878.0, + "step": 19968 + }, + { + "epoch": 2.5402620531738966, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.92383098602295, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8939129114151001, + "num_tokens": 775447101.0, + "step": 19969 + }, + { + "epoch": 2.5403892634524867, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.840339660644531, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.902633547782898, + "num_tokens": 775480720.0, + "step": 19970 + }, + { + "epoch": 2.5405164737310777, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.779424667358398, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8992007374763489, + "num_tokens": 775523737.0, + "step": 19971 + }, + { + "epoch": 2.540643684009668, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.82360553741455, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8980932235717773, + "num_tokens": 775559569.0, + "step": 19972 + }, + { + "epoch": 2.5407708942882588, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.78380298614502, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.899198055267334, + "num_tokens": 775602580.0, + "step": 19973 + }, + { + "epoch": 2.540898104566849, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.791099548339844, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.9022233486175537, + "num_tokens": 775642865.0, + "step": 19974 + }, + { + "epoch": 2.5410253148454394, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.002184867858887, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8959224820137024, + "num_tokens": 775678147.0, + "step": 19975 + }, + { + "epoch": 2.54115252512403, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.92180347442627, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8923509120941162, + "num_tokens": 775709065.0, + "step": 19976 + }, + { + "epoch": 2.5412797354026204, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.83499813079834, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8977389335632324, + "num_tokens": 775745660.0, + "step": 19977 + }, + { + "epoch": 2.541406945681211, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.828060150146484, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.894773542881012, + "num_tokens": 775783115.0, + "step": 19978 + }, + { + "epoch": 2.5415341559598015, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.962607383728027, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8964390754699707, + "num_tokens": 775818489.0, + "step": 19979 + }, + { + "epoch": 2.541661366238392, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.908220291137695, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.9032941460609436, + "num_tokens": 775850832.0, + "step": 19980 + }, + { + "epoch": 2.5417885765169825, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.8281888961792, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.9037430286407471, + "num_tokens": 775888566.0, + "step": 19981 + }, + { + "epoch": 2.541915786795573, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.783618927001953, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8990079164505005, + "num_tokens": 775933721.0, + "step": 19982 + }, + { + "epoch": 2.5420429970741636, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.906679153442383, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8945596218109131, + "num_tokens": 775968166.0, + "step": 19983 + }, + { + "epoch": 2.542170207352754, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.88465690612793, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.897140622138977, + "num_tokens": 776009834.0, + "step": 19984 + }, + { + "epoch": 2.5422974176313446, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.012222290039062e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.845306396484375, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8968864679336548, + "num_tokens": 776049817.0, + "step": 19985 + }, + { + "epoch": 2.542424627909935, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.854314804077148, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.906425952911377, + "num_tokens": 776088614.0, + "step": 19986 + }, + { + "epoch": 2.5425518381885257, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.85698127746582, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.9018388986587524, + "num_tokens": 776127655.0, + "step": 19987 + }, + { + "epoch": 2.5426790484671162, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.875534057617188, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.898789644241333, + "num_tokens": 776167391.0, + "step": 19988 + }, + { + "epoch": 2.5428062587457068, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.895331382751465, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.9055611491203308, + "num_tokens": 776202936.0, + "step": 19989 + }, + { + "epoch": 2.5429334690242973, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.872842788696289, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8877321481704712, + "num_tokens": 776241505.0, + "step": 19990 + }, + { + "epoch": 2.543060679302888, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.942520141601562, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8859737515449524, + "num_tokens": 776280442.0, + "step": 19991 + }, + { + "epoch": 2.5431878895814783, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.93156623840332, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9086079001426697, + "num_tokens": 776314381.0, + "step": 19992 + }, + { + "epoch": 2.5433150998600684, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.958552360534668, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.908394992351532, + "num_tokens": 776345440.0, + "step": 19993 + }, + { + "epoch": 2.5434423101386594, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.82628059387207, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.9031147956848145, + "num_tokens": 776385075.0, + "step": 19994 + }, + { + "epoch": 2.5435695204172495, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.880777359008789, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8965181708335876, + "num_tokens": 776422498.0, + "step": 19995 + }, + { + "epoch": 2.5436967306958405, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.028243064880371, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8961608409881592, + "num_tokens": 776453198.0, + "step": 19996 + }, + { + "epoch": 2.5438239409744305, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.159311294555664, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8935204744338989, + "num_tokens": 776494809.0, + "step": 19997 + }, + { + "epoch": 2.5439511512530215, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.78470230102539, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8960136771202087, + "num_tokens": 776537563.0, + "step": 19998 + }, + { + "epoch": 2.5440783615316116, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.842759132385254, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8985328674316406, + "num_tokens": 776581339.0, + "step": 19999 + }, + { + "epoch": 2.544205571810202, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.825565338134766, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8991575837135315, + "num_tokens": 776622872.0, + "step": 20000 + }, + { + "epoch": 2.5443327820887927, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.915928840637207, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8951309323310852, + "num_tokens": 776662125.0, + "step": 20001 + }, + { + "epoch": 2.544459992367383, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.896611213684082, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8892600536346436, + "num_tokens": 776701296.0, + "step": 20002 + }, + { + "epoch": 2.5445872026459737, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.868513107299805, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8836454153060913, + "num_tokens": 776743468.0, + "step": 20003 + }, + { + "epoch": 2.5447144129245642, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.028814315795898, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8821181058883667, + "num_tokens": 776778102.0, + "step": 20004 + }, + { + "epoch": 2.5448416232031548, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.851435661315918, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8867192268371582, + "num_tokens": 776823259.0, + "step": 20005 + }, + { + "epoch": 2.5449688334817453, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.943038940429688, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.89728844165802, + "num_tokens": 776857669.0, + "step": 20006 + }, + { + "epoch": 2.545096043760336, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.859420776367188, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.9008840322494507, + "num_tokens": 776892182.0, + "step": 20007 + }, + { + "epoch": 2.5452232540389264, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.922258377075195, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9046726226806641, + "num_tokens": 776924300.0, + "step": 20008 + }, + { + "epoch": 2.545350464317517, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.823111534118652, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9103666543960571, + "num_tokens": 776963486.0, + "step": 20009 + }, + { + "epoch": 2.5454776745961074, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.995600700378418, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8929292559623718, + "num_tokens": 776999912.0, + "step": 20010 + }, + { + "epoch": 2.545604884874698, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.862617492675781, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.9012877941131592, + "num_tokens": 777041098.0, + "step": 20011 + }, + { + "epoch": 2.5457320951532885, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.887228965759277, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8868723511695862, + "num_tokens": 777078982.0, + "step": 20012 + }, + { + "epoch": 2.545859305431879, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.933006286621094, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8981057405471802, + "num_tokens": 777113806.0, + "step": 20013 + }, + { + "epoch": 2.5459865157104695, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.82291030883789, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8972188234329224, + "num_tokens": 777153154.0, + "step": 20014 + }, + { + "epoch": 2.54611372598906, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.85972785949707, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9093426465988159, + "num_tokens": 777187118.0, + "step": 20015 + }, + { + "epoch": 2.5462409362676506, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.998785018920898, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8837712407112122, + "num_tokens": 777221232.0, + "step": 20016 + }, + { + "epoch": 2.546368146546241, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.976563453674316, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8890040516853333, + "num_tokens": 777261179.0, + "step": 20017 + }, + { + "epoch": 2.546495356824831, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.849695205688477, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8982369303703308, + "num_tokens": 777298033.0, + "step": 20018 + }, + { + "epoch": 2.546622567103422, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.93562126159668, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8931407332420349, + "num_tokens": 777334131.0, + "step": 20019 + }, + { + "epoch": 2.5467497773820122, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.8483247756958, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8944427967071533, + "num_tokens": 777372107.0, + "step": 20020 + }, + { + "epoch": 2.546876987660603, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.908072471618652, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.895209550857544, + "num_tokens": 777408654.0, + "step": 20021 + }, + { + "epoch": 2.5470041979391933, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.835355758666992, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8964879512786865, + "num_tokens": 777451499.0, + "step": 20022 + }, + { + "epoch": 2.547131408217784, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.882610321044922, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8973897695541382, + "num_tokens": 777487695.0, + "step": 20023 + }, + { + "epoch": 2.5472586184963744, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.927002906799316, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.9021450281143188, + "num_tokens": 777525396.0, + "step": 20024 + }, + { + "epoch": 2.547385828774965, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.872712135314941, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9064775705337524, + "num_tokens": 777563975.0, + "step": 20025 + }, + { + "epoch": 2.5475130390535554, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.062517166137695, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8958605527877808, + "num_tokens": 777596144.0, + "step": 20026 + }, + { + "epoch": 2.547640249332146, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.873480796813965, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8977883458137512, + "num_tokens": 777632461.0, + "step": 20027 + }, + { + "epoch": 2.5477674596107365, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.947443008422852, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8959486484527588, + "num_tokens": 777672184.0, + "step": 20028 + }, + { + "epoch": 2.547894669889327, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.825834274291992, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8925131559371948, + "num_tokens": 777713304.0, + "step": 20029 + }, + { + "epoch": 2.5480218801679175, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.890921592712402, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8923028111457825, + "num_tokens": 777749815.0, + "step": 20030 + }, + { + "epoch": 2.548149090446508, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.867545127868652, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9039347171783447, + "num_tokens": 777788243.0, + "step": 20031 + }, + { + "epoch": 2.5482763007250986, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.695623397827148, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9155256748199463, + "num_tokens": 777834834.0, + "step": 20032 + }, + { + "epoch": 2.548403511003689, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.939587593078613, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8980862498283386, + "num_tokens": 777874519.0, + "step": 20033 + }, + { + "epoch": 2.5485307212822796, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.908387184143066, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8940173983573914, + "num_tokens": 777914446.0, + "step": 20034 + }, + { + "epoch": 2.54865793156087, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.799595832824707, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.901671290397644, + "num_tokens": 777955125.0, + "step": 20035 + }, + { + "epoch": 2.5487851418394607, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.964804649353027, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8941519856452942, + "num_tokens": 777993952.0, + "step": 20036 + }, + { + "epoch": 2.5489123521180512, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.969552040100098, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8860583901405334, + "num_tokens": 778030008.0, + "step": 20037 + }, + { + "epoch": 2.5490395623966418, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.857047080993652, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8995637893676758, + "num_tokens": 778068900.0, + "step": 20038 + }, + { + "epoch": 2.5491667726752323, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.848075866699219, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9054611325263977, + "num_tokens": 778104115.0, + "step": 20039 + }, + { + "epoch": 2.549293982953823, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.799989700317383, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9014154672622681, + "num_tokens": 778151158.0, + "step": 20040 + }, + { + "epoch": 2.5494211932324133, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.893038749694824, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9009183645248413, + "num_tokens": 778193618.0, + "step": 20041 + }, + { + "epoch": 2.549548403511004, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.77824878692627, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8906806707382202, + "num_tokens": 778237007.0, + "step": 20042 + }, + { + "epoch": 2.549675613789594, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.921247482299805, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8977132439613342, + "num_tokens": 778273458.0, + "step": 20043 + }, + { + "epoch": 2.549802824068185, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.838234901428223, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8936961889266968, + "num_tokens": 778315382.0, + "step": 20044 + }, + { + "epoch": 2.549930034346775, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.932787895202637, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8950012922286987, + "num_tokens": 778354438.0, + "step": 20045 + }, + { + "epoch": 2.550057244625366, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.86111831665039, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8998532295227051, + "num_tokens": 778396930.0, + "step": 20046 + }, + { + "epoch": 2.550184454903956, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.986311912536621, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8735136389732361, + "num_tokens": 778437715.0, + "step": 20047 + }, + { + "epoch": 2.5503116651825466, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.810739517211914, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8948632478713989, + "num_tokens": 778480564.0, + "step": 20048 + }, + { + "epoch": 2.550438875461137, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.911677360534668, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8972773551940918, + "num_tokens": 778513383.0, + "step": 20049 + }, + { + "epoch": 2.5505660857397277, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.8567476272583, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8830515146255493, + "num_tokens": 778555629.0, + "step": 20050 + }, + { + "epoch": 2.550693296018318, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.974143981933594, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9069905281066895, + "num_tokens": 778586384.0, + "step": 20051 + }, + { + "epoch": 2.5508205062969087, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.007925033569336, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.895166277885437, + "num_tokens": 778617232.0, + "step": 20052 + }, + { + "epoch": 2.5509477165754992, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.869879722595215, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8972353339195251, + "num_tokens": 778656819.0, + "step": 20053 + }, + { + "epoch": 2.5510749268540898, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.867532730102539, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.89792400598526, + "num_tokens": 778696435.0, + "step": 20054 + }, + { + "epoch": 2.5512021371326803, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.782344818115234, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.902437686920166, + "num_tokens": 778739589.0, + "step": 20055 + }, + { + "epoch": 2.551329347411271, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.90772819519043, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9015903472900391, + "num_tokens": 778778214.0, + "step": 20056 + }, + { + "epoch": 2.5514565576898613, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.896142959594727, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.896080493927002, + "num_tokens": 778816178.0, + "step": 20057 + }, + { + "epoch": 2.551583767968452, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.79991626739502, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9087526798248291, + "num_tokens": 778857930.0, + "step": 20058 + }, + { + "epoch": 2.5517109782470424, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.961625099182129, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8976680040359497, + "num_tokens": 778890501.0, + "step": 20059 + }, + { + "epoch": 2.551838188525633, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.060601234436035, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.9009619951248169, + "num_tokens": 778921179.0, + "step": 20060 + }, + { + "epoch": 2.5519653988042235, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.796844482421875, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.902647078037262, + "num_tokens": 778960271.0, + "step": 20061 + }, + { + "epoch": 2.552092609082814, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.909113883972168, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.9005560874938965, + "num_tokens": 778999334.0, + "step": 20062 + }, + { + "epoch": 2.5522198193614045, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.832548141479492, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8958680629730225, + "num_tokens": 779043833.0, + "step": 20063 + }, + { + "epoch": 2.552347029639995, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.902789115905762, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8985992670059204, + "num_tokens": 779081950.0, + "step": 20064 + }, + { + "epoch": 2.5524742399185856, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.824066162109375, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.9034727215766907, + "num_tokens": 779124482.0, + "step": 20065 + }, + { + "epoch": 2.5526014501971757, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.90262508392334, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8975179195404053, + "num_tokens": 779166441.0, + "step": 20066 + }, + { + "epoch": 2.5527286604757666, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.973001480102539, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8842467069625854, + "num_tokens": 779203191.0, + "step": 20067 + }, + { + "epoch": 2.5528558707543567, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.813850402832031, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8980982303619385, + "num_tokens": 779248998.0, + "step": 20068 + }, + { + "epoch": 2.5529830810329477, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.81135368347168, + "learning_rate": 1e-06, + "loss": 0.2629, + "mean_token_accuracy": 0.9109950065612793, + "num_tokens": 779290323.0, + "step": 20069 + }, + { + "epoch": 2.5531102913115378, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.877614974975586, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.9020437002182007, + "num_tokens": 779328923.0, + "step": 20070 + }, + { + "epoch": 2.5532375015901287, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.007688522338867, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8968830108642578, + "num_tokens": 779363405.0, + "step": 20071 + }, + { + "epoch": 2.553364711868719, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.871849060058594, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.9014050364494324, + "num_tokens": 779402258.0, + "step": 20072 + }, + { + "epoch": 2.5534919221473094, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.80951976776123, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9119004011154175, + "num_tokens": 779440293.0, + "step": 20073 + }, + { + "epoch": 2.5536191324259, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.826642990112305, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9106112122535706, + "num_tokens": 779481872.0, + "step": 20074 + }, + { + "epoch": 2.5537463427044904, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.978752136230469, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8974617719650269, + "num_tokens": 779517550.0, + "step": 20075 + }, + { + "epoch": 2.553873552983081, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.918900489807129, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.895000696182251, + "num_tokens": 779561724.0, + "step": 20076 + }, + { + "epoch": 2.5540007632616715, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.85350513458252, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8955546021461487, + "num_tokens": 779597380.0, + "step": 20077 + }, + { + "epoch": 2.554127973540262, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.895319938659668, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8873569965362549, + "num_tokens": 779634226.0, + "step": 20078 + }, + { + "epoch": 2.5542551838188525, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.877791404724121, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8892812132835388, + "num_tokens": 779672326.0, + "step": 20079 + }, + { + "epoch": 2.554382394097443, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.831031799316406, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9111499786376953, + "num_tokens": 779708356.0, + "step": 20080 + }, + { + "epoch": 2.5545096043760336, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.012475967407227, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8794141411781311, + "num_tokens": 779742683.0, + "step": 20081 + }, + { + "epoch": 2.554636814654624, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.059906005859375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.89975357055664, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8968039155006409, + "num_tokens": 779780149.0, + "step": 20082 + }, + { + "epoch": 2.5547640249332146, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.838350296020508, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.889886200428009, + "num_tokens": 779823716.0, + "step": 20083 + }, + { + "epoch": 2.554891235211805, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.870139122009277, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.901024580001831, + "num_tokens": 779862369.0, + "step": 20084 + }, + { + "epoch": 2.5550184454903957, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.755826950073242, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8937521576881409, + "num_tokens": 779906229.0, + "step": 20085 + }, + { + "epoch": 2.5551456557689862, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.798392295837402, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.9107711911201477, + "num_tokens": 779944841.0, + "step": 20086 + }, + { + "epoch": 2.5552728660475768, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.822959899902344, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8980579376220703, + "num_tokens": 779986610.0, + "step": 20087 + }, + { + "epoch": 2.5554000763261673, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.763136863708496, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9057105779647827, + "num_tokens": 780028960.0, + "step": 20088 + }, + { + "epoch": 2.555527286604758, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.939362525939941, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9027067422866821, + "num_tokens": 780060970.0, + "step": 20089 + }, + { + "epoch": 2.5556544968833483, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.905765533447266, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8969639539718628, + "num_tokens": 780101456.0, + "step": 20090 + }, + { + "epoch": 2.5557817071619384, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.807807922363281, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9057601690292358, + "num_tokens": 780142934.0, + "step": 20091 + }, + { + "epoch": 2.5559089174405294, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.975740432739258, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8907051086425781, + "num_tokens": 780181033.0, + "step": 20092 + }, + { + "epoch": 2.5560361277191195, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.88567066192627, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.898045539855957, + "num_tokens": 780218821.0, + "step": 20093 + }, + { + "epoch": 2.5561633379977104, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.030603408813477, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8961280584335327, + "num_tokens": 780254777.0, + "step": 20094 + }, + { + "epoch": 2.5562905482763005, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.936314582824707, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9018890857696533, + "num_tokens": 780287350.0, + "step": 20095 + }, + { + "epoch": 2.5564177585548915, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.808121681213379, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9085806012153625, + "num_tokens": 780324740.0, + "step": 20096 + }, + { + "epoch": 2.5565449688334816, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.781886100769043, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.9069561958312988, + "num_tokens": 780367541.0, + "step": 20097 + }, + { + "epoch": 2.556672179112072, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.931121826171875, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8897238969802856, + "num_tokens": 780407236.0, + "step": 20098 + }, + { + "epoch": 2.5567993893906626, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.835033416748047, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9070993065834045, + "num_tokens": 780442370.0, + "step": 20099 + }, + { + "epoch": 2.556926599669253, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.0667142868042, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8873379230499268, + "num_tokens": 780479063.0, + "step": 20100 + }, + { + "epoch": 2.5570538099478437, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.950058937072754, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8934217691421509, + "num_tokens": 780512675.0, + "step": 20101 + }, + { + "epoch": 2.5571810202264342, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.083158493041992, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8954219818115234, + "num_tokens": 780549627.0, + "step": 20102 + }, + { + "epoch": 2.5573082305050248, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.894451141357422, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8940520286560059, + "num_tokens": 780592248.0, + "step": 20103 + }, + { + "epoch": 2.5574354407836153, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.989293098449707, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8845363855361938, + "num_tokens": 780634972.0, + "step": 20104 + }, + { + "epoch": 2.557562651062206, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.99343204498291, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8905578851699829, + "num_tokens": 780669774.0, + "step": 20105 + }, + { + "epoch": 2.5576898613407963, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.83009147644043, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8953624367713928, + "num_tokens": 780705351.0, + "step": 20106 + }, + { + "epoch": 2.557817071619387, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.890679359436035, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8968597054481506, + "num_tokens": 780741698.0, + "step": 20107 + }, + { + "epoch": 2.5579442818979774, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.911699295043945, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.907250165939331, + "num_tokens": 780776268.0, + "step": 20108 + }, + { + "epoch": 2.558071492176568, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.93007755279541, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8864914178848267, + "num_tokens": 780813252.0, + "step": 20109 + }, + { + "epoch": 2.5581987024551585, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.807793617248535, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.908197283744812, + "num_tokens": 780854909.0, + "step": 20110 + }, + { + "epoch": 2.558325912733749, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.82944107055664, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8953434228897095, + "num_tokens": 780896045.0, + "step": 20111 + }, + { + "epoch": 2.5584531230123395, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.050955772399902, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8827914595603943, + "num_tokens": 780927203.0, + "step": 20112 + }, + { + "epoch": 2.55858033329093, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.783303260803223, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.900755763053894, + "num_tokens": 780966769.0, + "step": 20113 + }, + { + "epoch": 2.5587075435695206, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.824620246887207, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8910281658172607, + "num_tokens": 781005705.0, + "step": 20114 + }, + { + "epoch": 2.558834753848111, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.875486373901367, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8902723789215088, + "num_tokens": 781047502.0, + "step": 20115 + }, + { + "epoch": 2.558961964126701, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.880158424377441, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.899665117263794, + "num_tokens": 781083071.0, + "step": 20116 + }, + { + "epoch": 2.559089174405292, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.933889389038086, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8958109021186829, + "num_tokens": 781120408.0, + "step": 20117 + }, + { + "epoch": 2.5592163846838822, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.765876770019531, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.89897620677948, + "num_tokens": 781161072.0, + "step": 20118 + }, + { + "epoch": 2.559343594962473, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.812948226928711, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.9001925587654114, + "num_tokens": 781198674.0, + "step": 20119 + }, + { + "epoch": 2.5594708052410633, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.870214462280273, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8938523530960083, + "num_tokens": 781237596.0, + "step": 20120 + }, + { + "epoch": 2.559598015519654, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.845443725585938, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8968416452407837, + "num_tokens": 781274122.0, + "step": 20121 + }, + { + "epoch": 2.5597252257982444, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.823382377624512, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.9040585160255432, + "num_tokens": 781314086.0, + "step": 20122 + }, + { + "epoch": 2.559852436076835, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.846528053283691, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.9080521464347839, + "num_tokens": 781354622.0, + "step": 20123 + }, + { + "epoch": 2.5599796463554254, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.843811988830566, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9089101552963257, + "num_tokens": 781390095.0, + "step": 20124 + }, + { + "epoch": 2.560106856634016, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.923347473144531, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8895538449287415, + "num_tokens": 781433440.0, + "step": 20125 + }, + { + "epoch": 2.5602340669126065, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.734947204589844, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8956495523452759, + "num_tokens": 781472733.0, + "step": 20126 + }, + { + "epoch": 2.560361277191197, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.885997772216797, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8992906212806702, + "num_tokens": 781514163.0, + "step": 20127 + }, + { + "epoch": 2.5604884874697875, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.86378288269043, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9076005220413208, + "num_tokens": 781548524.0, + "step": 20128 + }, + { + "epoch": 2.560615697748378, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.900476455688477, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.9052378535270691, + "num_tokens": 781581476.0, + "step": 20129 + }, + { + "epoch": 2.5607429080269686, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.003399848937988, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8911027312278748, + "num_tokens": 781616238.0, + "step": 20130 + }, + { + "epoch": 2.560870118305559, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.811912536621094, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8924310207366943, + "num_tokens": 781658930.0, + "step": 20131 + }, + { + "epoch": 2.5609973285841496, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.868975639343262, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.903827965259552, + "num_tokens": 781694045.0, + "step": 20132 + }, + { + "epoch": 2.56112453886274, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.842982292175293, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.9053128957748413, + "num_tokens": 781732887.0, + "step": 20133 + }, + { + "epoch": 2.5612517491413307, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.786792755126953, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9107820391654968, + "num_tokens": 781774688.0, + "step": 20134 + }, + { + "epoch": 2.561378959419921, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.963923454284668, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8937520980834961, + "num_tokens": 781817020.0, + "step": 20135 + }, + { + "epoch": 2.5615061696985117, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.95527458190918, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9001960754394531, + "num_tokens": 781851895.0, + "step": 20136 + }, + { + "epoch": 2.5616333799771023, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.952620506286621, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8952114582061768, + "num_tokens": 781886630.0, + "step": 20137 + }, + { + "epoch": 2.561760590255693, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.921051025390625, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8954960703849792, + "num_tokens": 781921111.0, + "step": 20138 + }, + { + "epoch": 2.5618878005342833, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.814010620117188, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9127531051635742, + "num_tokens": 781962319.0, + "step": 20139 + }, + { + "epoch": 2.562015010812874, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.93423843383789, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9091277122497559, + "num_tokens": 781994625.0, + "step": 20140 + }, + { + "epoch": 2.562142221091464, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.928034782409668, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9104405641555786, + "num_tokens": 782028730.0, + "step": 20141 + }, + { + "epoch": 2.562269431370055, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.927624702453613, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9019339084625244, + "num_tokens": 782064412.0, + "step": 20142 + }, + { + "epoch": 2.562396641648645, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.984631538391113, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8997447490692139, + "num_tokens": 782097004.0, + "step": 20143 + }, + { + "epoch": 2.562523851927236, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.77096939086914, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9086917042732239, + "num_tokens": 782140166.0, + "step": 20144 + }, + { + "epoch": 2.562651062205826, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.97922420501709, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8978606462478638, + "num_tokens": 782177429.0, + "step": 20145 + }, + { + "epoch": 2.5627782724844166, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.847105026245117, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8957386016845703, + "num_tokens": 782219930.0, + "step": 20146 + }, + { + "epoch": 2.562905482763007, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.874124526977539, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9014337062835693, + "num_tokens": 782261149.0, + "step": 20147 + }, + { + "epoch": 2.5630326930415976, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.0137300491333, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.89565509557724, + "num_tokens": 782294222.0, + "step": 20148 + }, + { + "epoch": 2.563159903320188, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.107589721679688e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.909173011779785, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8893614411354065, + "num_tokens": 782333705.0, + "step": 20149 + }, + { + "epoch": 2.5632871135987787, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.95598316192627, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8889853954315186, + "num_tokens": 782368007.0, + "step": 20150 + }, + { + "epoch": 2.5634143238773692, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.92704963684082, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.9002630710601807, + "num_tokens": 782408558.0, + "step": 20151 + }, + { + "epoch": 2.5635415341559598, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.870871543884277, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.9006458520889282, + "num_tokens": 782447567.0, + "step": 20152 + }, + { + "epoch": 2.5636687444345503, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.838242530822754, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9024173021316528, + "num_tokens": 782491448.0, + "step": 20153 + }, + { + "epoch": 2.563795954713141, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.70385456085205, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9101839065551758, + "num_tokens": 782532968.0, + "step": 20154 + }, + { + "epoch": 2.5639231649917313, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.987569808959961, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.9057228565216064, + "num_tokens": 782577144.0, + "step": 20155 + }, + { + "epoch": 2.564050375270322, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.876912117004395, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.9073913097381592, + "num_tokens": 782617340.0, + "step": 20156 + }, + { + "epoch": 2.5641775855489124, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.889920234680176, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8959194421768188, + "num_tokens": 782655721.0, + "step": 20157 + }, + { + "epoch": 2.564304795827503, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.879572868347168, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.897822916507721, + "num_tokens": 782701493.0, + "step": 20158 + }, + { + "epoch": 2.5644320061060935, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.865599632263184, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.900182843208313, + "num_tokens": 782741395.0, + "step": 20159 + }, + { + "epoch": 2.564559216384684, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.018776893615723, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8787397742271423, + "num_tokens": 782779343.0, + "step": 20160 + }, + { + "epoch": 2.5646864266632745, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.915277481079102, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.906633734703064, + "num_tokens": 782819977.0, + "step": 20161 + }, + { + "epoch": 2.564813636941865, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.963617324829102, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8811119794845581, + "num_tokens": 782860235.0, + "step": 20162 + }, + { + "epoch": 2.5649408472204556, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.777345657348633, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8986865878105164, + "num_tokens": 782903266.0, + "step": 20163 + }, + { + "epoch": 2.5650680574990457, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.902517318725586, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9080969095230103, + "num_tokens": 782938792.0, + "step": 20164 + }, + { + "epoch": 2.5651952677776366, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.845061302185059, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.9075164794921875, + "num_tokens": 782980577.0, + "step": 20165 + }, + { + "epoch": 2.5653224780562267, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.923213958740234, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8856414556503296, + "num_tokens": 783019241.0, + "step": 20166 + }, + { + "epoch": 2.5654496883348177, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.994823455810547, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9075811505317688, + "num_tokens": 783050557.0, + "step": 20167 + }, + { + "epoch": 2.5655768986134078, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.898983001708984, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.89495450258255, + "num_tokens": 783088036.0, + "step": 20168 + }, + { + "epoch": 2.5657041088919987, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.773685455322266, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8932376503944397, + "num_tokens": 783134244.0, + "step": 20169 + }, + { + "epoch": 2.565831319170589, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.835808753967285, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9078889489173889, + "num_tokens": 783175837.0, + "step": 20170 + }, + { + "epoch": 2.5659585294491793, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.93319320678711, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9024257063865662, + "num_tokens": 783212791.0, + "step": 20171 + }, + { + "epoch": 2.56608573972777, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.905384063720703, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8931289911270142, + "num_tokens": 783249357.0, + "step": 20172 + }, + { + "epoch": 2.5662129500063604, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.80844783782959, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.9037339687347412, + "num_tokens": 783290166.0, + "step": 20173 + }, + { + "epoch": 2.566340160284951, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.912994384765625, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.9032731056213379, + "num_tokens": 783326354.0, + "step": 20174 + }, + { + "epoch": 2.5664673705635415, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.822121620178223, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.897713303565979, + "num_tokens": 783366415.0, + "step": 20175 + }, + { + "epoch": 2.566594580842132, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.906047821044922, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.906335711479187, + "num_tokens": 783402843.0, + "step": 20176 + }, + { + "epoch": 2.5667217911207225, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.850449562072754, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8978253602981567, + "num_tokens": 783446118.0, + "step": 20177 + }, + { + "epoch": 2.566849001399313, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.891485214233398, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8948329091072083, + "num_tokens": 783483949.0, + "step": 20178 + }, + { + "epoch": 2.5669762116779036, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.92778205871582, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.89706951379776, + "num_tokens": 783524692.0, + "step": 20179 + }, + { + "epoch": 2.567103421956494, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.953052520751953, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.9012687802314758, + "num_tokens": 783560581.0, + "step": 20180 + }, + { + "epoch": 2.5672306322350846, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.93253231048584, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.903142511844635, + "num_tokens": 783594016.0, + "step": 20181 + }, + { + "epoch": 2.567357842513675, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.775520324707031, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9048557281494141, + "num_tokens": 783635613.0, + "step": 20182 + }, + { + "epoch": 2.5674850527922657, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.891246795654297, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8944372534751892, + "num_tokens": 783675752.0, + "step": 20183 + }, + { + "epoch": 2.567612263070856, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.005824089050293, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8959579467773438, + "num_tokens": 783708794.0, + "step": 20184 + }, + { + "epoch": 2.5677394733494467, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.864076614379883, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8965606689453125, + "num_tokens": 783752434.0, + "step": 20185 + }, + { + "epoch": 2.5678666836280373, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.803374290466309, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9122787714004517, + "num_tokens": 783794841.0, + "step": 20186 + }, + { + "epoch": 2.567993893906628, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.944637298583984, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8872098922729492, + "num_tokens": 783833415.0, + "step": 20187 + }, + { + "epoch": 2.5681211041852183, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.907696723937988, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8906278610229492, + "num_tokens": 783871930.0, + "step": 20188 + }, + { + "epoch": 2.5682483144638084, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.90317153930664, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8914528489112854, + "num_tokens": 783906319.0, + "step": 20189 + }, + { + "epoch": 2.5683755247423994, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.816965103149414, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9022136926651001, + "num_tokens": 783948850.0, + "step": 20190 + }, + { + "epoch": 2.5685027350209895, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.868083953857422, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.9017248153686523, + "num_tokens": 783990667.0, + "step": 20191 + }, + { + "epoch": 2.5686299452995804, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.885969161987305, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.896178662776947, + "num_tokens": 784031905.0, + "step": 20192 + }, + { + "epoch": 2.5687571555781705, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.969056129455566, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.9007706642150879, + "num_tokens": 784062796.0, + "step": 20193 + }, + { + "epoch": 2.5688843658567615, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.769430160522461, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8947463035583496, + "num_tokens": 784104013.0, + "step": 20194 + }, + { + "epoch": 2.5690115761353516, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.880598068237305, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8896983861923218, + "num_tokens": 784150215.0, + "step": 20195 + }, + { + "epoch": 2.569138786413942, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.825439453125, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.9018347859382629, + "num_tokens": 784191849.0, + "step": 20196 + }, + { + "epoch": 2.5692659966925326, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.8731107711792, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8911212086677551, + "num_tokens": 784231321.0, + "step": 20197 + }, + { + "epoch": 2.569393206971123, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.796879768371582, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9118303656578064, + "num_tokens": 784272573.0, + "step": 20198 + }, + { + "epoch": 2.5695204172497137, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.96198844909668, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8961663246154785, + "num_tokens": 784308086.0, + "step": 20199 + }, + { + "epoch": 2.5696476275283042, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.958696365356445, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8941282033920288, + "num_tokens": 784347521.0, + "step": 20200 + }, + { + "epoch": 2.5697748378068948, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.953597068786621, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8740036487579346, + "num_tokens": 784389126.0, + "step": 20201 + }, + { + "epoch": 2.5699020480854853, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.895920753479004, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8901844024658203, + "num_tokens": 784426898.0, + "step": 20202 + }, + { + "epoch": 2.570029258364076, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.93787670135498, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8806462287902832, + "num_tokens": 784465342.0, + "step": 20203 + }, + { + "epoch": 2.5701564686426663, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.094076156616211, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.885804295539856, + "num_tokens": 784500662.0, + "step": 20204 + }, + { + "epoch": 2.570283678921257, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.79170036315918, + "learning_rate": 1e-06, + "loss": 0.2591, + "mean_token_accuracy": 0.9135116338729858, + "num_tokens": 784538495.0, + "step": 20205 + }, + { + "epoch": 2.5704108891998474, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.899365425109863, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8869823217391968, + "num_tokens": 784577264.0, + "step": 20206 + }, + { + "epoch": 2.570538099478438, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.911404609680176, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.9034769535064697, + "num_tokens": 784614043.0, + "step": 20207 + }, + { + "epoch": 2.5706653097570284, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.92509937286377, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.9015994668006897, + "num_tokens": 784644539.0, + "step": 20208 + }, + { + "epoch": 2.570792520035619, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.920746803283691, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8983027935028076, + "num_tokens": 784687687.0, + "step": 20209 + }, + { + "epoch": 2.5709197303142095, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.888864517211914, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.9001289010047913, + "num_tokens": 784727179.0, + "step": 20210 + }, + { + "epoch": 2.5710469405928, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.916792869567871, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8987805843353271, + "num_tokens": 784765435.0, + "step": 20211 + }, + { + "epoch": 2.5711741508713906, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.908513069152832, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8921718597412109, + "num_tokens": 784805988.0, + "step": 20212 + }, + { + "epoch": 2.571301361149981, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.940569877624512, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.9013712406158447, + "num_tokens": 784837113.0, + "step": 20213 + }, + { + "epoch": 2.571428571428571, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.940942764282227, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.899330735206604, + "num_tokens": 784873707.0, + "step": 20214 + }, + { + "epoch": 2.571555781707162, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.933164596557617, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8935930728912354, + "num_tokens": 784908899.0, + "step": 20215 + }, + { + "epoch": 2.5716829919857522, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.909200668334961, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8903770446777344, + "num_tokens": 784942897.0, + "step": 20216 + }, + { + "epoch": 2.571810202264343, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.83338451385498, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9047800302505493, + "num_tokens": 784982783.0, + "step": 20217 + }, + { + "epoch": 2.5719374125429333, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.932088851928711, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.891648530960083, + "num_tokens": 785019300.0, + "step": 20218 + }, + { + "epoch": 2.572064622821524, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.1552734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.936023712158203, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8953408598899841, + "num_tokens": 785061096.0, + "step": 20219 + }, + { + "epoch": 2.5721918331001143, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.805718421936035, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9145858883857727, + "num_tokens": 785096095.0, + "step": 20220 + }, + { + "epoch": 2.572319043378705, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.868609428405762, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8941507339477539, + "num_tokens": 785136142.0, + "step": 20221 + }, + { + "epoch": 2.5724462536572954, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.826507568359375, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.9006098508834839, + "num_tokens": 785176011.0, + "step": 20222 + }, + { + "epoch": 2.572573463935886, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.897233009338379, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9044635891914368, + "num_tokens": 785213611.0, + "step": 20223 + }, + { + "epoch": 2.5727006742144765, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.875496864318848, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9010365605354309, + "num_tokens": 785250607.0, + "step": 20224 + }, + { + "epoch": 2.572827884493067, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.86514663696289, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8968312740325928, + "num_tokens": 785292523.0, + "step": 20225 + }, + { + "epoch": 2.5729550947716575, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.022340774536133, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8844031095504761, + "num_tokens": 785325781.0, + "step": 20226 + }, + { + "epoch": 2.573082305050248, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.819546699523926, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8956466913223267, + "num_tokens": 785370489.0, + "step": 20227 + }, + { + "epoch": 2.5732095153288386, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.912200927734375, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8913097977638245, + "num_tokens": 785407782.0, + "step": 20228 + }, + { + "epoch": 2.573336725607429, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.96590805053711, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.9013202786445618, + "num_tokens": 785443785.0, + "step": 20229 + }, + { + "epoch": 2.5734639358860196, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.877899169921875, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9033639430999756, + "num_tokens": 785479662.0, + "step": 20230 + }, + { + "epoch": 2.57359114616461, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.016035079956055, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.9002946615219116, + "num_tokens": 785511730.0, + "step": 20231 + }, + { + "epoch": 2.5737183564432007, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.885578155517578, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8936554193496704, + "num_tokens": 785547754.0, + "step": 20232 + }, + { + "epoch": 2.573845566721791, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.824947357177734, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9075171947479248, + "num_tokens": 785588912.0, + "step": 20233 + }, + { + "epoch": 2.5739727770003817, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.861661911010742, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.9001563787460327, + "num_tokens": 785624826.0, + "step": 20234 + }, + { + "epoch": 2.5740999872789723, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.9799165725708, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8880821466445923, + "num_tokens": 785663656.0, + "step": 20235 + }, + { + "epoch": 2.574227197557563, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.812071800231934, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8992719054222107, + "num_tokens": 785712359.0, + "step": 20236 + }, + { + "epoch": 2.5743544078361533, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.853562355041504, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8955380320549011, + "num_tokens": 785756690.0, + "step": 20237 + }, + { + "epoch": 2.574481618114744, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.963692665100098, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8976668119430542, + "num_tokens": 785790732.0, + "step": 20238 + }, + { + "epoch": 2.574608828393334, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.852642059326172, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9054269790649414, + "num_tokens": 785828918.0, + "step": 20239 + }, + { + "epoch": 2.574736038671925, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.877452850341797, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.9020606279373169, + "num_tokens": 785868439.0, + "step": 20240 + }, + { + "epoch": 2.574863248950515, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.829246520996094, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.9059008359909058, + "num_tokens": 785909655.0, + "step": 20241 + }, + { + "epoch": 2.574990459229106, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.884309768676758, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.900301992893219, + "num_tokens": 785951398.0, + "step": 20242 + }, + { + "epoch": 2.575117669507696, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.971712112426758, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8958848118782043, + "num_tokens": 785988024.0, + "step": 20243 + }, + { + "epoch": 2.5752448797862866, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.860729217529297, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9102261662483215, + "num_tokens": 786033078.0, + "step": 20244 + }, + { + "epoch": 2.575372090064877, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.908884048461914, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8915199041366577, + "num_tokens": 786074555.0, + "step": 20245 + }, + { + "epoch": 2.5754993003434676, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.855408668518066, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9121947288513184, + "num_tokens": 786109421.0, + "step": 20246 + }, + { + "epoch": 2.575626510622058, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.886212348937988, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8953942656517029, + "num_tokens": 786150355.0, + "step": 20247 + }, + { + "epoch": 2.5757537209006487, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.896133422851562, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8990775346755981, + "num_tokens": 786190918.0, + "step": 20248 + }, + { + "epoch": 2.575880931179239, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.848200798034668, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.9064540863037109, + "num_tokens": 786226670.0, + "step": 20249 + }, + { + "epoch": 2.5760081414578297, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.907191276550293, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9055734872817993, + "num_tokens": 786267377.0, + "step": 20250 + }, + { + "epoch": 2.5761353517364203, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.905868530273438, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.904531717300415, + "num_tokens": 786302869.0, + "step": 20251 + }, + { + "epoch": 2.576262562015011, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.921076774597168, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8869270086288452, + "num_tokens": 786343627.0, + "step": 20252 + }, + { + "epoch": 2.5763897722936013, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.88119888305664, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9020443558692932, + "num_tokens": 786379280.0, + "step": 20253 + }, + { + "epoch": 2.576516982572192, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.874707221984863, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.9005893468856812, + "num_tokens": 786416050.0, + "step": 20254 + }, + { + "epoch": 2.5766441928507824, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.795701026916504, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9114499092102051, + "num_tokens": 786456753.0, + "step": 20255 + }, + { + "epoch": 2.576771403129373, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.872416496276855, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9019728899002075, + "num_tokens": 786494628.0, + "step": 20256 + }, + { + "epoch": 2.5768986134079634, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.969320297241211, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8924281597137451, + "num_tokens": 786528822.0, + "step": 20257 + }, + { + "epoch": 2.577025823686554, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.934606552124023, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.9024231433868408, + "num_tokens": 786567228.0, + "step": 20258 + }, + { + "epoch": 2.5771530339651445, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.759041786193848, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9205279350280762, + "num_tokens": 786605104.0, + "step": 20259 + }, + { + "epoch": 2.577280244243735, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.912504196166992, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9036216139793396, + "num_tokens": 786640349.0, + "step": 20260 + }, + { + "epoch": 2.5774074545223256, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.90638256072998, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8983480334281921, + "num_tokens": 786679963.0, + "step": 20261 + }, + { + "epoch": 2.5775346648009156, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.91014575958252, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.9005230665206909, + "num_tokens": 786714750.0, + "step": 20262 + }, + { + "epoch": 2.5776618750795066, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.798050880432129, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8976415991783142, + "num_tokens": 786756654.0, + "step": 20263 + }, + { + "epoch": 2.5777890853580967, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.858824729919434, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8946080207824707, + "num_tokens": 786795909.0, + "step": 20264 + }, + { + "epoch": 2.5779162956366877, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.839729309082031, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9026058912277222, + "num_tokens": 786836568.0, + "step": 20265 + }, + { + "epoch": 2.5780435059152778, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.938371658325195, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8910664916038513, + "num_tokens": 786867846.0, + "step": 20266 + }, + { + "epoch": 2.5781707161938687, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.826290130615234, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9073147773742676, + "num_tokens": 786910142.0, + "step": 20267 + }, + { + "epoch": 2.578297926472459, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.87182331085205, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8935384750366211, + "num_tokens": 786948343.0, + "step": 20268 + }, + { + "epoch": 2.5784251367510493, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.904740333557129, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8961825370788574, + "num_tokens": 786986478.0, + "step": 20269 + }, + { + "epoch": 2.57855234702964, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.828845024108887, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.9014623165130615, + "num_tokens": 787026312.0, + "step": 20270 + }, + { + "epoch": 2.5786795573082304, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.840713500976562, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9119181632995605, + "num_tokens": 787064948.0, + "step": 20271 + }, + { + "epoch": 2.578806767586821, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.90054988861084, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9085723161697388, + "num_tokens": 787097650.0, + "step": 20272 + }, + { + "epoch": 2.5789339778654115, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.863462448120117, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8901294469833374, + "num_tokens": 787135769.0, + "step": 20273 + }, + { + "epoch": 2.579061188144002, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.807129859924316, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9107852578163147, + "num_tokens": 787177827.0, + "step": 20274 + }, + { + "epoch": 2.5791883984225925, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.805986404418945, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8987264037132263, + "num_tokens": 787218950.0, + "step": 20275 + }, + { + "epoch": 2.579315608701183, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.786015510559082, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.9017167091369629, + "num_tokens": 787262316.0, + "step": 20276 + }, + { + "epoch": 2.5794428189797736, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.922832489013672, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8965190649032593, + "num_tokens": 787301171.0, + "step": 20277 + }, + { + "epoch": 2.579570029258364, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.872233390808105, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8959214091300964, + "num_tokens": 787337037.0, + "step": 20278 + }, + { + "epoch": 2.5796972395369546, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.924560546875, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8855419158935547, + "num_tokens": 787379089.0, + "step": 20279 + }, + { + "epoch": 2.579824449815545, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.806265830993652, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.9023193717002869, + "num_tokens": 787416304.0, + "step": 20280 + }, + { + "epoch": 2.5799516600941357, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.870092391967773, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8876112699508667, + "num_tokens": 787459229.0, + "step": 20281 + }, + { + "epoch": 2.580078870372726, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.879739761352539, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.903367280960083, + "num_tokens": 787498148.0, + "step": 20282 + }, + { + "epoch": 2.5802060806513167, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.82265567779541, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9041392803192139, + "num_tokens": 787539281.0, + "step": 20283 + }, + { + "epoch": 2.5803332909299073, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.842050552368164, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.9095025062561035, + "num_tokens": 787579253.0, + "step": 20284 + }, + { + "epoch": 2.580460501208498, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.979399681091309, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8948578834533691, + "num_tokens": 787613897.0, + "step": 20285 + }, + { + "epoch": 2.5805877114870883, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.844436645507812, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8931816220283508, + "num_tokens": 787654556.0, + "step": 20286 + }, + { + "epoch": 2.5807149217656784, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.83564281463623, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8935304880142212, + "num_tokens": 787698127.0, + "step": 20287 + }, + { + "epoch": 2.5808421320442694, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.926312446594238, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8947916626930237, + "num_tokens": 787734788.0, + "step": 20288 + }, + { + "epoch": 2.5809693423228595, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.870306015014648, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8888348937034607, + "num_tokens": 787774317.0, + "step": 20289 + }, + { + "epoch": 2.5810965526014504, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.759796142578125, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.908869743347168, + "num_tokens": 787817370.0, + "step": 20290 + }, + { + "epoch": 2.5812237628800405, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.976675987243652, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8938022255897522, + "num_tokens": 787854589.0, + "step": 20291 + }, + { + "epoch": 2.5813509731586315, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.742851257324219, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8912605047225952, + "num_tokens": 787901810.0, + "step": 20292 + }, + { + "epoch": 2.5814781834372216, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.87508487701416, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.9030202627182007, + "num_tokens": 787939768.0, + "step": 20293 + }, + { + "epoch": 2.581605393715812, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.813855171203613, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9084641933441162, + "num_tokens": 787979713.0, + "step": 20294 + }, + { + "epoch": 2.5817326039944026, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.848788261413574, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8882976174354553, + "num_tokens": 788020769.0, + "step": 20295 + }, + { + "epoch": 2.581859814272993, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.831765174865723, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.9055420756340027, + "num_tokens": 788066413.0, + "step": 20296 + }, + { + "epoch": 2.5819870245515837, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.803421974182129, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.9034001231193542, + "num_tokens": 788105922.0, + "step": 20297 + }, + { + "epoch": 2.582114234830174, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.933485984802246, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.896260678768158, + "num_tokens": 788143151.0, + "step": 20298 + }, + { + "epoch": 2.5822414451087647, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.855477333068848, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8911826610565186, + "num_tokens": 788183993.0, + "step": 20299 + }, + { + "epoch": 2.5823686553873553, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.93067741394043, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8949350118637085, + "num_tokens": 788222985.0, + "step": 20300 + }, + { + "epoch": 2.582495865665946, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.772392272949219, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9160379767417908, + "num_tokens": 788265795.0, + "step": 20301 + }, + { + "epoch": 2.5826230759445363, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 10.033183097839355, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8774084448814392, + "num_tokens": 788302885.0, + "step": 20302 + }, + { + "epoch": 2.582750286223127, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.887341499328613, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9124965071678162, + "num_tokens": 788340505.0, + "step": 20303 + }, + { + "epoch": 2.5828774965017174, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.888965606689453, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8898881077766418, + "num_tokens": 788376113.0, + "step": 20304 + }, + { + "epoch": 2.583004706780308, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.926338195800781, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8918930292129517, + "num_tokens": 788411251.0, + "step": 20305 + }, + { + "epoch": 2.5831319170588984, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.73740291595459, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.9020441174507141, + "num_tokens": 788450042.0, + "step": 20306 + }, + { + "epoch": 2.583259127337489, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.92048168182373, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8921586275100708, + "num_tokens": 788491119.0, + "step": 20307 + }, + { + "epoch": 2.5833863376160795, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.864699363708496, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.900687038898468, + "num_tokens": 788527655.0, + "step": 20308 + }, + { + "epoch": 2.58351354789467, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.960710525512695, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.9024631977081299, + "num_tokens": 788561022.0, + "step": 20309 + }, + { + "epoch": 2.5836407581732606, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.839341163635254, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8927664756774902, + "num_tokens": 788600972.0, + "step": 20310 + }, + { + "epoch": 2.583767968451851, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.911901473999023, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8911714553833008, + "num_tokens": 788636683.0, + "step": 20311 + }, + { + "epoch": 2.583895178730441, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.898425102233887, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9021629095077515, + "num_tokens": 788670755.0, + "step": 20312 + }, + { + "epoch": 2.584022389009032, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.800882339477539, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9027674198150635, + "num_tokens": 788710614.0, + "step": 20313 + }, + { + "epoch": 2.5841495992876222, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.852006912231445, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9024922847747803, + "num_tokens": 788747846.0, + "step": 20314 + }, + { + "epoch": 2.584276809566213, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.986898422241211, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8889481425285339, + "num_tokens": 788788269.0, + "step": 20315 + }, + { + "epoch": 2.5844040198448033, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.805182456970215, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.9084028005599976, + "num_tokens": 788823282.0, + "step": 20316 + }, + { + "epoch": 2.584531230123394, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.863755226135254, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.887926459312439, + "num_tokens": 788870594.0, + "step": 20317 + }, + { + "epoch": 2.5846584404019843, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.868972778320312, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.9066799879074097, + "num_tokens": 788909070.0, + "step": 20318 + }, + { + "epoch": 2.584785650680575, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.875746726989746, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.889299750328064, + "num_tokens": 788947069.0, + "step": 20319 + }, + { + "epoch": 2.5849128609591654, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.818999290466309, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.9061472415924072, + "num_tokens": 788984872.0, + "step": 20320 + }, + { + "epoch": 2.585040071237756, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.837994575500488, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9077049493789673, + "num_tokens": 789021573.0, + "step": 20321 + }, + { + "epoch": 2.5851672815163464, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.976492881774902, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.9003863334655762, + "num_tokens": 789059249.0, + "step": 20322 + }, + { + "epoch": 2.585294491794937, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.83194351196289, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9089682102203369, + "num_tokens": 789099053.0, + "step": 20323 + }, + { + "epoch": 2.5854217020735275, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.98864459991455, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8933080434799194, + "num_tokens": 789135497.0, + "step": 20324 + }, + { + "epoch": 2.585548912352118, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.846240997314453, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9057320952415466, + "num_tokens": 789174161.0, + "step": 20325 + }, + { + "epoch": 2.5856761226307086, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.93554401397705, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8977512121200562, + "num_tokens": 789212791.0, + "step": 20326 + }, + { + "epoch": 2.585803332909299, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.841866493225098, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8935245275497437, + "num_tokens": 789252703.0, + "step": 20327 + }, + { + "epoch": 2.5859305431878896, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.856014251708984, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8944770097732544, + "num_tokens": 789293152.0, + "step": 20328 + }, + { + "epoch": 2.58605775346648, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.899504661560059, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8875950574874878, + "num_tokens": 789338132.0, + "step": 20329 + }, + { + "epoch": 2.5861849637450707, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.03116512298584, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8798136115074158, + "num_tokens": 789371536.0, + "step": 20330 + }, + { + "epoch": 2.586312174023661, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.772470474243164, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.905367910861969, + "num_tokens": 789418861.0, + "step": 20331 + }, + { + "epoch": 2.5864393843022517, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.894742012023926, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8977872729301453, + "num_tokens": 789453961.0, + "step": 20332 + }, + { + "epoch": 2.5865665945808423, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.839685440063477, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.9041720628738403, + "num_tokens": 789489870.0, + "step": 20333 + }, + { + "epoch": 2.586693804859433, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.926570892333984, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8877685070037842, + "num_tokens": 789530992.0, + "step": 20334 + }, + { + "epoch": 2.5868210151380233, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.790543556213379, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.9006060361862183, + "num_tokens": 789572409.0, + "step": 20335 + }, + { + "epoch": 2.586948225416614, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.819036483764648, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.89375901222229, + "num_tokens": 789618617.0, + "step": 20336 + }, + { + "epoch": 2.587075435695204, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.903644561767578, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8980128765106201, + "num_tokens": 789656160.0, + "step": 20337 + }, + { + "epoch": 2.587202645973795, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.007719039916992, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8967525959014893, + "num_tokens": 789690917.0, + "step": 20338 + }, + { + "epoch": 2.587329856252385, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.762741088867188, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9072990417480469, + "num_tokens": 789735979.0, + "step": 20339 + }, + { + "epoch": 2.587457066530976, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.810652732849121, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8992440104484558, + "num_tokens": 789781579.0, + "step": 20340 + }, + { + "epoch": 2.587584276809566, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.940484046936035, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9110104441642761, + "num_tokens": 789813488.0, + "step": 20341 + }, + { + "epoch": 2.5877114870881566, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.910879135131836, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8964205980300903, + "num_tokens": 789847481.0, + "step": 20342 + }, + { + "epoch": 2.587838697366747, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.948567390441895, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8918224573135376, + "num_tokens": 789887982.0, + "step": 20343 + }, + { + "epoch": 2.5879659076453376, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.814221382141113, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9016997814178467, + "num_tokens": 789927475.0, + "step": 20344 + }, + { + "epoch": 2.588093117923928, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.804061889648438, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.9038392305374146, + "num_tokens": 789967827.0, + "step": 20345 + }, + { + "epoch": 2.5882203282025187, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.844778060913086, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8904813528060913, + "num_tokens": 790011308.0, + "step": 20346 + }, + { + "epoch": 2.588347538481109, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.844491004943848, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8875057697296143, + "num_tokens": 790053071.0, + "step": 20347 + }, + { + "epoch": 2.5884747487596997, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.947229385375977, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8897795081138611, + "num_tokens": 790095208.0, + "step": 20348 + }, + { + "epoch": 2.5886019590382903, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.836310386657715, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9172886610031128, + "num_tokens": 790131725.0, + "step": 20349 + }, + { + "epoch": 2.588729169316881, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.864031791687012, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.9023158550262451, + "num_tokens": 790172681.0, + "step": 20350 + }, + { + "epoch": 2.5888563795954713, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.906487464904785, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8880336284637451, + "num_tokens": 790218201.0, + "step": 20351 + }, + { + "epoch": 2.588983589874062, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.889121055603027, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8920854330062866, + "num_tokens": 790254619.0, + "step": 20352 + }, + { + "epoch": 2.5891108001526524, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.770745277404785, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9081689119338989, + "num_tokens": 790300108.0, + "step": 20353 + }, + { + "epoch": 2.589238010431243, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.91065788269043, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9052169322967529, + "num_tokens": 790337390.0, + "step": 20354 + }, + { + "epoch": 2.5893652207098334, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.014144897460938, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8892004489898682, + "num_tokens": 790376172.0, + "step": 20355 + }, + { + "epoch": 2.589492430988424, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.818912506103516, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.9028983116149902, + "num_tokens": 790415603.0, + "step": 20356 + }, + { + "epoch": 2.5896196412670145, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.867888450622559, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.9017273187637329, + "num_tokens": 790458404.0, + "step": 20357 + }, + { + "epoch": 2.589746851545605, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.96247386932373, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8884965181350708, + "num_tokens": 790500719.0, + "step": 20358 + }, + { + "epoch": 2.5898740618241956, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.964948654174805, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.9021919965744019, + "num_tokens": 790537052.0, + "step": 20359 + }, + { + "epoch": 2.5900012721027856, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.7872314453125, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.9028942584991455, + "num_tokens": 790578152.0, + "step": 20360 + }, + { + "epoch": 2.5901284823813766, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.906846046447754, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9020684957504272, + "num_tokens": 790617254.0, + "step": 20361 + }, + { + "epoch": 2.5902556926599667, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.954901695251465, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.886290967464447, + "num_tokens": 790657109.0, + "step": 20362 + }, + { + "epoch": 2.5903829029385577, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.0095796585083, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9126501679420471, + "num_tokens": 790693288.0, + "step": 20363 + }, + { + "epoch": 2.5905101132171477, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.833892822265625, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.9017983675003052, + "num_tokens": 790737350.0, + "step": 20364 + }, + { + "epoch": 2.5906373234957387, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.912525177001953, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8838070631027222, + "num_tokens": 790777944.0, + "step": 20365 + }, + { + "epoch": 2.590764533774329, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.83960247039795, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9025020003318787, + "num_tokens": 790821848.0, + "step": 20366 + }, + { + "epoch": 2.5908917440529193, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.926332473754883, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8863017559051514, + "num_tokens": 790857382.0, + "step": 20367 + }, + { + "epoch": 2.59101895433151, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.982394218444824, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8947896361351013, + "num_tokens": 790897364.0, + "step": 20368 + }, + { + "epoch": 2.5911461646101004, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.814192771911621, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.9045308232307434, + "num_tokens": 790936103.0, + "step": 20369 + }, + { + "epoch": 2.591273374888691, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.00067138671875, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.903276801109314, + "num_tokens": 790976322.0, + "step": 20370 + }, + { + "epoch": 2.5914005851672814, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.8850736618042, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8956971168518066, + "num_tokens": 791016167.0, + "step": 20371 + }, + { + "epoch": 2.591527795445872, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.857270240783691, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.907289445400238, + "num_tokens": 791054138.0, + "step": 20372 + }, + { + "epoch": 2.5916550057244625, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.75836181640625, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.9066729545593262, + "num_tokens": 791100239.0, + "step": 20373 + }, + { + "epoch": 2.591782216003053, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.839842796325684, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.9011030793190002, + "num_tokens": 791140110.0, + "step": 20374 + }, + { + "epoch": 2.5919094262816436, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.992406845092773, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8788939714431763, + "num_tokens": 791176473.0, + "step": 20375 + }, + { + "epoch": 2.592036636560234, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.987478256225586, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8886032104492188, + "num_tokens": 791208659.0, + "step": 20376 + }, + { + "epoch": 2.5921638468388246, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.006958961486816, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9118388295173645, + "num_tokens": 791247671.0, + "step": 20377 + }, + { + "epoch": 2.592291057117415, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.791274070739746, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9024274349212646, + "num_tokens": 791289372.0, + "step": 20378 + }, + { + "epoch": 2.5924182673960057, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.820008277893066, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.9034743309020996, + "num_tokens": 791328592.0, + "step": 20379 + }, + { + "epoch": 2.592545477674596, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.905191421508789, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.89015793800354, + "num_tokens": 791372319.0, + "step": 20380 + }, + { + "epoch": 2.5926726879531867, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.931838989257812, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8820065855979919, + "num_tokens": 791410656.0, + "step": 20381 + }, + { + "epoch": 2.5927998982317773, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.898706436157227, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8886663317680359, + "num_tokens": 791448519.0, + "step": 20382 + }, + { + "epoch": 2.592927108510368, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.857742309570312, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8965423703193665, + "num_tokens": 791486126.0, + "step": 20383 + }, + { + "epoch": 2.5930543187889583, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.920878410339355, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8970364332199097, + "num_tokens": 791521147.0, + "step": 20384 + }, + { + "epoch": 2.5931815290675484, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.879400253295898, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8976955413818359, + "num_tokens": 791564981.0, + "step": 20385 + }, + { + "epoch": 2.5933087393461394, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.907733917236328, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8975656032562256, + "num_tokens": 791603087.0, + "step": 20386 + }, + { + "epoch": 2.5934359496247295, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.835749626159668, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9133402109146118, + "num_tokens": 791644857.0, + "step": 20387 + }, + { + "epoch": 2.5935631599033204, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.84333610534668, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8984315395355225, + "num_tokens": 791690956.0, + "step": 20388 + }, + { + "epoch": 2.5936903701819105, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.901327133178711, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.9010272026062012, + "num_tokens": 791733784.0, + "step": 20389 + }, + { + "epoch": 2.5938175804605015, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.87417221069336, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9029322862625122, + "num_tokens": 791773968.0, + "step": 20390 + }, + { + "epoch": 2.5939447907390916, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.859085083007812, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.901087760925293, + "num_tokens": 791814351.0, + "step": 20391 + }, + { + "epoch": 2.594072001017682, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.884645462036133, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9090369343757629, + "num_tokens": 791850736.0, + "step": 20392 + }, + { + "epoch": 2.5941992112962726, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.917994499206543, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8951669931411743, + "num_tokens": 791891923.0, + "step": 20393 + }, + { + "epoch": 2.594326421574863, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.907608032226562, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8892452716827393, + "num_tokens": 791929154.0, + "step": 20394 + }, + { + "epoch": 2.5944536318534537, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 10.05955982208252, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8910979628562927, + "num_tokens": 791964833.0, + "step": 20395 + }, + { + "epoch": 2.594580842132044, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.773734092712402, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.903335452079773, + "num_tokens": 792007434.0, + "step": 20396 + }, + { + "epoch": 2.5947080524106347, + "ewc_loss": 0.02099609375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1861324310302734e-05, + "grad_norm": 9.877545356750488, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8985132575035095, + "num_tokens": 792044874.0, + "step": 20397 + }, + { + "epoch": 2.5948352626892253, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.904099464416504, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.9038722515106201, + "num_tokens": 792082711.0, + "step": 20398 + }, + { + "epoch": 2.594962472967816, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.938935279846191, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8879116177558899, + "num_tokens": 792114861.0, + "step": 20399 + }, + { + "epoch": 2.5950896832464063, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.892926216125488, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8856284022331238, + "num_tokens": 792155838.0, + "step": 20400 + }, + { + "epoch": 2.595216893524997, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.840799331665039, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8943954706192017, + "num_tokens": 792200208.0, + "step": 20401 + }, + { + "epoch": 2.5953441038035874, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.884414672851562, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.9028425216674805, + "num_tokens": 792236309.0, + "step": 20402 + }, + { + "epoch": 2.595471314082178, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.804503440856934, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9117137789726257, + "num_tokens": 792277345.0, + "step": 20403 + }, + { + "epoch": 2.5955985243607684, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.898566246032715, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8952571153640747, + "num_tokens": 792315809.0, + "step": 20404 + }, + { + "epoch": 2.595725734639359, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.974822044372559, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8874375820159912, + "num_tokens": 792352038.0, + "step": 20405 + }, + { + "epoch": 2.5958529449179495, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.970250129699707, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8980038166046143, + "num_tokens": 792382481.0, + "step": 20406 + }, + { + "epoch": 2.59598015519654, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.84228229522705, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.912834644317627, + "num_tokens": 792418206.0, + "step": 20407 + }, + { + "epoch": 2.5961073654751305, + "ewc_loss": 0.0211181640625, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.1920928955078125e-05, + "grad_norm": 9.756759643554688, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8985604643821716, + "num_tokens": 792462414.0, + "step": 20408 + }, + { + "epoch": 2.596234575753721, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 10.161321640014648, + "learning_rate": 1e-06, + "loss": 0.3933, + "mean_token_accuracy": 0.8661508560180664, + "num_tokens": 792494382.0, + "step": 20409 + }, + { + "epoch": 2.596361786032311, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.877118110656738, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8960555791854858, + "num_tokens": 792535509.0, + "step": 20410 + }, + { + "epoch": 2.596488996310902, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.804594993591309, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.892986536026001, + "num_tokens": 792577132.0, + "step": 20411 + }, + { + "epoch": 2.596616206589492, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.871236801147461, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9074983596801758, + "num_tokens": 792618734.0, + "step": 20412 + }, + { + "epoch": 2.596743416868083, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.92341136932373, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.9000792503356934, + "num_tokens": 792651679.0, + "step": 20413 + }, + { + "epoch": 2.5968706271466733, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.953128814697266, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8922373056411743, + "num_tokens": 792692355.0, + "step": 20414 + }, + { + "epoch": 2.596997837425264, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.812565803527832, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9078353643417358, + "num_tokens": 792734510.0, + "step": 20415 + }, + { + "epoch": 2.5971250477038543, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.976227760314941, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8894195556640625, + "num_tokens": 792775030.0, + "step": 20416 + }, + { + "epoch": 2.597252257982445, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.018489837646484, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.896308422088623, + "num_tokens": 792808650.0, + "step": 20417 + }, + { + "epoch": 2.5973794682610354, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.880444526672363, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.9008147120475769, + "num_tokens": 792843617.0, + "step": 20418 + }, + { + "epoch": 2.597506678539626, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.002372741699219, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8974287509918213, + "num_tokens": 792878211.0, + "step": 20419 + }, + { + "epoch": 2.5976338888182164, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.801740646362305, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8888950347900391, + "num_tokens": 792922831.0, + "step": 20420 + }, + { + "epoch": 2.597761099096807, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.851826667785645, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.9046783447265625, + "num_tokens": 792960056.0, + "step": 20421 + }, + { + "epoch": 2.5978883093753975, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.873970985412598, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9026724100112915, + "num_tokens": 792998946.0, + "step": 20422 + }, + { + "epoch": 2.598015519653988, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.904969215393066, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8989584445953369, + "num_tokens": 793037036.0, + "step": 20423 + }, + { + "epoch": 2.5981427299325786, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.936681747436523, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.898306667804718, + "num_tokens": 793075174.0, + "step": 20424 + }, + { + "epoch": 2.598269940211169, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.868677139282227, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.9016426801681519, + "num_tokens": 793114062.0, + "step": 20425 + }, + { + "epoch": 2.5983971504897596, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.91759967803955, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8961869478225708, + "num_tokens": 793154017.0, + "step": 20426 + }, + { + "epoch": 2.59852436076835, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.95478630065918, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8885992765426636, + "num_tokens": 793190400.0, + "step": 20427 + }, + { + "epoch": 2.5986515710469407, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.92390251159668, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8957836031913757, + "num_tokens": 793232020.0, + "step": 20428 + }, + { + "epoch": 2.598778781325531, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.768388748168945, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.9039379358291626, + "num_tokens": 793272253.0, + "step": 20429 + }, + { + "epoch": 2.5989059916041217, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.902321815490723, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8945106267929077, + "num_tokens": 793310353.0, + "step": 20430 + }, + { + "epoch": 2.5990332018827123, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.857614517211914, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8988763093948364, + "num_tokens": 793350171.0, + "step": 20431 + }, + { + "epoch": 2.599160412161303, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.860753059387207, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8986716866493225, + "num_tokens": 793392313.0, + "step": 20432 + }, + { + "epoch": 2.5992876224398933, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.853774070739746, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9078277945518494, + "num_tokens": 793430975.0, + "step": 20433 + }, + { + "epoch": 2.599414832718484, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.008737564086914, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8951531648635864, + "num_tokens": 793466090.0, + "step": 20434 + }, + { + "epoch": 2.599542042997074, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.877437591552734, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8867719769477844, + "num_tokens": 793509173.0, + "step": 20435 + }, + { + "epoch": 2.599669253275665, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.83998966217041, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8877346515655518, + "num_tokens": 793555602.0, + "step": 20436 + }, + { + "epoch": 2.599796463554255, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.998563766479492, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.889201283454895, + "num_tokens": 793598565.0, + "step": 20437 + }, + { + "epoch": 2.599923673832846, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.838651657104492, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9015639424324036, + "num_tokens": 793632640.0, + "step": 20438 + }, + { + "epoch": 2.600050884111436, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.202957153320312e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.876143455505371, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.9010075330734253, + "num_tokens": 793675267.0, + "step": 20439 + }, + { + "epoch": 2.6001780943900266, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.852664947509766, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.9010934233665466, + "num_tokens": 793709662.0, + "step": 20440 + }, + { + "epoch": 2.600305304668617, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.035135269165039, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8918160200119019, + "num_tokens": 793741571.0, + "step": 20441 + }, + { + "epoch": 2.6004325149472076, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.869202613830566, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8966200351715088, + "num_tokens": 793781871.0, + "step": 20442 + }, + { + "epoch": 2.600559725225798, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.76053524017334, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.912070631980896, + "num_tokens": 793823593.0, + "step": 20443 + }, + { + "epoch": 2.6006869355043887, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.88252067565918, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8976364135742188, + "num_tokens": 793866716.0, + "step": 20444 + }, + { + "epoch": 2.600814145782979, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.9063138961792, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.897057294845581, + "num_tokens": 793908032.0, + "step": 20445 + }, + { + "epoch": 2.6009413560615697, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.936615943908691, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8922229409217834, + "num_tokens": 793941126.0, + "step": 20446 + }, + { + "epoch": 2.6010685663401603, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.894186973571777, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.895857572555542, + "num_tokens": 793981965.0, + "step": 20447 + }, + { + "epoch": 2.601195776618751, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.847078323364258, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9083529710769653, + "num_tokens": 794019364.0, + "step": 20448 + }, + { + "epoch": 2.6013229868973413, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.928561210632324, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8963362574577332, + "num_tokens": 794057364.0, + "step": 20449 + }, + { + "epoch": 2.601450197175932, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.052278518676758, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8903260231018066, + "num_tokens": 794094727.0, + "step": 20450 + }, + { + "epoch": 2.6015774074545224, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.863165855407715, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8962618112564087, + "num_tokens": 794136336.0, + "step": 20451 + }, + { + "epoch": 2.601704617733113, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.916224479675293, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.9011213779449463, + "num_tokens": 794177710.0, + "step": 20452 + }, + { + "epoch": 2.6018318280117034, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.925092697143555, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8936147689819336, + "num_tokens": 794217795.0, + "step": 20453 + }, + { + "epoch": 2.601959038290294, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.858946800231934, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8975841403007507, + "num_tokens": 794254718.0, + "step": 20454 + }, + { + "epoch": 2.6020862485688845, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.889374732971191, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8890160322189331, + "num_tokens": 794295886.0, + "step": 20455 + }, + { + "epoch": 2.602213458847475, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.017992973327637, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.9017314910888672, + "num_tokens": 794336015.0, + "step": 20456 + }, + { + "epoch": 2.6023406691260655, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.88640022277832, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.897839367389679, + "num_tokens": 794372878.0, + "step": 20457 + }, + { + "epoch": 2.6024678794046556, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.949875831604004, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8996450901031494, + "num_tokens": 794408805.0, + "step": 20458 + }, + { + "epoch": 2.6025950896832466, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.018393516540527, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8934346437454224, + "num_tokens": 794443678.0, + "step": 20459 + }, + { + "epoch": 2.6027222999618367, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.788397789001465, + "learning_rate": 1e-06, + "loss": 0.2474, + "mean_token_accuracy": 0.913750410079956, + "num_tokens": 794484698.0, + "step": 20460 + }, + { + "epoch": 2.6028495102404277, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.8909273147583, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.9007223844528198, + "num_tokens": 794522553.0, + "step": 20461 + }, + { + "epoch": 2.6029767205190177, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.979303359985352, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8797256350517273, + "num_tokens": 794560885.0, + "step": 20462 + }, + { + "epoch": 2.6031039307976087, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.853957176208496, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8995972871780396, + "num_tokens": 794598226.0, + "step": 20463 + }, + { + "epoch": 2.603231141076199, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.839200973510742, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9070813655853271, + "num_tokens": 794640132.0, + "step": 20464 + }, + { + "epoch": 2.6033583513547893, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.794361114501953, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8895193934440613, + "num_tokens": 794687390.0, + "step": 20465 + }, + { + "epoch": 2.60348556163338, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.96739673614502, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8988491892814636, + "num_tokens": 794723783.0, + "step": 20466 + }, + { + "epoch": 2.6036127719119704, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.00823974609375, + "learning_rate": 1e-06, + "loss": 0.3597, + "mean_token_accuracy": 0.8786315321922302, + "num_tokens": 794763122.0, + "step": 20467 + }, + { + "epoch": 2.603739982190561, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.95374584197998, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8838144540786743, + "num_tokens": 794804505.0, + "step": 20468 + }, + { + "epoch": 2.6038671924691514, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.89318561553955, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8937749862670898, + "num_tokens": 794846422.0, + "step": 20469 + }, + { + "epoch": 2.603994402747742, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.860482215881348, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.893044650554657, + "num_tokens": 794890729.0, + "step": 20470 + }, + { + "epoch": 2.6041216130263325, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.076742172241211, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8886709213256836, + "num_tokens": 794924390.0, + "step": 20471 + }, + { + "epoch": 2.604248823304923, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.888655662536621, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.9032278060913086, + "num_tokens": 794961149.0, + "step": 20472 + }, + { + "epoch": 2.6043760335835135, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.946562767028809, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8985540866851807, + "num_tokens": 794995932.0, + "step": 20473 + }, + { + "epoch": 2.604503243862104, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.942662239074707, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9068030118942261, + "num_tokens": 795027683.0, + "step": 20474 + }, + { + "epoch": 2.6046304541406946, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.883768081665039, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8987330198287964, + "num_tokens": 795066303.0, + "step": 20475 + }, + { + "epoch": 2.604757664419285, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.9998779296875, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.9002894759178162, + "num_tokens": 795103595.0, + "step": 20476 + }, + { + "epoch": 2.6048848746978757, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.893817901611328, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.907333493232727, + "num_tokens": 795144402.0, + "step": 20477 + }, + { + "epoch": 2.605012084976466, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.975093841552734, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8985172510147095, + "num_tokens": 795183364.0, + "step": 20478 + }, + { + "epoch": 2.6051392952550567, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.870086669921875, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8892867565155029, + "num_tokens": 795228103.0, + "step": 20479 + }, + { + "epoch": 2.6052665055336472, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.881611824035645, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8914897441864014, + "num_tokens": 795268247.0, + "step": 20480 + }, + { + "epoch": 2.6053937158122378, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.838995933532715, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.897599458694458, + "num_tokens": 795310995.0, + "step": 20481 + }, + { + "epoch": 2.6055209260908283, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.922451972961426, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.9016863703727722, + "num_tokens": 795349108.0, + "step": 20482 + }, + { + "epoch": 2.6056481363694184, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.978684425354004, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9004798531532288, + "num_tokens": 795384704.0, + "step": 20483 + }, + { + "epoch": 2.6057753466480094, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.870136260986328, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9002795815467834, + "num_tokens": 795422493.0, + "step": 20484 + }, + { + "epoch": 2.6059025569265994, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.989282608032227, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8867098689079285, + "num_tokens": 795459518.0, + "step": 20485 + }, + { + "epoch": 2.6060297672051904, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.812292098999023, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8980254530906677, + "num_tokens": 795503719.0, + "step": 20486 + }, + { + "epoch": 2.6061569774837805, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.94112491607666, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9051580429077148, + "num_tokens": 795538592.0, + "step": 20487 + }, + { + "epoch": 2.6062841877623715, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.902268409729004, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8891632556915283, + "num_tokens": 795582811.0, + "step": 20488 + }, + { + "epoch": 2.6064113980409616, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.83230209350586, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.9014990329742432, + "num_tokens": 795624794.0, + "step": 20489 + }, + { + "epoch": 2.606538608319552, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.941329002380371, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.9011859893798828, + "num_tokens": 795663390.0, + "step": 20490 + }, + { + "epoch": 2.6066658185981426, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.881570816040039, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8988063335418701, + "num_tokens": 795700348.0, + "step": 20491 + }, + { + "epoch": 2.606793028876733, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.032748222351074, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.9035919904708862, + "num_tokens": 795734258.0, + "step": 20492 + }, + { + "epoch": 2.6069202391553237, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.79455852508545, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.902289628982544, + "num_tokens": 795779454.0, + "step": 20493 + }, + { + "epoch": 2.607047449433914, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.893169403076172, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9122580885887146, + "num_tokens": 795818171.0, + "step": 20494 + }, + { + "epoch": 2.6071746597125047, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.943160057067871, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8978297114372253, + "num_tokens": 795857060.0, + "step": 20495 + }, + { + "epoch": 2.6073018699910953, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.956294059753418, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8942005634307861, + "num_tokens": 795895824.0, + "step": 20496 + }, + { + "epoch": 2.607429080269686, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.869622230529785, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8943529725074768, + "num_tokens": 795939745.0, + "step": 20497 + }, + { + "epoch": 2.6075562905482763, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.936676979064941, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9021189212799072, + "num_tokens": 795978637.0, + "step": 20498 + }, + { + "epoch": 2.607683500826867, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.891119003295898, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.9081858992576599, + "num_tokens": 796017823.0, + "step": 20499 + }, + { + "epoch": 2.6078107111054574, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.077627182006836, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8890954256057739, + "num_tokens": 796051969.0, + "step": 20500 + }, + { + "epoch": 2.607937921384048, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.860721588134766, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8998663425445557, + "num_tokens": 796088466.0, + "step": 20501 + }, + { + "epoch": 2.6080651316626384, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.948905944824219, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8933142423629761, + "num_tokens": 796132206.0, + "step": 20502 + }, + { + "epoch": 2.608192341941229, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.804831504821777, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.9045521020889282, + "num_tokens": 796179121.0, + "step": 20503 + }, + { + "epoch": 2.6083195522198195, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.877692222595215, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9037044048309326, + "num_tokens": 796220092.0, + "step": 20504 + }, + { + "epoch": 2.60844676249841, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.910435676574707, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8979838490486145, + "num_tokens": 796257611.0, + "step": 20505 + }, + { + "epoch": 2.6085739727770005, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.879859924316406, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8931716680526733, + "num_tokens": 796297888.0, + "step": 20506 + }, + { + "epoch": 2.608701183055591, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.920768737792969, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8999652862548828, + "num_tokens": 796334629.0, + "step": 20507 + }, + { + "epoch": 2.608828393334181, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.03174114227295, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8967132568359375, + "num_tokens": 796368660.0, + "step": 20508 + }, + { + "epoch": 2.608955603612772, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.876386642456055, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.900171160697937, + "num_tokens": 796407480.0, + "step": 20509 + }, + { + "epoch": 2.609082813891362, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.922835350036621, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.9029068946838379, + "num_tokens": 796443555.0, + "step": 20510 + }, + { + "epoch": 2.609210024169953, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.93605899810791, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.898332953453064, + "num_tokens": 796477012.0, + "step": 20511 + }, + { + "epoch": 2.6093372344485433, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.855772018432617, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8987741470336914, + "num_tokens": 796515496.0, + "step": 20512 + }, + { + "epoch": 2.609464444727134, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.901457786560059, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8915903568267822, + "num_tokens": 796554369.0, + "step": 20513 + }, + { + "epoch": 2.6095916550057243, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.970884323120117, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8884007334709167, + "num_tokens": 796591327.0, + "step": 20514 + }, + { + "epoch": 2.609718865284315, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.83533763885498, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.906745195388794, + "num_tokens": 796632625.0, + "step": 20515 + }, + { + "epoch": 2.6098460755629054, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.905351638793945, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9044268727302551, + "num_tokens": 796667966.0, + "step": 20516 + }, + { + "epoch": 2.609973285841496, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.94729995727539, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9150253534317017, + "num_tokens": 796702588.0, + "step": 20517 + }, + { + "epoch": 2.6101004961200864, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.977445602416992, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8976420164108276, + "num_tokens": 796738982.0, + "step": 20518 + }, + { + "epoch": 2.610227706398677, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.899435043334961, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.9078738689422607, + "num_tokens": 796773948.0, + "step": 20519 + }, + { + "epoch": 2.6103549166772675, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.943968772888184, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.9030482769012451, + "num_tokens": 796809256.0, + "step": 20520 + }, + { + "epoch": 2.610482126955858, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.922614097595215, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9052057862281799, + "num_tokens": 796849494.0, + "step": 20521 + }, + { + "epoch": 2.6106093372344485, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.862051963806152, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.9066015481948853, + "num_tokens": 796891638.0, + "step": 20522 + }, + { + "epoch": 2.610736547513039, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.250640869140625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.9252347946167, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8878509998321533, + "num_tokens": 796930674.0, + "step": 20523 + }, + { + "epoch": 2.6108637577916296, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.909612655639648, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.9015518426895142, + "num_tokens": 796969855.0, + "step": 20524 + }, + { + "epoch": 2.61099096807022, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.90351676940918, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8839739561080933, + "num_tokens": 797009834.0, + "step": 20525 + }, + { + "epoch": 2.6111181783488107, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.951225280761719, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8929239511489868, + "num_tokens": 797045756.0, + "step": 20526 + }, + { + "epoch": 2.611245388627401, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.905234336853027, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8901368975639343, + "num_tokens": 797090951.0, + "step": 20527 + }, + { + "epoch": 2.6113725989059917, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.835591316223145, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.894768238067627, + "num_tokens": 797135880.0, + "step": 20528 + }, + { + "epoch": 2.6114998091845822, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.852499961853027, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.9067033529281616, + "num_tokens": 797173831.0, + "step": 20529 + }, + { + "epoch": 2.6116270194631728, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.821497917175293, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8912909030914307, + "num_tokens": 797217715.0, + "step": 20530 + }, + { + "epoch": 2.6117542297417633, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.009965896606445, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8965486884117126, + "num_tokens": 797254793.0, + "step": 20531 + }, + { + "epoch": 2.611881440020354, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.950496673583984, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.9015653133392334, + "num_tokens": 797291490.0, + "step": 20532 + }, + { + "epoch": 2.612008650298944, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.895162582397461, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9068428874015808, + "num_tokens": 797327533.0, + "step": 20533 + }, + { + "epoch": 2.612135860577535, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.904692649841309, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8990577459335327, + "num_tokens": 797371575.0, + "step": 20534 + }, + { + "epoch": 2.612263070856125, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.903585433959961, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9023878574371338, + "num_tokens": 797405378.0, + "step": 20535 + }, + { + "epoch": 2.612390281134716, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.924962043762207, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8915290832519531, + "num_tokens": 797443395.0, + "step": 20536 + }, + { + "epoch": 2.612517491413306, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.73452377319336, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8993435502052307, + "num_tokens": 797488226.0, + "step": 20537 + }, + { + "epoch": 2.6126447016918966, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.860923767089844, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.9048120975494385, + "num_tokens": 797530607.0, + "step": 20538 + }, + { + "epoch": 2.612771911970487, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.920632362365723, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9091877937316895, + "num_tokens": 797564063.0, + "step": 20539 + }, + { + "epoch": 2.6128991222490776, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.830549240112305, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.9022505283355713, + "num_tokens": 797602964.0, + "step": 20540 + }, + { + "epoch": 2.613026332527668, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.900298118591309, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8995494842529297, + "num_tokens": 797643528.0, + "step": 20541 + }, + { + "epoch": 2.6131535428062587, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.8952054977417, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.9001584053039551, + "num_tokens": 797679526.0, + "step": 20542 + }, + { + "epoch": 2.613280753084849, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.833822250366211, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8921948671340942, + "num_tokens": 797721089.0, + "step": 20543 + }, + { + "epoch": 2.6134079633634397, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.915926933288574, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8925046324729919, + "num_tokens": 797762814.0, + "step": 20544 + }, + { + "epoch": 2.6135351736420303, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.842450141906738, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9124540090560913, + "num_tokens": 797801622.0, + "step": 20545 + }, + { + "epoch": 2.613662383920621, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.985604286193848, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8839102983474731, + "num_tokens": 797839010.0, + "step": 20546 + }, + { + "epoch": 2.6137895941992113, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.80895709991455, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9119316339492798, + "num_tokens": 797875894.0, + "step": 20547 + }, + { + "epoch": 2.613916804477802, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.909261703491211, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.9015757441520691, + "num_tokens": 797921336.0, + "step": 20548 + }, + { + "epoch": 2.6140440147563924, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.891690254211426, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8918960690498352, + "num_tokens": 797960546.0, + "step": 20549 + }, + { + "epoch": 2.614171225034983, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.901786804199219, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8965793251991272, + "num_tokens": 797995138.0, + "step": 20550 + }, + { + "epoch": 2.6142984353135734, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.943135261535645, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9073854684829712, + "num_tokens": 798030740.0, + "step": 20551 + }, + { + "epoch": 2.614425645592164, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.786023139953613, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9015100002288818, + "num_tokens": 798076832.0, + "step": 20552 + }, + { + "epoch": 2.6145528558707545, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.957303047180176, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.9004207849502563, + "num_tokens": 798114719.0, + "step": 20553 + }, + { + "epoch": 2.614680066149345, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.841926574707031, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9013398885726929, + "num_tokens": 798159341.0, + "step": 20554 + }, + { + "epoch": 2.6148072764279355, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.864784240722656, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9094754457473755, + "num_tokens": 798195290.0, + "step": 20555 + }, + { + "epoch": 2.6149344867065256, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.95950984954834, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8989894390106201, + "num_tokens": 798235711.0, + "step": 20556 + }, + { + "epoch": 2.6150616969851166, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.872172355651855, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8905054926872253, + "num_tokens": 798281528.0, + "step": 20557 + }, + { + "epoch": 2.6151889072637067, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.92076301574707, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.912190318107605, + "num_tokens": 798319244.0, + "step": 20558 + }, + { + "epoch": 2.6153161175422976, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.875679969787598, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9029852151870728, + "num_tokens": 798359364.0, + "step": 20559 + }, + { + "epoch": 2.6154433278208877, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.93129825592041, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8792237043380737, + "num_tokens": 798402260.0, + "step": 20560 + }, + { + "epoch": 2.6155705380994787, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.881391525268555, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.9021161794662476, + "num_tokens": 798440847.0, + "step": 20561 + }, + { + "epoch": 2.615697748378069, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.868083953857422, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8998605012893677, + "num_tokens": 798482109.0, + "step": 20562 + }, + { + "epoch": 2.6158249586566593, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.895389556884766, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9022319912910461, + "num_tokens": 798519316.0, + "step": 20563 + }, + { + "epoch": 2.61595216893525, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.907894134521484, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8962323665618896, + "num_tokens": 798561089.0, + "step": 20564 + }, + { + "epoch": 2.6160793792138404, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.818546295166016, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8916642665863037, + "num_tokens": 798602989.0, + "step": 20565 + }, + { + "epoch": 2.616206589492431, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.936477661132812, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.889478862285614, + "num_tokens": 798642664.0, + "step": 20566 + }, + { + "epoch": 2.6163337997710214, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.935012817382812, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8999582529067993, + "num_tokens": 798682695.0, + "step": 20567 + }, + { + "epoch": 2.616461010049612, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.935331344604492, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9045862555503845, + "num_tokens": 798718990.0, + "step": 20568 + }, + { + "epoch": 2.6165882203282025, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.979783058166504, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.892671525478363, + "num_tokens": 798757910.0, + "step": 20569 + }, + { + "epoch": 2.616715430606793, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 10.005668640136719, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.9025159478187561, + "num_tokens": 798791107.0, + "step": 20570 + }, + { + "epoch": 2.6168426408853835, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.99046516418457, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.880959153175354, + "num_tokens": 798832467.0, + "step": 20571 + }, + { + "epoch": 2.616969851163974, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.93363094329834, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8902225494384766, + "num_tokens": 798869960.0, + "step": 20572 + }, + { + "epoch": 2.6170970614425646, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.926287651062012, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8946921825408936, + "num_tokens": 798909593.0, + "step": 20573 + }, + { + "epoch": 2.617224271721155, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.923656463623047, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9096814393997192, + "num_tokens": 798945012.0, + "step": 20574 + }, + { + "epoch": 2.6173514819997457, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.900989532470703, + "learning_rate": 1e-06, + "loss": 0.247, + "mean_token_accuracy": 0.9172540903091431, + "num_tokens": 798979735.0, + "step": 20575 + }, + { + "epoch": 2.617478692278336, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.874341011047363, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.9002832174301147, + "num_tokens": 799022170.0, + "step": 20576 + }, + { + "epoch": 2.6176059025569267, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.897273063659668, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8881314396858215, + "num_tokens": 799060431.0, + "step": 20577 + }, + { + "epoch": 2.6177331128355172, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.87650203704834, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8911634683609009, + "num_tokens": 799099937.0, + "step": 20578 + }, + { + "epoch": 2.6178603231141078, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.927010536193848, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8896598815917969, + "num_tokens": 799142562.0, + "step": 20579 + }, + { + "epoch": 2.6179875333926983, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.005352973937988, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8841885924339294, + "num_tokens": 799178895.0, + "step": 20580 + }, + { + "epoch": 2.6181147436712884, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.814075469970703, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8942781090736389, + "num_tokens": 799225817.0, + "step": 20581 + }, + { + "epoch": 2.6182419539498794, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.879531860351562, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8984643816947937, + "num_tokens": 799266362.0, + "step": 20582 + }, + { + "epoch": 2.6183691642284694, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.958104133605957, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8934353590011597, + "num_tokens": 799306761.0, + "step": 20583 + }, + { + "epoch": 2.6184963745070604, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.9412260055542, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9090745449066162, + "num_tokens": 799338118.0, + "step": 20584 + }, + { + "epoch": 2.6186235847856505, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.016179084777832, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.884263277053833, + "num_tokens": 799370782.0, + "step": 20585 + }, + { + "epoch": 2.6187507950642415, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.84575080871582, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.9017884731292725, + "num_tokens": 799413872.0, + "step": 20586 + }, + { + "epoch": 2.6188780053428315, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.907530784606934, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8940466046333313, + "num_tokens": 799456123.0, + "step": 20587 + }, + { + "epoch": 2.619005215621422, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.846823692321777, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8984194993972778, + "num_tokens": 799500346.0, + "step": 20588 + }, + { + "epoch": 2.6191324259000126, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.899044036865234, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9033573865890503, + "num_tokens": 799537096.0, + "step": 20589 + }, + { + "epoch": 2.619259636178603, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.931235313415527, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9088460803031921, + "num_tokens": 799575282.0, + "step": 20590 + }, + { + "epoch": 2.6193868464571937, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.971536636352539, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.897346019744873, + "num_tokens": 799613050.0, + "step": 20591 + }, + { + "epoch": 2.619514056735784, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.864381790161133, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.9021322727203369, + "num_tokens": 799648518.0, + "step": 20592 + }, + { + "epoch": 2.6196412670143747, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.877583503723145, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.9184695482254028, + "num_tokens": 799685565.0, + "step": 20593 + }, + { + "epoch": 2.6197684772929652, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.88581371307373, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8941761255264282, + "num_tokens": 799723954.0, + "step": 20594 + }, + { + "epoch": 2.6198956875715558, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.939281463623047, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8951350450515747, + "num_tokens": 799764598.0, + "step": 20595 + }, + { + "epoch": 2.6200228978501463, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.934453964233398, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8920033574104309, + "num_tokens": 799803112.0, + "step": 20596 + }, + { + "epoch": 2.620150108128737, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.79763126373291, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9025537967681885, + "num_tokens": 799843031.0, + "step": 20597 + }, + { + "epoch": 2.6202773184073274, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.010592460632324, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8912476301193237, + "num_tokens": 799878316.0, + "step": 20598 + }, + { + "epoch": 2.620404528685918, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.981382369995117, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8935483694076538, + "num_tokens": 799916811.0, + "step": 20599 + }, + { + "epoch": 2.6205317389645084, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.114973068237305, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8805449604988098, + "num_tokens": 799952259.0, + "step": 20600 + }, + { + "epoch": 2.620658949243099, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.921783447265625, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.9088545441627502, + "num_tokens": 799988646.0, + "step": 20601 + }, + { + "epoch": 2.6207861595216895, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.825287818908691, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8847624063491821, + "num_tokens": 800036223.0, + "step": 20602 + }, + { + "epoch": 2.62091336980028, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.952418327331543, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9081164598464966, + "num_tokens": 800071580.0, + "step": 20603 + }, + { + "epoch": 2.6210405800788705, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.950769424438477, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.9013322591781616, + "num_tokens": 800108205.0, + "step": 20604 + }, + { + "epoch": 2.621167790357461, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.967671394348145, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8999907970428467, + "num_tokens": 800143885.0, + "step": 20605 + }, + { + "epoch": 2.621295000636051, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.91711139678955, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9047466516494751, + "num_tokens": 800178043.0, + "step": 20606 + }, + { + "epoch": 2.621422210914642, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.825488090515137, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8991841673851013, + "num_tokens": 800218215.0, + "step": 20607 + }, + { + "epoch": 2.621549421193232, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.936976432800293, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.900626540184021, + "num_tokens": 800259501.0, + "step": 20608 + }, + { + "epoch": 2.621676631471823, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.93108081817627, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8926831483840942, + "num_tokens": 800298259.0, + "step": 20609 + }, + { + "epoch": 2.6218038417504133, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.917341232299805, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8872720003128052, + "num_tokens": 800336683.0, + "step": 20610 + }, + { + "epoch": 2.621931052029004, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.925297737121582, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8996579647064209, + "num_tokens": 800378268.0, + "step": 20611 + }, + { + "epoch": 2.6220582623075943, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.948134422302246, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8958274126052856, + "num_tokens": 800420121.0, + "step": 20612 + }, + { + "epoch": 2.622185472586185, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.88814926147461, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8933916091918945, + "num_tokens": 800464331.0, + "step": 20613 + }, + { + "epoch": 2.6223126828647754, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.948637008666992, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8994753956794739, + "num_tokens": 800499778.0, + "step": 20614 + }, + { + "epoch": 2.622439893143366, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.961677551269531, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8896644115447998, + "num_tokens": 800536484.0, + "step": 20615 + }, + { + "epoch": 2.6225671034219564, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.979559898376465, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9087438583374023, + "num_tokens": 800573471.0, + "step": 20616 + }, + { + "epoch": 2.622694313700547, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.933209419250488, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.902703046798706, + "num_tokens": 800612172.0, + "step": 20617 + }, + { + "epoch": 2.6228215239791375, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.004225730895996, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8982348442077637, + "num_tokens": 800647267.0, + "step": 20618 + }, + { + "epoch": 2.622948734257728, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.846068382263184, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8995656967163086, + "num_tokens": 800693293.0, + "step": 20619 + }, + { + "epoch": 2.6230759445363185, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.880938529968262, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9086427688598633, + "num_tokens": 800728631.0, + "step": 20620 + }, + { + "epoch": 2.623203154814909, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.983071327209473, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.9001258611679077, + "num_tokens": 800763620.0, + "step": 20621 + }, + { + "epoch": 2.6233303650934996, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.900799751281738, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.898784875869751, + "num_tokens": 800799885.0, + "step": 20622 + }, + { + "epoch": 2.62345757537209, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.984288215637207, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.890954852104187, + "num_tokens": 800840709.0, + "step": 20623 + }, + { + "epoch": 2.6235847856506807, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.87020492553711, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8946259021759033, + "num_tokens": 800885261.0, + "step": 20624 + }, + { + "epoch": 2.623711995929271, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.000736236572266, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8951374292373657, + "num_tokens": 800923812.0, + "step": 20625 + }, + { + "epoch": 2.6238392062078617, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.951770782470703, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8979865312576294, + "num_tokens": 800959058.0, + "step": 20626 + }, + { + "epoch": 2.6239664164864522, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.83708381652832, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.9007461667060852, + "num_tokens": 800996212.0, + "step": 20627 + }, + { + "epoch": 2.6240936267650428, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.902619361877441, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9100465774536133, + "num_tokens": 801037190.0, + "step": 20628 + }, + { + "epoch": 2.6242208370436333, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.852889060974121, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9089001417160034, + "num_tokens": 801077891.0, + "step": 20629 + }, + { + "epoch": 2.624348047322224, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.95505142211914, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9086118936538696, + "num_tokens": 801111879.0, + "step": 20630 + }, + { + "epoch": 2.624475257600814, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.997957229614258, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9115369915962219, + "num_tokens": 801142199.0, + "step": 20631 + }, + { + "epoch": 2.624602467879405, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.904586791992188, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8960476517677307, + "num_tokens": 801183479.0, + "step": 20632 + }, + { + "epoch": 2.624729678157995, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.900837898254395, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8914556503295898, + "num_tokens": 801224544.0, + "step": 20633 + }, + { + "epoch": 2.624856888436586, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.854714393615723, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.9039468765258789, + "num_tokens": 801262035.0, + "step": 20634 + }, + { + "epoch": 2.624984098715176, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.875614166259766, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.9011020660400391, + "num_tokens": 801298805.0, + "step": 20635 + }, + { + "epoch": 2.6251113089937665, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.865623474121094, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9067375659942627, + "num_tokens": 801340383.0, + "step": 20636 + }, + { + "epoch": 2.625238519272357, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.847899436950684, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.9010472893714905, + "num_tokens": 801381046.0, + "step": 20637 + }, + { + "epoch": 2.6253657295509476, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.900982856750488, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.907162606716156, + "num_tokens": 801420207.0, + "step": 20638 + }, + { + "epoch": 2.625492939829538, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 10.000267028808594, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8916955590248108, + "num_tokens": 801454003.0, + "step": 20639 + }, + { + "epoch": 2.6256201501081287, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.836536407470703, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9021162986755371, + "num_tokens": 801498954.0, + "step": 20640 + }, + { + "epoch": 2.625747360386719, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.921051979064941, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8892999291419983, + "num_tokens": 801537691.0, + "step": 20641 + }, + { + "epoch": 2.6258745706653097, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.075026512145996, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8908920288085938, + "num_tokens": 801574104.0, + "step": 20642 + }, + { + "epoch": 2.6260017809439002, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.772493362426758, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.907124400138855, + "num_tokens": 801617451.0, + "step": 20643 + }, + { + "epoch": 2.6261289912224908, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.919290542602539, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8890106678009033, + "num_tokens": 801660577.0, + "step": 20644 + }, + { + "epoch": 2.6262562015010813, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.029973983764648, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8916962742805481, + "num_tokens": 801698410.0, + "step": 20645 + }, + { + "epoch": 2.626383411779672, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.939448356628418, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8911133408546448, + "num_tokens": 801735397.0, + "step": 20646 + }, + { + "epoch": 2.6265106220582624, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.860645294189453, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8936848640441895, + "num_tokens": 801776735.0, + "step": 20647 + }, + { + "epoch": 2.626637832336853, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.972899436950684, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.9026864767074585, + "num_tokens": 801814903.0, + "step": 20648 + }, + { + "epoch": 2.6267650426154434, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.86161994934082, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8943841457366943, + "num_tokens": 801855714.0, + "step": 20649 + }, + { + "epoch": 2.626892252894034, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.838783264160156, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.9010198712348938, + "num_tokens": 801896598.0, + "step": 20650 + }, + { + "epoch": 2.6270194631726245, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.924899101257324, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9080843925476074, + "num_tokens": 801932598.0, + "step": 20651 + }, + { + "epoch": 2.627146673451215, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.964313507080078, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.875780463218689, + "num_tokens": 801974037.0, + "step": 20652 + }, + { + "epoch": 2.6272738837298055, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.861130714416504, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8929352164268494, + "num_tokens": 802020436.0, + "step": 20653 + }, + { + "epoch": 2.6274010940083956, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.969657897949219, + "learning_rate": 1e-06, + "loss": 0.2629, + "mean_token_accuracy": 0.9091996550559998, + "num_tokens": 802048421.0, + "step": 20654 + }, + { + "epoch": 2.6275283042869866, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.924784660339355, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8976103067398071, + "num_tokens": 802086359.0, + "step": 20655 + }, + { + "epoch": 2.6276555145655767, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.886382102966309, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.9031307697296143, + "num_tokens": 802121800.0, + "step": 20656 + }, + { + "epoch": 2.6277827248441676, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.944703102111816, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8896523118019104, + "num_tokens": 802162563.0, + "step": 20657 + }, + { + "epoch": 2.6279099351227577, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.013640403747559, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.9028828740119934, + "num_tokens": 802198301.0, + "step": 20658 + }, + { + "epoch": 2.6280371454013487, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.901249885559082, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8915382027626038, + "num_tokens": 802234274.0, + "step": 20659 + }, + { + "epoch": 2.628164355679939, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.189506530761719, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8809006214141846, + "num_tokens": 802263131.0, + "step": 20660 + }, + { + "epoch": 2.6282915659585293, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.9624662399292, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.896490216255188, + "num_tokens": 802298432.0, + "step": 20661 + }, + { + "epoch": 2.62841877623712, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.876596450805664, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9022133350372314, + "num_tokens": 802337073.0, + "step": 20662 + }, + { + "epoch": 2.6285459865157104, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.954875946044922, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8933461904525757, + "num_tokens": 802376535.0, + "step": 20663 + }, + { + "epoch": 2.628673196794301, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.824482917785645, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8966101408004761, + "num_tokens": 802415825.0, + "step": 20664 + }, + { + "epoch": 2.6288004070728914, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.95065975189209, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.9062811136245728, + "num_tokens": 802450048.0, + "step": 20665 + }, + { + "epoch": 2.628927617351482, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.111797332763672, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8962539434432983, + "num_tokens": 802481386.0, + "step": 20666 + }, + { + "epoch": 2.6290548276300725, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.935321807861328, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9019379615783691, + "num_tokens": 802519159.0, + "step": 20667 + }, + { + "epoch": 2.629182037908663, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.810504913330078, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9071868658065796, + "num_tokens": 802557961.0, + "step": 20668 + }, + { + "epoch": 2.6293092481872535, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.923816680908203, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.9005409479141235, + "num_tokens": 802599493.0, + "step": 20669 + }, + { + "epoch": 2.629436458465844, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.960081100463867, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8956665992736816, + "num_tokens": 802634892.0, + "step": 20670 + }, + { + "epoch": 2.6295636687444346, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.010336875915527, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.9008170366287231, + "num_tokens": 802670540.0, + "step": 20671 + }, + { + "epoch": 2.629690879023025, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.883636474609375, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8838738203048706, + "num_tokens": 802712834.0, + "step": 20672 + }, + { + "epoch": 2.6298180893016156, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.9246244430542, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.9038516879081726, + "num_tokens": 802751692.0, + "step": 20673 + }, + { + "epoch": 2.629945299580206, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.887182235717773, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.904527485370636, + "num_tokens": 802788103.0, + "step": 20674 + }, + { + "epoch": 2.6300725098587967, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.96436882019043, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8981943130493164, + "num_tokens": 802825114.0, + "step": 20675 + }, + { + "epoch": 2.6301997201373872, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.860916137695312, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.9039353728294373, + "num_tokens": 802864467.0, + "step": 20676 + }, + { + "epoch": 2.6303269304159778, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.83825969696045, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.9008606672286987, + "num_tokens": 802906714.0, + "step": 20677 + }, + { + "epoch": 2.6304541406945683, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.884424209594727, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.89803147315979, + "num_tokens": 802947606.0, + "step": 20678 + }, + { + "epoch": 2.6305813509731584, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.914216041564941, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.9018267393112183, + "num_tokens": 802987601.0, + "step": 20679 + }, + { + "epoch": 2.6307085612517493, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.847709655761719, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9111689329147339, + "num_tokens": 803024375.0, + "step": 20680 + }, + { + "epoch": 2.6308357715303394, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.97790241241455, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9090873003005981, + "num_tokens": 803062409.0, + "step": 20681 + }, + { + "epoch": 2.6309629818089304, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.921515464782715, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8876770734786987, + "num_tokens": 803099445.0, + "step": 20682 + }, + { + "epoch": 2.6310901920875205, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.81519603729248, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8950439095497131, + "num_tokens": 803153582.0, + "step": 20683 + }, + { + "epoch": 2.6312174023661115, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.791064262390137, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.894049882888794, + "num_tokens": 803193690.0, + "step": 20684 + }, + { + "epoch": 2.6313446126447015, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.833634376525879, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.899959921836853, + "num_tokens": 803239916.0, + "step": 20685 + }, + { + "epoch": 2.631471822923292, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.000287055969238, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8934473395347595, + "num_tokens": 803276918.0, + "step": 20686 + }, + { + "epoch": 2.6315990332018826, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.901755332946777, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8956379890441895, + "num_tokens": 803321288.0, + "step": 20687 + }, + { + "epoch": 2.631726243480473, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.906736373901367, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8962312936782837, + "num_tokens": 803358885.0, + "step": 20688 + }, + { + "epoch": 2.6318534537590637, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.035109519958496, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8938556909561157, + "num_tokens": 803395503.0, + "step": 20689 + }, + { + "epoch": 2.631980664037654, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.816786766052246, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8932015299797058, + "num_tokens": 803435427.0, + "step": 20690 + }, + { + "epoch": 2.6321078743162447, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.902047157287598, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8934142589569092, + "num_tokens": 803477000.0, + "step": 20691 + }, + { + "epoch": 2.6322350845948352, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.897327423095703, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8953578472137451, + "num_tokens": 803516501.0, + "step": 20692 + }, + { + "epoch": 2.6323622948734258, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.846214294433594, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8903217911720276, + "num_tokens": 803560629.0, + "step": 20693 + }, + { + "epoch": 2.6324895051520163, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.94640064239502, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8972395062446594, + "num_tokens": 803601286.0, + "step": 20694 + }, + { + "epoch": 2.632616715430607, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.811525344848633, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.908467173576355, + "num_tokens": 803639093.0, + "step": 20695 + }, + { + "epoch": 2.6327439257091974, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.966188430786133, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8881697654724121, + "num_tokens": 803678898.0, + "step": 20696 + }, + { + "epoch": 2.632871135987788, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.059072494506836, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8878331184387207, + "num_tokens": 803713998.0, + "step": 20697 + }, + { + "epoch": 2.6329983462663784, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.809316635131836, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8965021371841431, + "num_tokens": 803756359.0, + "step": 20698 + }, + { + "epoch": 2.633125556544969, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.941217422485352, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8990960121154785, + "num_tokens": 803790421.0, + "step": 20699 + }, + { + "epoch": 2.6332527668235595, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.905844688415527, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8968868255615234, + "num_tokens": 803831375.0, + "step": 20700 + }, + { + "epoch": 2.63337997710215, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.892438888549805, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8986014723777771, + "num_tokens": 803877142.0, + "step": 20701 + }, + { + "epoch": 2.6335071873807405, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.141584396362305, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8937105536460876, + "num_tokens": 803912620.0, + "step": 20702 + }, + { + "epoch": 2.633634397659331, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.942455291748047, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8843768835067749, + "num_tokens": 803953968.0, + "step": 20703 + }, + { + "epoch": 2.633761607937921, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.890854835510254, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9020631313323975, + "num_tokens": 803990042.0, + "step": 20704 + }, + { + "epoch": 2.633888818216512, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.878265380859375, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8976322412490845, + "num_tokens": 804031765.0, + "step": 20705 + }, + { + "epoch": 2.634016028495102, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.878617286682129, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.9038519859313965, + "num_tokens": 804066642.0, + "step": 20706 + }, + { + "epoch": 2.634143238773693, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.89548397064209, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8955108523368835, + "num_tokens": 804109086.0, + "step": 20707 + }, + { + "epoch": 2.6342704490522832, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.84569263458252, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.90496826171875, + "num_tokens": 804148433.0, + "step": 20708 + }, + { + "epoch": 2.6343976593308738, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.904510498046875, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8947809338569641, + "num_tokens": 804190403.0, + "step": 20709 + }, + { + "epoch": 2.6345248696094643, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.967023849487305, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8968839645385742, + "num_tokens": 804229716.0, + "step": 20710 + }, + { + "epoch": 2.634652079888055, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.89786434173584, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8983674049377441, + "num_tokens": 804265445.0, + "step": 20711 + }, + { + "epoch": 2.6347792901666454, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.899069786071777, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9076578617095947, + "num_tokens": 804301723.0, + "step": 20712 + }, + { + "epoch": 2.634906500445236, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.902087211608887, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.898647129535675, + "num_tokens": 804339312.0, + "step": 20713 + }, + { + "epoch": 2.6350337107238264, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.860079765319824, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8979560136795044, + "num_tokens": 804379950.0, + "step": 20714 + }, + { + "epoch": 2.635160921002417, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.895171165466309, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8949188590049744, + "num_tokens": 804419864.0, + "step": 20715 + }, + { + "epoch": 2.6352881312810075, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.98606014251709, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8961048722267151, + "num_tokens": 804457112.0, + "step": 20716 + }, + { + "epoch": 2.635415341559598, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.81392765045166, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8949260711669922, + "num_tokens": 804498592.0, + "step": 20717 + }, + { + "epoch": 2.6355425518381885, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.965496063232422, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.9027044177055359, + "num_tokens": 804531772.0, + "step": 20718 + }, + { + "epoch": 2.635669762116779, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.918584823608398, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8871902823448181, + "num_tokens": 804569358.0, + "step": 20719 + }, + { + "epoch": 2.6357969723953696, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.940858840942383, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8921770453453064, + "num_tokens": 804606964.0, + "step": 20720 + }, + { + "epoch": 2.63592418267396, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.88066577911377, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8996769785881042, + "num_tokens": 804643439.0, + "step": 20721 + }, + { + "epoch": 2.6360513929525506, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.96260929107666, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8982535600662231, + "num_tokens": 804680033.0, + "step": 20722 + }, + { + "epoch": 2.636178603231141, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.96563720703125, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8902404308319092, + "num_tokens": 804715189.0, + "step": 20723 + }, + { + "epoch": 2.6363058135097317, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.984951972961426, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8956798911094666, + "num_tokens": 804754341.0, + "step": 20724 + }, + { + "epoch": 2.6364330237883222, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.89599323272705, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9131494760513306, + "num_tokens": 804786343.0, + "step": 20725 + }, + { + "epoch": 2.6365602340669128, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.04350757598877, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8981530666351318, + "num_tokens": 804816572.0, + "step": 20726 + }, + { + "epoch": 2.636687444345503, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.89719295501709, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8964446783065796, + "num_tokens": 804857508.0, + "step": 20727 + }, + { + "epoch": 2.636814654624094, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.895105361938477, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9084665775299072, + "num_tokens": 804897181.0, + "step": 20728 + }, + { + "epoch": 2.636941864902684, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.885536193847656, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.900134801864624, + "num_tokens": 804935586.0, + "step": 20729 + }, + { + "epoch": 2.637069075181275, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.941242218017578, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8980370759963989, + "num_tokens": 804972038.0, + "step": 20730 + }, + { + "epoch": 2.637196285459865, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.90949821472168, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8831828832626343, + "num_tokens": 805017655.0, + "step": 20731 + }, + { + "epoch": 2.637323495738456, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.878776550292969, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9096868634223938, + "num_tokens": 805056958.0, + "step": 20732 + }, + { + "epoch": 2.637450706017046, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.945667266845703, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8981642723083496, + "num_tokens": 805094229.0, + "step": 20733 + }, + { + "epoch": 2.6375779162956365, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.903438568115234, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8952745199203491, + "num_tokens": 805135850.0, + "step": 20734 + }, + { + "epoch": 2.637705126574227, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.908266067504883, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8952004909515381, + "num_tokens": 805177024.0, + "step": 20735 + }, + { + "epoch": 2.6378323368528176, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.030277252197266, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8879019021987915, + "num_tokens": 805212854.0, + "step": 20736 + }, + { + "epoch": 2.637959547131408, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.818486213684082, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9081742763519287, + "num_tokens": 805251860.0, + "step": 20737 + }, + { + "epoch": 2.6380867574099987, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.1075439453125, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.905945360660553, + "num_tokens": 805280825.0, + "step": 20738 + }, + { + "epoch": 2.638213967688589, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.887145042419434, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9028638601303101, + "num_tokens": 805319488.0, + "step": 20739 + }, + { + "epoch": 2.6383411779671797, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.91827392578125, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.895275354385376, + "num_tokens": 805358840.0, + "step": 20740 + }, + { + "epoch": 2.6384683882457702, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.024587631225586, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.894357442855835, + "num_tokens": 805391395.0, + "step": 20741 + }, + { + "epoch": 2.6385955985243608, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.828634262084961, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8903738260269165, + "num_tokens": 805432624.0, + "step": 20742 + }, + { + "epoch": 2.6387228088029513, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.892876625061035, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.900169312953949, + "num_tokens": 805472007.0, + "step": 20743 + }, + { + "epoch": 2.638850019081542, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.804207801818848, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9140323400497437, + "num_tokens": 805510232.0, + "step": 20744 + }, + { + "epoch": 2.6389772293601323, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.925329208374023, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9040929079055786, + "num_tokens": 805544493.0, + "step": 20745 + }, + { + "epoch": 2.639104439638723, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.887951850891113, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8869370222091675, + "num_tokens": 805587378.0, + "step": 20746 + }, + { + "epoch": 2.6392316499173134, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.813108444213867, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8985285758972168, + "num_tokens": 805633637.0, + "step": 20747 + }, + { + "epoch": 2.639358860195904, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.957110404968262, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8890655636787415, + "num_tokens": 805670080.0, + "step": 20748 + }, + { + "epoch": 2.6394860704744945, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.026975631713867, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8879605531692505, + "num_tokens": 805706068.0, + "step": 20749 + }, + { + "epoch": 2.639613280753085, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.860326766967773, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.9048894047737122, + "num_tokens": 805744236.0, + "step": 20750 + }, + { + "epoch": 2.6397404910316755, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.874418258666992, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.9065937399864197, + "num_tokens": 805780726.0, + "step": 20751 + }, + { + "epoch": 2.6398677013102656, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.041704177856445, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.9022656679153442, + "num_tokens": 805812128.0, + "step": 20752 + }, + { + "epoch": 2.6399949115888566, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.740578651428223, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9075106382369995, + "num_tokens": 805853831.0, + "step": 20753 + }, + { + "epoch": 2.6401221218674467, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.841631889343262, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8950380086898804, + "num_tokens": 805902675.0, + "step": 20754 + }, + { + "epoch": 2.6402493321460376, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.944138526916504, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9011006355285645, + "num_tokens": 805941760.0, + "step": 20755 + }, + { + "epoch": 2.6403765424246277, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.90472412109375, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.9035696983337402, + "num_tokens": 805977212.0, + "step": 20756 + }, + { + "epoch": 2.6405037527032187, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.907830238342285, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8971164226531982, + "num_tokens": 806021622.0, + "step": 20757 + }, + { + "epoch": 2.6406309629818088, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.788724899291992, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.90670245885849, + "num_tokens": 806063959.0, + "step": 20758 + }, + { + "epoch": 2.6407581732603993, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.007264137268066, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.9056026339530945, + "num_tokens": 806096761.0, + "step": 20759 + }, + { + "epoch": 2.64088538353899, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.868855476379395, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9145265221595764, + "num_tokens": 806138966.0, + "step": 20760 + }, + { + "epoch": 2.6410125938175804, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.856158256530762, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.9009121656417847, + "num_tokens": 806182958.0, + "step": 20761 + }, + { + "epoch": 2.641139804096171, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.051037788391113, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.883736252784729, + "num_tokens": 806216156.0, + "step": 20762 + }, + { + "epoch": 2.6412670143747614, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.983759880065918, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.9044042825698853, + "num_tokens": 806249625.0, + "step": 20763 + }, + { + "epoch": 2.641394224653352, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.021161079406738, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8932480812072754, + "num_tokens": 806285524.0, + "step": 20764 + }, + { + "epoch": 2.6415214349319425, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.886199951171875, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8987188935279846, + "num_tokens": 806325587.0, + "step": 20765 + }, + { + "epoch": 2.641648645210533, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.877320289611816, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8962692022323608, + "num_tokens": 806363201.0, + "step": 20766 + }, + { + "epoch": 2.6417758554891235, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.928099632263184, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.898446798324585, + "num_tokens": 806398962.0, + "step": 20767 + }, + { + "epoch": 2.641903065767714, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.81712818145752, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.9017153382301331, + "num_tokens": 806447158.0, + "step": 20768 + }, + { + "epoch": 2.6420302760463046, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.922187805175781, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.905105710029602, + "num_tokens": 806483560.0, + "step": 20769 + }, + { + "epoch": 2.642157486324895, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.868069648742676, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9027230739593506, + "num_tokens": 806522236.0, + "step": 20770 + }, + { + "epoch": 2.6422846966034856, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.931732177734375, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8970394730567932, + "num_tokens": 806561903.0, + "step": 20771 + }, + { + "epoch": 2.642411906882076, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.910035133361816, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.9014550447463989, + "num_tokens": 806595746.0, + "step": 20772 + }, + { + "epoch": 2.6425391171606667, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.920219421386719, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.897698163986206, + "num_tokens": 806635202.0, + "step": 20773 + }, + { + "epoch": 2.6426663274392572, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.88731861114502, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8856521844863892, + "num_tokens": 806678125.0, + "step": 20774 + }, + { + "epoch": 2.6427935377178478, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.947683334350586, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8929367661476135, + "num_tokens": 806722936.0, + "step": 20775 + }, + { + "epoch": 2.6429207479964383, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.953792572021484, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8944042325019836, + "num_tokens": 806763493.0, + "step": 20776 + }, + { + "epoch": 2.6430479582750284, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.948615074157715, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8993603587150574, + "num_tokens": 806801866.0, + "step": 20777 + }, + { + "epoch": 2.6431751685536193, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.999588012695312, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.9010506272315979, + "num_tokens": 806842345.0, + "step": 20778 + }, + { + "epoch": 2.6433023788322094, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.812082290649414, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9031728506088257, + "num_tokens": 806884988.0, + "step": 20779 + }, + { + "epoch": 2.6434295891108004, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.896095275878906, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.9014502763748169, + "num_tokens": 806921595.0, + "step": 20780 + }, + { + "epoch": 2.6435567993893905, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.886674880981445, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.9016165137290955, + "num_tokens": 806963848.0, + "step": 20781 + }, + { + "epoch": 2.6436840096679814, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.835649490356445, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9105473756790161, + "num_tokens": 807003629.0, + "step": 20782 + }, + { + "epoch": 2.6438112199465715, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.901270866394043, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9019722938537598, + "num_tokens": 807043063.0, + "step": 20783 + }, + { + "epoch": 2.643938430225162, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.930397033691406, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9032406806945801, + "num_tokens": 807082316.0, + "step": 20784 + }, + { + "epoch": 2.6440656405037526, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.016103744506836, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.8790975213050842, + "num_tokens": 807123281.0, + "step": 20785 + }, + { + "epoch": 2.644192850782343, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.028019905090332, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8753179311752319, + "num_tokens": 807158567.0, + "step": 20786 + }, + { + "epoch": 2.6443200610609336, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.873607635498047, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.9009484052658081, + "num_tokens": 807203787.0, + "step": 20787 + }, + { + "epoch": 2.644447271339524, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.879837989807129, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.9071958661079407, + "num_tokens": 807243108.0, + "step": 20788 + }, + { + "epoch": 2.6445744816181147, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.956319808959961, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9045886993408203, + "num_tokens": 807280471.0, + "step": 20789 + }, + { + "epoch": 2.6447016918967052, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.968091011047363, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8943201303482056, + "num_tokens": 807321765.0, + "step": 20790 + }, + { + "epoch": 2.6448289021752958, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.808056831359863, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8979631066322327, + "num_tokens": 807366477.0, + "step": 20791 + }, + { + "epoch": 2.6449561124538863, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 10.116385459899902, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8865363597869873, + "num_tokens": 807398911.0, + "step": 20792 + }, + { + "epoch": 2.645083322732477, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.894887924194336, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9040660262107849, + "num_tokens": 807436234.0, + "step": 20793 + }, + { + "epoch": 2.6452105330110673, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.960694313049316, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8936164379119873, + "num_tokens": 807469772.0, + "step": 20794 + }, + { + "epoch": 2.645337743289658, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.957795143127441, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8974459767341614, + "num_tokens": 807507276.0, + "step": 20795 + }, + { + "epoch": 2.6454649535682484, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.804919242858887, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.9060849547386169, + "num_tokens": 807545464.0, + "step": 20796 + }, + { + "epoch": 2.645592163846839, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.912943840026855, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9094517827033997, + "num_tokens": 807579450.0, + "step": 20797 + }, + { + "epoch": 2.6457193741254295, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.922456741333008, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.898076593875885, + "num_tokens": 807619038.0, + "step": 20798 + }, + { + "epoch": 2.64584658440402, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.90049934387207, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9086945652961731, + "num_tokens": 807656482.0, + "step": 20799 + }, + { + "epoch": 2.6459737946826105, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.86763858795166, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.90635085105896, + "num_tokens": 807696933.0, + "step": 20800 + }, + { + "epoch": 2.646101004961201, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.986194610595703, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.9082618355751038, + "num_tokens": 807733395.0, + "step": 20801 + }, + { + "epoch": 2.646228215239791, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.946747779846191, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8920335173606873, + "num_tokens": 807771489.0, + "step": 20802 + }, + { + "epoch": 2.646355425518382, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.038971900939941, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.890710711479187, + "num_tokens": 807807494.0, + "step": 20803 + }, + { + "epoch": 2.646482635796972, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.872973442077637, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8994094729423523, + "num_tokens": 807850109.0, + "step": 20804 + }, + { + "epoch": 2.646609846075563, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.924314498901367, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8986502885818481, + "num_tokens": 807889285.0, + "step": 20805 + }, + { + "epoch": 2.6467370563541532, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.961342811584473, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8967611789703369, + "num_tokens": 807931522.0, + "step": 20806 + }, + { + "epoch": 2.6468642666327438, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.871624946594238, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.901826024055481, + "num_tokens": 807971559.0, + "step": 20807 + }, + { + "epoch": 2.6469914769113343, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.997694969177246, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9067650437355042, + "num_tokens": 808008129.0, + "step": 20808 + }, + { + "epoch": 2.647118687189925, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.89756965637207, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9024381041526794, + "num_tokens": 808047484.0, + "step": 20809 + }, + { + "epoch": 2.6472458974685154, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.064068794250488, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8810336589813232, + "num_tokens": 808083954.0, + "step": 20810 + }, + { + "epoch": 2.647373107747106, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.917031288146973, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8939768075942993, + "num_tokens": 808124745.0, + "step": 20811 + }, + { + "epoch": 2.6475003180256964, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.962509155273438, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8939617872238159, + "num_tokens": 808161996.0, + "step": 20812 + }, + { + "epoch": 2.647627528304287, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.964922904968262, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9006634950637817, + "num_tokens": 808202423.0, + "step": 20813 + }, + { + "epoch": 2.6477547385828775, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.931076049804688, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8987106680870056, + "num_tokens": 808234773.0, + "step": 20814 + }, + { + "epoch": 2.647881948861468, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.950376510620117, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.879460334777832, + "num_tokens": 808280731.0, + "step": 20815 + }, + { + "epoch": 2.6480091591400585, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.881244659423828, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8949323296546936, + "num_tokens": 808320477.0, + "step": 20816 + }, + { + "epoch": 2.648136369418649, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.909980773925781, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9094468951225281, + "num_tokens": 808357644.0, + "step": 20817 + }, + { + "epoch": 2.6482635796972396, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.840437889099121, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.9027036428451538, + "num_tokens": 808400290.0, + "step": 20818 + }, + { + "epoch": 2.64839078997583, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.870277404785156, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8951834440231323, + "num_tokens": 808447888.0, + "step": 20819 + }, + { + "epoch": 2.6485180002544206, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.97443675994873, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8904659748077393, + "num_tokens": 808483664.0, + "step": 20820 + }, + { + "epoch": 2.648645210533011, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.86072826385498, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9062052965164185, + "num_tokens": 808523795.0, + "step": 20821 + }, + { + "epoch": 2.6487724208116017, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.010395050048828, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8944333791732788, + "num_tokens": 808559733.0, + "step": 20822 + }, + { + "epoch": 2.648899631090192, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.982645988464355, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8899726867675781, + "num_tokens": 808597783.0, + "step": 20823 + }, + { + "epoch": 2.6490268413687827, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.814197540283203, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.9077537059783936, + "num_tokens": 808637777.0, + "step": 20824 + }, + { + "epoch": 2.649154051647373, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.94604206085205, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.9083330631256104, + "num_tokens": 808672750.0, + "step": 20825 + }, + { + "epoch": 2.649281261925964, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.941021919250488, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8898537755012512, + "num_tokens": 808710861.0, + "step": 20826 + }, + { + "epoch": 2.649408472204554, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.855362892150879, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.9027106761932373, + "num_tokens": 808750103.0, + "step": 20827 + }, + { + "epoch": 2.649535682483145, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.868812561035156, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.9001379013061523, + "num_tokens": 808793121.0, + "step": 20828 + }, + { + "epoch": 2.649662892761735, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.925337791442871, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8987247943878174, + "num_tokens": 808834890.0, + "step": 20829 + }, + { + "epoch": 2.649790103040326, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.899836540222168, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8992183804512024, + "num_tokens": 808872134.0, + "step": 20830 + }, + { + "epoch": 2.649917313318916, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.805458068847656, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9027414917945862, + "num_tokens": 808918460.0, + "step": 20831 + }, + { + "epoch": 2.6500445235975065, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.941964149475098, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.9065159559249878, + "num_tokens": 808954008.0, + "step": 20832 + }, + { + "epoch": 2.650171733876097, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.963175773620605, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9028033018112183, + "num_tokens": 808990994.0, + "step": 20833 + }, + { + "epoch": 2.6502989441546876, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.875433921813965, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9120277166366577, + "num_tokens": 809030076.0, + "step": 20834 + }, + { + "epoch": 2.650426154433278, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.965547561645508, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9079209566116333, + "num_tokens": 809065442.0, + "step": 20835 + }, + { + "epoch": 2.6505533647118686, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.959229469299316, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8821299076080322, + "num_tokens": 809106117.0, + "step": 20836 + }, + { + "epoch": 2.650680574990459, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.970036506652832, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8915826082229614, + "num_tokens": 809144897.0, + "step": 20837 + }, + { + "epoch": 2.6508077852690497, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.867269515991211, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8962457776069641, + "num_tokens": 809188906.0, + "step": 20838 + }, + { + "epoch": 2.6509349955476402, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.953709602355957, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9138665199279785, + "num_tokens": 809218495.0, + "step": 20839 + }, + { + "epoch": 2.6510622058262308, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.987279891967773, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8965276479721069, + "num_tokens": 809256162.0, + "step": 20840 + }, + { + "epoch": 2.6511894161048213, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.889735221862793, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8965860605239868, + "num_tokens": 809293868.0, + "step": 20841 + }, + { + "epoch": 2.651316626383412, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.857829093933105, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.915681779384613, + "num_tokens": 809335430.0, + "step": 20842 + }, + { + "epoch": 2.6514438366620023, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.952118873596191, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8961716890335083, + "num_tokens": 809372078.0, + "step": 20843 + }, + { + "epoch": 2.651571046940593, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.87313175201416, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8994222283363342, + "num_tokens": 809412081.0, + "step": 20844 + }, + { + "epoch": 2.6516982572191834, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.87380313873291, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9114943742752075, + "num_tokens": 809454131.0, + "step": 20845 + }, + { + "epoch": 2.651825467497774, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.870880126953125, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8941465616226196, + "num_tokens": 809496514.0, + "step": 20846 + }, + { + "epoch": 2.6519526777763645, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.922863960266113, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8980274200439453, + "num_tokens": 809534968.0, + "step": 20847 + }, + { + "epoch": 2.652079888054955, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.878972053527832, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.907384991645813, + "num_tokens": 809575927.0, + "step": 20848 + }, + { + "epoch": 2.6522070983335455, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.946414947509766, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8946058750152588, + "num_tokens": 809614009.0, + "step": 20849 + }, + { + "epoch": 2.6523343086121356, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.928370475769043, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.9032660126686096, + "num_tokens": 809650776.0, + "step": 20850 + }, + { + "epoch": 2.6524615188907266, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.886429786682129, + "learning_rate": 1e-06, + "loss": 0.2461, + "mean_token_accuracy": 0.9163211584091187, + "num_tokens": 809688956.0, + "step": 20851 + }, + { + "epoch": 2.6525887291693167, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.930416107177734, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8994541168212891, + "num_tokens": 809726502.0, + "step": 20852 + }, + { + "epoch": 2.6527159394479076, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.002012252807617, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8963265419006348, + "num_tokens": 809765116.0, + "step": 20853 + }, + { + "epoch": 2.6528431497264977, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.955387115478516, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8799098134040833, + "num_tokens": 809802455.0, + "step": 20854 + }, + { + "epoch": 2.6529703600050887, + "ewc_loss": 0.021240234375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.877685546875, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.9002912044525146, + "num_tokens": 809841668.0, + "step": 20855 + }, + { + "epoch": 2.6530975702836788, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.890341758728027, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8965415954589844, + "num_tokens": 809881558.0, + "step": 20856 + }, + { + "epoch": 2.6532247805622693, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.983171463012695, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8876751661300659, + "num_tokens": 809918532.0, + "step": 20857 + }, + { + "epoch": 2.65335199084086, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.025907516479492, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8910980820655823, + "num_tokens": 809957816.0, + "step": 20858 + }, + { + "epoch": 2.6534792011194503, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.850894927978516, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8912642002105713, + "num_tokens": 809999659.0, + "step": 20859 + }, + { + "epoch": 2.653606411398041, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.001751899719238, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8851831555366516, + "num_tokens": 810038467.0, + "step": 20860 + }, + { + "epoch": 2.6537336216766314, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.98200798034668, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8949006199836731, + "num_tokens": 810073558.0, + "step": 20861 + }, + { + "epoch": 2.653860831955222, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.897483825683594, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8984806537628174, + "num_tokens": 810110771.0, + "step": 20862 + }, + { + "epoch": 2.6539880422338125, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.927436828613281, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9022377133369446, + "num_tokens": 810149187.0, + "step": 20863 + }, + { + "epoch": 2.654115252512403, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.889275550842285, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8994319438934326, + "num_tokens": 810189250.0, + "step": 20864 + }, + { + "epoch": 2.6542424627909935, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.887811660766602, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8947482705116272, + "num_tokens": 810233304.0, + "step": 20865 + }, + { + "epoch": 2.654369673069584, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.017516136169434, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.9049519300460815, + "num_tokens": 810267023.0, + "step": 20866 + }, + { + "epoch": 2.6544968833481746, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.973593711853027, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8989583253860474, + "num_tokens": 810302707.0, + "step": 20867 + }, + { + "epoch": 2.654624093626765, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.82900333404541, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.889731764793396, + "num_tokens": 810350079.0, + "step": 20868 + }, + { + "epoch": 2.6547513039053556, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.93148136138916, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.890650749206543, + "num_tokens": 810391396.0, + "step": 20869 + }, + { + "epoch": 2.654878514183946, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.918434143066406, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.888960599899292, + "num_tokens": 810430685.0, + "step": 20870 + }, + { + "epoch": 2.6550057244625367, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.060215950012207, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8824734687805176, + "num_tokens": 810463638.0, + "step": 20871 + }, + { + "epoch": 2.655132934741127, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.989676475524902, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8978338241577148, + "num_tokens": 810501720.0, + "step": 20872 + }, + { + "epoch": 2.6552601450197177, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.86503791809082, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.9043541550636292, + "num_tokens": 810541574.0, + "step": 20873 + }, + { + "epoch": 2.6553873552983083, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.011592864990234, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9044623970985413, + "num_tokens": 810572659.0, + "step": 20874 + }, + { + "epoch": 2.6555145655768984, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.83343505859375, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.9020442366600037, + "num_tokens": 810614595.0, + "step": 20875 + }, + { + "epoch": 2.6556417758554893, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.959189414978027, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.892137885093689, + "num_tokens": 810653364.0, + "step": 20876 + }, + { + "epoch": 2.6557689861340794, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.919748306274414, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8958086967468262, + "num_tokens": 810695166.0, + "step": 20877 + }, + { + "epoch": 2.6558961964126704, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.888307571411133, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8872051239013672, + "num_tokens": 810738092.0, + "step": 20878 + }, + { + "epoch": 2.6560234066912605, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.994813919067383, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.898206889629364, + "num_tokens": 810779618.0, + "step": 20879 + }, + { + "epoch": 2.6561506169698514, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.847830772399902, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9064662456512451, + "num_tokens": 810815801.0, + "step": 20880 + }, + { + "epoch": 2.6562778272484415, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.903058052062988, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.9030513763427734, + "num_tokens": 810854185.0, + "step": 20881 + }, + { + "epoch": 2.656405037527032, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.069812774658203, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8915263414382935, + "num_tokens": 810887400.0, + "step": 20882 + }, + { + "epoch": 2.6565322478056226, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.926591873168945, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.9027979969978333, + "num_tokens": 810928698.0, + "step": 20883 + }, + { + "epoch": 2.656659458084213, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.976420402526855, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8910707235336304, + "num_tokens": 810968943.0, + "step": 20884 + }, + { + "epoch": 2.6567866683628036, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.027071952819824, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8907408714294434, + "num_tokens": 811013828.0, + "step": 20885 + }, + { + "epoch": 2.656913878641394, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.98779582977295, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8997911810874939, + "num_tokens": 811049706.0, + "step": 20886 + }, + { + "epoch": 2.6570410889199847, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.015275955200195, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8832607865333557, + "num_tokens": 811086690.0, + "step": 20887 + }, + { + "epoch": 2.6571682991985752, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.902934074401855, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9033923149108887, + "num_tokens": 811126288.0, + "step": 20888 + }, + { + "epoch": 2.6572955094771658, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.989996910095215, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8988672494888306, + "num_tokens": 811160660.0, + "step": 20889 + }, + { + "epoch": 2.6574227197557563, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.953485488891602, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.9005985856056213, + "num_tokens": 811200490.0, + "step": 20890 + }, + { + "epoch": 2.657549930034347, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.082148551940918, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.896364688873291, + "num_tokens": 811229807.0, + "step": 20891 + }, + { + "epoch": 2.6576771403129373, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.90564250946045, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9114938378334045, + "num_tokens": 811263092.0, + "step": 20892 + }, + { + "epoch": 2.657804350591528, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.816557884216309, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9098456501960754, + "num_tokens": 811304099.0, + "step": 20893 + }, + { + "epoch": 2.6579315608701184, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.962890625, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8878477811813354, + "num_tokens": 811346343.0, + "step": 20894 + }, + { + "epoch": 2.658058771148709, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.955753326416016, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8983402252197266, + "num_tokens": 811386034.0, + "step": 20895 + }, + { + "epoch": 2.6581859814272994, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.906946182250977, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.9002973437309265, + "num_tokens": 811429985.0, + "step": 20896 + }, + { + "epoch": 2.65831319170589, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.967551231384277, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.9023171663284302, + "num_tokens": 811466821.0, + "step": 20897 + }, + { + "epoch": 2.6584404019844805, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.015800476074219, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8809236288070679, + "num_tokens": 811503523.0, + "step": 20898 + }, + { + "epoch": 2.658567612263071, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.856932640075684, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9003698825836182, + "num_tokens": 811544040.0, + "step": 20899 + }, + { + "epoch": 2.658694822541661, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.905679702758789, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8993279933929443, + "num_tokens": 811585182.0, + "step": 20900 + }, + { + "epoch": 2.658822032820252, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.884452819824219, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9040289521217346, + "num_tokens": 811625942.0, + "step": 20901 + }, + { + "epoch": 2.658949243098842, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.953706741333008, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8883265256881714, + "num_tokens": 811669124.0, + "step": 20902 + }, + { + "epoch": 2.659076453377433, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.942195892333984, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8971900939941406, + "num_tokens": 811709727.0, + "step": 20903 + }, + { + "epoch": 2.6592036636560232, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.951513290405273, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8852346539497375, + "num_tokens": 811752022.0, + "step": 20904 + }, + { + "epoch": 2.6593308739346138, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.940519332885742, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8990896344184875, + "num_tokens": 811791086.0, + "step": 20905 + }, + { + "epoch": 2.6594580842132043, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.021485328674316, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8937830924987793, + "num_tokens": 811824792.0, + "step": 20906 + }, + { + "epoch": 2.659585294491795, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.947846412658691, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.897159218788147, + "num_tokens": 811861792.0, + "step": 20907 + }, + { + "epoch": 2.6597125047703853, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.98745346069336, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8835062980651855, + "num_tokens": 811900182.0, + "step": 20908 + }, + { + "epoch": 2.659839715048976, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.923373222351074, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8874766230583191, + "num_tokens": 811938784.0, + "step": 20909 + }, + { + "epoch": 2.6599669253275664, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.883014678955078, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8992156982421875, + "num_tokens": 811974774.0, + "step": 20910 + }, + { + "epoch": 2.660094135606157, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.053784370422363, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8871217966079712, + "num_tokens": 812011662.0, + "step": 20911 + }, + { + "epoch": 2.6602213458847475, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.857644081115723, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8856415748596191, + "num_tokens": 812058049.0, + "step": 20912 + }, + { + "epoch": 2.660348556163338, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.942828178405762, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.910529613494873, + "num_tokens": 812091992.0, + "step": 20913 + }, + { + "epoch": 2.6604757664419285, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.897505760192871, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9110861420631409, + "num_tokens": 812130373.0, + "step": 20914 + }, + { + "epoch": 2.660602976720519, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.821492195129395, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.9061577320098877, + "num_tokens": 812177029.0, + "step": 20915 + }, + { + "epoch": 2.6607301869991096, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.928702354431152, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9107615351676941, + "num_tokens": 812217640.0, + "step": 20916 + }, + { + "epoch": 2.6608573972777, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.943721771240234, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8942118287086487, + "num_tokens": 812254305.0, + "step": 20917 + }, + { + "epoch": 2.6609846075562906, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.879864692687988, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.9052161574363708, + "num_tokens": 812296626.0, + "step": 20918 + }, + { + "epoch": 2.661111817834881, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.93105697631836, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8992941379547119, + "num_tokens": 812329168.0, + "step": 20919 + }, + { + "epoch": 2.6612390281134717, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.947956085205078, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.9095417261123657, + "num_tokens": 812364806.0, + "step": 20920 + }, + { + "epoch": 2.661366238392062, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.026472091674805, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8949955105781555, + "num_tokens": 812401946.0, + "step": 20921 + }, + { + "epoch": 2.6614934486706527, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.912792205810547, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9151469469070435, + "num_tokens": 812435874.0, + "step": 20922 + }, + { + "epoch": 2.661620658949243, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.863551139831543, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8964901566505432, + "num_tokens": 812477953.0, + "step": 20923 + }, + { + "epoch": 2.661747869227834, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.936090469360352, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.9064387083053589, + "num_tokens": 812515546.0, + "step": 20924 + }, + { + "epoch": 2.661875079506424, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.938384056091309, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8997281789779663, + "num_tokens": 812552369.0, + "step": 20925 + }, + { + "epoch": 2.662002289785015, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.951083183288574, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8997503519058228, + "num_tokens": 812590014.0, + "step": 20926 + }, + { + "epoch": 2.662129500063605, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.988712310791016, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8880406618118286, + "num_tokens": 812630735.0, + "step": 20927 + }, + { + "epoch": 2.662256710342196, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.81716537475586, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.905521810054779, + "num_tokens": 812671616.0, + "step": 20928 + }, + { + "epoch": 2.662383920620786, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.964561462402344, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8849320411682129, + "num_tokens": 812715480.0, + "step": 20929 + }, + { + "epoch": 2.6625111308993765, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.873385429382324, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8982737064361572, + "num_tokens": 812758981.0, + "step": 20930 + }, + { + "epoch": 2.662638341177967, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.909462928771973, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9038610458374023, + "num_tokens": 812796415.0, + "step": 20931 + }, + { + "epoch": 2.6627655514565576, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.943931579589844, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8922173976898193, + "num_tokens": 812832396.0, + "step": 20932 + }, + { + "epoch": 2.662892761735148, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.987509727478027, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.899762749671936, + "num_tokens": 812870055.0, + "step": 20933 + }, + { + "epoch": 2.6630199720137386, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.974003791809082, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8943924903869629, + "num_tokens": 812911387.0, + "step": 20934 + }, + { + "epoch": 2.663147182292329, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.804510116577148, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8979367613792419, + "num_tokens": 812953357.0, + "step": 20935 + }, + { + "epoch": 2.6632743925709197, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.010669708251953, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8875685930252075, + "num_tokens": 812990775.0, + "step": 20936 + }, + { + "epoch": 2.66340160284951, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.94568920135498, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8987597227096558, + "num_tokens": 813025489.0, + "step": 20937 + }, + { + "epoch": 2.6635288131281007, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.852066040039062, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.9054155349731445, + "num_tokens": 813069550.0, + "step": 20938 + }, + { + "epoch": 2.6636560234066913, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.919174194335938, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.9004248380661011, + "num_tokens": 813108558.0, + "step": 20939 + }, + { + "epoch": 2.663783233685282, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.953529357910156, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.9028816223144531, + "num_tokens": 813146893.0, + "step": 20940 + }, + { + "epoch": 2.6639104439638723, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.298324584960938e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.960336685180664, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8948681354522705, + "num_tokens": 813187727.0, + "step": 20941 + }, + { + "epoch": 2.664037654242463, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.918113708496094, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8948525190353394, + "num_tokens": 813229850.0, + "step": 20942 + }, + { + "epoch": 2.6641648645210534, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.909185409545898, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8944272398948669, + "num_tokens": 813265923.0, + "step": 20943 + }, + { + "epoch": 2.664292074799644, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.89374828338623, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8890899419784546, + "num_tokens": 813311432.0, + "step": 20944 + }, + { + "epoch": 2.6644192850782344, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.763882637023926, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9040186405181885, + "num_tokens": 813357458.0, + "step": 20945 + }, + { + "epoch": 2.664546495356825, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.024940490722656, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.9002809524536133, + "num_tokens": 813390971.0, + "step": 20946 + }, + { + "epoch": 2.6646737056354155, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.915999412536621, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8961267471313477, + "num_tokens": 813434212.0, + "step": 20947 + }, + { + "epoch": 2.6648009159140056, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.95795726776123, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8961388468742371, + "num_tokens": 813476163.0, + "step": 20948 + }, + { + "epoch": 2.6649281261925966, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.049675941467285, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8939580917358398, + "num_tokens": 813509729.0, + "step": 20949 + }, + { + "epoch": 2.6650553364711866, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.95227336883545, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8933249711990356, + "num_tokens": 813546056.0, + "step": 20950 + }, + { + "epoch": 2.6651825467497776, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.937933921813965, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8871392011642456, + "num_tokens": 813585726.0, + "step": 20951 + }, + { + "epoch": 2.6653097570283677, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.999482154846191, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.9016901850700378, + "num_tokens": 813622264.0, + "step": 20952 + }, + { + "epoch": 2.6654369673069587, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.882230758666992, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8931612968444824, + "num_tokens": 813664726.0, + "step": 20953 + }, + { + "epoch": 2.6655641775855488, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.973467826843262, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8851681351661682, + "num_tokens": 813707920.0, + "step": 20954 + }, + { + "epoch": 2.6656913878641393, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.857563972473145, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.9048923254013062, + "num_tokens": 813751256.0, + "step": 20955 + }, + { + "epoch": 2.66581859814273, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.899341583251953, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8836051225662231, + "num_tokens": 813790586.0, + "step": 20956 + }, + { + "epoch": 2.6659458084213203, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.012261390686035, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8927180767059326, + "num_tokens": 813831364.0, + "step": 20957 + }, + { + "epoch": 2.666073018699911, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.921113967895508, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8962041139602661, + "num_tokens": 813871668.0, + "step": 20958 + }, + { + "epoch": 2.6662002289785014, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.901490211486816, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9036158919334412, + "num_tokens": 813909690.0, + "step": 20959 + }, + { + "epoch": 2.666327439257092, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.992645263671875, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8827978372573853, + "num_tokens": 813945893.0, + "step": 20960 + }, + { + "epoch": 2.6664546495356825, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.8759765625, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8959139585494995, + "num_tokens": 813988375.0, + "step": 20961 + }, + { + "epoch": 2.666581859814273, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.843305587768555, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8970619440078735, + "num_tokens": 814030896.0, + "step": 20962 + }, + { + "epoch": 2.6667090700928635, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.904345512390137, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.909395694732666, + "num_tokens": 814071288.0, + "step": 20963 + }, + { + "epoch": 2.666836280371454, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.94140911102295, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.9056344032287598, + "num_tokens": 814105668.0, + "step": 20964 + }, + { + "epoch": 2.6669634906500446, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.853754997253418, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.9004205465316772, + "num_tokens": 814149326.0, + "step": 20965 + }, + { + "epoch": 2.667090700928635, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.100637435913086, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.884433388710022, + "num_tokens": 814184124.0, + "step": 20966 + }, + { + "epoch": 2.6672179112072256, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.927477836608887, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8990697860717773, + "num_tokens": 814218357.0, + "step": 20967 + }, + { + "epoch": 2.667345121485816, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.84308910369873, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9123783111572266, + "num_tokens": 814261426.0, + "step": 20968 + }, + { + "epoch": 2.6674723317644067, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.98098087310791, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9062120318412781, + "num_tokens": 814290923.0, + "step": 20969 + }, + { + "epoch": 2.667599542042997, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.957660675048828, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8988319039344788, + "num_tokens": 814327242.0, + "step": 20970 + }, + { + "epoch": 2.6677267523215877, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.024463653564453, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.9047921895980835, + "num_tokens": 814365278.0, + "step": 20971 + }, + { + "epoch": 2.6678539626001783, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.888250350952148, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9058085083961487, + "num_tokens": 814405450.0, + "step": 20972 + }, + { + "epoch": 2.6679811728787683, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.981186866760254, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8987195491790771, + "num_tokens": 814444977.0, + "step": 20973 + }, + { + "epoch": 2.6681083831573593, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.935259819030762, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.9069647789001465, + "num_tokens": 814483110.0, + "step": 20974 + }, + { + "epoch": 2.6682355934359494, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.867956161499023, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8985404968261719, + "num_tokens": 814531018.0, + "step": 20975 + }, + { + "epoch": 2.6683628037145404, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.035879135131836, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8824608325958252, + "num_tokens": 814567989.0, + "step": 20976 + }, + { + "epoch": 2.6684900139931305, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.895573616027832, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.9059247970581055, + "num_tokens": 814605591.0, + "step": 20977 + }, + { + "epoch": 2.668617224271721, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.93038272857666, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8994629979133606, + "num_tokens": 814644528.0, + "step": 20978 + }, + { + "epoch": 2.6687444345503115, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.936264991760254, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9071546792984009, + "num_tokens": 814681884.0, + "step": 20979 + }, + { + "epoch": 2.668871644828902, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.908534049987793, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8912245035171509, + "num_tokens": 814726189.0, + "step": 20980 + }, + { + "epoch": 2.6689988551074926, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.977953910827637, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.9050413370132446, + "num_tokens": 814765878.0, + "step": 20981 + }, + { + "epoch": 2.669126065386083, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.960855484008789, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8993133306503296, + "num_tokens": 814799773.0, + "step": 20982 + }, + { + "epoch": 2.6692532756646736, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.951454162597656, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8914774656295776, + "num_tokens": 814839786.0, + "step": 20983 + }, + { + "epoch": 2.669380485943264, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.93099308013916, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8971623182296753, + "num_tokens": 814874598.0, + "step": 20984 + }, + { + "epoch": 2.6695076962218547, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.90405559539795, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8930796980857849, + "num_tokens": 814915850.0, + "step": 20985 + }, + { + "epoch": 2.669634906500445, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.90850830078125, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9054427146911621, + "num_tokens": 814955178.0, + "step": 20986 + }, + { + "epoch": 2.6697621167790357, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.971684455871582, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.881024181842804, + "num_tokens": 815000390.0, + "step": 20987 + }, + { + "epoch": 2.6698893270576263, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.86103630065918, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8974311351776123, + "num_tokens": 815042722.0, + "step": 20988 + }, + { + "epoch": 2.670016537336217, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.949009895324707, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9049638509750366, + "num_tokens": 815080893.0, + "step": 20989 + }, + { + "epoch": 2.6701437476148073, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.967574119567871, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.9034581780433655, + "num_tokens": 815118315.0, + "step": 20990 + }, + { + "epoch": 2.670270957893398, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.045851707458496, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8888798356056213, + "num_tokens": 815153717.0, + "step": 20991 + }, + { + "epoch": 2.6703981681719884, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.002161979675293, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.880099892616272, + "num_tokens": 815190110.0, + "step": 20992 + }, + { + "epoch": 2.670525378450579, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.968672752380371, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9012138843536377, + "num_tokens": 815224985.0, + "step": 20993 + }, + { + "epoch": 2.6706525887291694, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.913614273071289, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.900506854057312, + "num_tokens": 815263704.0, + "step": 20994 + }, + { + "epoch": 2.67077979900776, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.890254974365234, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9012361168861389, + "num_tokens": 815302658.0, + "step": 20995 + }, + { + "epoch": 2.6709070092863505, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.880895614624023, + "learning_rate": 1e-06, + "loss": 0.2412, + "mean_token_accuracy": 0.917487382888794, + "num_tokens": 815334587.0, + "step": 20996 + }, + { + "epoch": 2.671034219564941, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.034223556518555, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8928825855255127, + "num_tokens": 815370423.0, + "step": 20997 + }, + { + "epoch": 2.671161429843531, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.896834373474121, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.9038419723510742, + "num_tokens": 815409548.0, + "step": 20998 + }, + { + "epoch": 2.671288640122122, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.880414009094238, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.9013081789016724, + "num_tokens": 815454062.0, + "step": 20999 + }, + { + "epoch": 2.671415850400712, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.974206924438477, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9135783314704895, + "num_tokens": 815491380.0, + "step": 21000 + }, + { + "epoch": 2.671543060679303, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.026017189025879, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8911611437797546, + "num_tokens": 815523661.0, + "step": 21001 + }, + { + "epoch": 2.6716702709578932, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.981999397277832, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8938331604003906, + "num_tokens": 815563283.0, + "step": 21002 + }, + { + "epoch": 2.6717974812364838, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.823347091674805, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.908510684967041, + "num_tokens": 815603221.0, + "step": 21003 + }, + { + "epoch": 2.6719246915150743, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.923884391784668, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9051893949508667, + "num_tokens": 815638137.0, + "step": 21004 + }, + { + "epoch": 2.672051901793665, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.078367233276367, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8993479013442993, + "num_tokens": 815679800.0, + "step": 21005 + }, + { + "epoch": 2.6721791120722553, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.845710754394531, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9020987153053284, + "num_tokens": 815721437.0, + "step": 21006 + }, + { + "epoch": 2.672306322350846, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.887269973754883, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8959053754806519, + "num_tokens": 815766583.0, + "step": 21007 + }, + { + "epoch": 2.6724335326294364, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.966535568237305, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8927679061889648, + "num_tokens": 815807587.0, + "step": 21008 + }, + { + "epoch": 2.672560742908027, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.021041870117188, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8924597501754761, + "num_tokens": 815847260.0, + "step": 21009 + }, + { + "epoch": 2.6726879531866174, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.917381286621094, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8944640755653381, + "num_tokens": 815885552.0, + "step": 21010 + }, + { + "epoch": 2.672815163465208, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.905787467956543, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8908027410507202, + "num_tokens": 815927987.0, + "step": 21011 + }, + { + "epoch": 2.6729423737437985, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.140198707580566, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8805122375488281, + "num_tokens": 815965167.0, + "step": 21012 + }, + { + "epoch": 2.673069584022389, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.905675888061523, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8921002149581909, + "num_tokens": 816005080.0, + "step": 21013 + }, + { + "epoch": 2.6731967943009796, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.88231086730957, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.9076312780380249, + "num_tokens": 816040785.0, + "step": 21014 + }, + { + "epoch": 2.67332400457957, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.962794303894043, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.9008182883262634, + "num_tokens": 816076749.0, + "step": 21015 + }, + { + "epoch": 2.6734512148581606, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.04671573638916, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8996337056159973, + "num_tokens": 816110015.0, + "step": 21016 + }, + { + "epoch": 2.673578425136751, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.0010404586792, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.9005618095397949, + "num_tokens": 816144456.0, + "step": 21017 + }, + { + "epoch": 2.6737056354153417, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.89741325378418, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8964642286300659, + "num_tokens": 816185849.0, + "step": 21018 + }, + { + "epoch": 2.673832845693932, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.84853744506836, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.9006162881851196, + "num_tokens": 816228375.0, + "step": 21019 + }, + { + "epoch": 2.6739600559725227, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.933771133422852, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9005787372589111, + "num_tokens": 816270109.0, + "step": 21020 + }, + { + "epoch": 2.674087266251113, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.958895683288574, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.899998664855957, + "num_tokens": 816308091.0, + "step": 21021 + }, + { + "epoch": 2.674214476529704, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.971938133239746, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9038106799125671, + "num_tokens": 816345675.0, + "step": 21022 + }, + { + "epoch": 2.674341686808294, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.982782363891602, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8833281993865967, + "num_tokens": 816386069.0, + "step": 21023 + }, + { + "epoch": 2.674468897086885, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.882244110107422, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.9046281576156616, + "num_tokens": 816425632.0, + "step": 21024 + }, + { + "epoch": 2.674596107365475, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.812281608581543, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8935315012931824, + "num_tokens": 816471878.0, + "step": 21025 + }, + { + "epoch": 2.674723317644066, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.932100296020508, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.9051721096038818, + "num_tokens": 816517224.0, + "step": 21026 + }, + { + "epoch": 2.674850527922656, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.832025527954102, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9036070108413696, + "num_tokens": 816559480.0, + "step": 21027 + }, + { + "epoch": 2.6749777382012465, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.928091049194336, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.90450519323349, + "num_tokens": 816598451.0, + "step": 21028 + }, + { + "epoch": 2.675104948479837, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.008184432983398, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8927285671234131, + "num_tokens": 816634872.0, + "step": 21029 + }, + { + "epoch": 2.6752321587584276, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.909455299377441, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9050021767616272, + "num_tokens": 816678406.0, + "step": 21030 + }, + { + "epoch": 2.675359369037018, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.839190483093262, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.903412401676178, + "num_tokens": 816721143.0, + "step": 21031 + }, + { + "epoch": 2.6754865793156086, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.931036949157715, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.901609480381012, + "num_tokens": 816756665.0, + "step": 21032 + }, + { + "epoch": 2.675613789594199, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.9078950881958, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8967927098274231, + "num_tokens": 816798683.0, + "step": 21033 + }, + { + "epoch": 2.6757409998727897, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.889605522155762, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.9010841250419617, + "num_tokens": 816838329.0, + "step": 21034 + }, + { + "epoch": 2.67586821015138, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.934175491333008, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9134237766265869, + "num_tokens": 816876750.0, + "step": 21035 + }, + { + "epoch": 2.6759954204299707, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.884932518005371, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9123752117156982, + "num_tokens": 816911167.0, + "step": 21036 + }, + { + "epoch": 2.6761226307085613, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.000618934631348, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8970598578453064, + "num_tokens": 816948784.0, + "step": 21037 + }, + { + "epoch": 2.676249840987152, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.01085376739502, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8925608992576599, + "num_tokens": 816987405.0, + "step": 21038 + }, + { + "epoch": 2.6763770512657423, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.863214492797852, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.9045969247817993, + "num_tokens": 817027219.0, + "step": 21039 + }, + { + "epoch": 2.676504261544333, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.805171012878418, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.9079698324203491, + "num_tokens": 817072417.0, + "step": 21040 + }, + { + "epoch": 2.6766314718229234, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.826163291931152, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9064452648162842, + "num_tokens": 817115932.0, + "step": 21041 + }, + { + "epoch": 2.676758682101514, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.877809524536133, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9068409204483032, + "num_tokens": 817154675.0, + "step": 21042 + }, + { + "epoch": 2.6768858923801044, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.999448776245117, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8942053318023682, + "num_tokens": 817193540.0, + "step": 21043 + }, + { + "epoch": 2.677013102658695, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.924580574035645, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.9011892676353455, + "num_tokens": 817232247.0, + "step": 21044 + }, + { + "epoch": 2.6771403129372855, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.940710067749023, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9041186571121216, + "num_tokens": 817272329.0, + "step": 21045 + }, + { + "epoch": 2.6772675232158756, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.832365989685059, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9062839150428772, + "num_tokens": 817309057.0, + "step": 21046 + }, + { + "epoch": 2.6773947334944665, + "ewc_loss": 0.0213623046875, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 10.038668632507324, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8909439444541931, + "num_tokens": 817348658.0, + "step": 21047 + }, + { + "epoch": 2.6775219437730566, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.053686141967773, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8874066472053528, + "num_tokens": 817379480.0, + "step": 21048 + }, + { + "epoch": 2.6776491540516476, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.019967079162598, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8972737789154053, + "num_tokens": 817413630.0, + "step": 21049 + }, + { + "epoch": 2.6777763643302377, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.82568073272705, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.9035007357597351, + "num_tokens": 817454230.0, + "step": 21050 + }, + { + "epoch": 2.6779035746088287, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.889126777648926, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.9035199880599976, + "num_tokens": 817488423.0, + "step": 21051 + }, + { + "epoch": 2.6780307848874187, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.34600830078125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.892476081848145, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8864258527755737, + "num_tokens": 817530369.0, + "step": 21052 + }, + { + "epoch": 2.6781579951660093, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.85358715057373, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.901633620262146, + "num_tokens": 817572231.0, + "step": 21053 + }, + { + "epoch": 2.6782852054446, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.937255859375, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9032283425331116, + "num_tokens": 817604652.0, + "step": 21054 + }, + { + "epoch": 2.6784124157231903, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.919136047363281, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.9026594161987305, + "num_tokens": 817641667.0, + "step": 21055 + }, + { + "epoch": 2.678539626001781, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.910504341125488, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9117565751075745, + "num_tokens": 817679709.0, + "step": 21056 + }, + { + "epoch": 2.6786668362803714, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.876246452331543, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8993252515792847, + "num_tokens": 817726499.0, + "step": 21057 + }, + { + "epoch": 2.678794046558962, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.936137199401855, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.9024453163146973, + "num_tokens": 817766036.0, + "step": 21058 + }, + { + "epoch": 2.6789212568375524, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.9476900100708, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.890326738357544, + "num_tokens": 817802698.0, + "step": 21059 + }, + { + "epoch": 2.679048467116143, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.8989896774292, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.9060840010643005, + "num_tokens": 817841105.0, + "step": 21060 + }, + { + "epoch": 2.6791756773947335, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.836350440979004, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9058299660682678, + "num_tokens": 817882347.0, + "step": 21061 + }, + { + "epoch": 2.679302887673324, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.067584037780762, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.888893187046051, + "num_tokens": 817917460.0, + "step": 21062 + }, + { + "epoch": 2.6794300979519146, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.963674545288086, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.891381561756134, + "num_tokens": 817954237.0, + "step": 21063 + }, + { + "epoch": 2.679557308230505, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.952749252319336, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8955298662185669, + "num_tokens": 817996928.0, + "step": 21064 + }, + { + "epoch": 2.6796845185090956, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.012507438659668, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8953202962875366, + "num_tokens": 818035373.0, + "step": 21065 + }, + { + "epoch": 2.679811728787686, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.968008041381836, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9121229648590088, + "num_tokens": 818069373.0, + "step": 21066 + }, + { + "epoch": 2.6799389390662767, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.897710800170898, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.9010192155838013, + "num_tokens": 818105176.0, + "step": 21067 + }, + { + "epoch": 2.680066149344867, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.903310775756836, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8959975838661194, + "num_tokens": 818146290.0, + "step": 21068 + }, + { + "epoch": 2.6801933596234577, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.814520835876465, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8988680839538574, + "num_tokens": 818188706.0, + "step": 21069 + }, + { + "epoch": 2.6803205699020483, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.936405181884766, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.892217218875885, + "num_tokens": 818227982.0, + "step": 21070 + }, + { + "epoch": 2.6804477801806383, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.856221199035645, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.9006083607673645, + "num_tokens": 818269897.0, + "step": 21071 + }, + { + "epoch": 2.6805749904592293, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.872939109802246, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8998233675956726, + "num_tokens": 818311052.0, + "step": 21072 + }, + { + "epoch": 2.6807022007378194, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.964154243469238, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8984507918357849, + "num_tokens": 818348022.0, + "step": 21073 + }, + { + "epoch": 2.6808294110164104, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.985689163208008, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8930660486221313, + "num_tokens": 818388886.0, + "step": 21074 + }, + { + "epoch": 2.6809566212950005, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.052709579467773, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8822925090789795, + "num_tokens": 818422213.0, + "step": 21075 + }, + { + "epoch": 2.681083831573591, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.910867691040039, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.901286244392395, + "num_tokens": 818461503.0, + "step": 21076 + }, + { + "epoch": 2.6812110418521815, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.92877197265625, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8969879746437073, + "num_tokens": 818498744.0, + "step": 21077 + }, + { + "epoch": 2.681338252130772, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.933091163635254, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9040950536727905, + "num_tokens": 818534508.0, + "step": 21078 + }, + { + "epoch": 2.6814654624093626, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.809847831726074, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.9003414511680603, + "num_tokens": 818582395.0, + "step": 21079 + }, + { + "epoch": 2.681592672687953, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.94826889038086, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9008887410163879, + "num_tokens": 818615994.0, + "step": 21080 + }, + { + "epoch": 2.6817198829665436, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.966915130615234, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8826876878738403, + "num_tokens": 818657014.0, + "step": 21081 + }, + { + "epoch": 2.681847093245134, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.959391593933105, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8815720081329346, + "num_tokens": 818694545.0, + "step": 21082 + }, + { + "epoch": 2.6819743035237247, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.901237487792969, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.9010416865348816, + "num_tokens": 818736503.0, + "step": 21083 + }, + { + "epoch": 2.682101513802315, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.920232772827148, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8934076428413391, + "num_tokens": 818774320.0, + "step": 21084 + }, + { + "epoch": 2.6822287240809057, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.94985580444336, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8990879058837891, + "num_tokens": 818809640.0, + "step": 21085 + }, + { + "epoch": 2.6823559343594963, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.86634635925293, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9036432504653931, + "num_tokens": 818849058.0, + "step": 21086 + }, + { + "epoch": 2.682483144638087, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.89294719696045, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.9020721912384033, + "num_tokens": 818889139.0, + "step": 21087 + }, + { + "epoch": 2.6826103549166773, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.939976692199707, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.9074916839599609, + "num_tokens": 818926360.0, + "step": 21088 + }, + { + "epoch": 2.682737565195268, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.858698844909668, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8935915231704712, + "num_tokens": 818970942.0, + "step": 21089 + }, + { + "epoch": 2.6828647754738584, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.993592262268066, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8964172601699829, + "num_tokens": 819004454.0, + "step": 21090 + }, + { + "epoch": 2.682991985752449, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.91995620727539, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8864637613296509, + "num_tokens": 819045645.0, + "step": 21091 + }, + { + "epoch": 2.6831191960310394, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.026123046875, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8971008062362671, + "num_tokens": 819075719.0, + "step": 21092 + }, + { + "epoch": 2.68324640630963, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.951279640197754, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8918383121490479, + "num_tokens": 819112157.0, + "step": 21093 + }, + { + "epoch": 2.6833736165882205, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.867780685424805, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8980470299720764, + "num_tokens": 819149067.0, + "step": 21094 + }, + { + "epoch": 2.683500826866811, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.981846809387207, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8949332237243652, + "num_tokens": 819185389.0, + "step": 21095 + }, + { + "epoch": 2.683628037145401, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.867202758789062, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9149289131164551, + "num_tokens": 819221306.0, + "step": 21096 + }, + { + "epoch": 2.683755247423992, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.85250473022461, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.9008800983428955, + "num_tokens": 819259697.0, + "step": 21097 + }, + { + "epoch": 2.683882457702582, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.13484001159668, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8866980075836182, + "num_tokens": 819291119.0, + "step": 21098 + }, + { + "epoch": 2.684009667981173, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.905099868774414, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8972290754318237, + "num_tokens": 819331158.0, + "step": 21099 + }, + { + "epoch": 2.684136878259763, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.987913131713867, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.9027096033096313, + "num_tokens": 819364165.0, + "step": 21100 + }, + { + "epoch": 2.6842640885383537, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.976553916931152, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8924898505210876, + "num_tokens": 819403926.0, + "step": 21101 + }, + { + "epoch": 2.6843912988169443, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.921354293823242, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8966563940048218, + "num_tokens": 819444247.0, + "step": 21102 + }, + { + "epoch": 2.684518509095535, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.959179878234863, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8926335573196411, + "num_tokens": 819484562.0, + "step": 21103 + }, + { + "epoch": 2.6846457193741253, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.03458309173584, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8971918225288391, + "num_tokens": 819521230.0, + "step": 21104 + }, + { + "epoch": 2.684772929652716, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.936102867126465, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.9069278836250305, + "num_tokens": 819560265.0, + "step": 21105 + }, + { + "epoch": 2.6849001399313064, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.907698631286621, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.901868462562561, + "num_tokens": 819596500.0, + "step": 21106 + }, + { + "epoch": 2.685027350209897, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.972647666931152, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9010336399078369, + "num_tokens": 819639585.0, + "step": 21107 + }, + { + "epoch": 2.6851545604884874, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.081762313842773, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9035042524337769, + "num_tokens": 819673636.0, + "step": 21108 + }, + { + "epoch": 2.685281770767078, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.104491233825684, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8900623321533203, + "num_tokens": 819710290.0, + "step": 21109 + }, + { + "epoch": 2.6854089810456685, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.054697036743164, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8929961323738098, + "num_tokens": 819743923.0, + "step": 21110 + }, + { + "epoch": 2.685536191324259, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.809930801391602, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8983138203620911, + "num_tokens": 819786427.0, + "step": 21111 + }, + { + "epoch": 2.6856634016028496, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.046812057495117, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8902777433395386, + "num_tokens": 819823724.0, + "step": 21112 + }, + { + "epoch": 2.68579061188144, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.905921936035156, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8996118307113647, + "num_tokens": 819859417.0, + "step": 21113 + }, + { + "epoch": 2.6859178221600306, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.894950866699219, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9091961979866028, + "num_tokens": 819895730.0, + "step": 21114 + }, + { + "epoch": 2.686045032438621, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.99401569366455, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8921021819114685, + "num_tokens": 819934813.0, + "step": 21115 + }, + { + "epoch": 2.6861722427172117, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.988094329833984, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9068104028701782, + "num_tokens": 819973218.0, + "step": 21116 + }, + { + "epoch": 2.686299452995802, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.018610954284668, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8998605012893677, + "num_tokens": 820009662.0, + "step": 21117 + }, + { + "epoch": 2.6864266632743927, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.906344413757324, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8990282416343689, + "num_tokens": 820047468.0, + "step": 21118 + }, + { + "epoch": 2.686553873552983, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.904305458068848, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8981309533119202, + "num_tokens": 820088576.0, + "step": 21119 + }, + { + "epoch": 2.686681083831574, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.097147941589355, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8723714351654053, + "num_tokens": 820129669.0, + "step": 21120 + }, + { + "epoch": 2.686808294110164, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.947606086730957, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.9061535596847534, + "num_tokens": 820163029.0, + "step": 21121 + }, + { + "epoch": 2.686935504388755, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.929396629333496, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8988282680511475, + "num_tokens": 820204762.0, + "step": 21122 + }, + { + "epoch": 2.687062714667345, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.894253730773926, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9090332984924316, + "num_tokens": 820239134.0, + "step": 21123 + }, + { + "epoch": 2.687189924945936, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.08415699005127, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8931108713150024, + "num_tokens": 820280374.0, + "step": 21124 + }, + { + "epoch": 2.687317135224526, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.023151397705078, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8919316530227661, + "num_tokens": 820321173.0, + "step": 21125 + }, + { + "epoch": 2.6874443455031165, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.892544746398926, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8991848230361938, + "num_tokens": 820362690.0, + "step": 21126 + }, + { + "epoch": 2.687571555781707, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.986773490905762, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8955010175704956, + "num_tokens": 820395882.0, + "step": 21127 + }, + { + "epoch": 2.6876987660602976, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.965386390686035, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8931257724761963, + "num_tokens": 820436147.0, + "step": 21128 + }, + { + "epoch": 2.687825976338888, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.957281112670898, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8992395401000977, + "num_tokens": 820472426.0, + "step": 21129 + }, + { + "epoch": 2.6879531866174786, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.001130104064941, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8838765025138855, + "num_tokens": 820507591.0, + "step": 21130 + }, + { + "epoch": 2.688080396896069, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.874895095825195, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.9028416872024536, + "num_tokens": 820547389.0, + "step": 21131 + }, + { + "epoch": 2.6882076071746597, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.055294036865234, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8964525461196899, + "num_tokens": 820579353.0, + "step": 21132 + }, + { + "epoch": 2.68833481745325, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.118242263793945, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.894591212272644, + "num_tokens": 820617847.0, + "step": 21133 + }, + { + "epoch": 2.6884620277318407, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.831785202026367, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8950034379959106, + "num_tokens": 820659098.0, + "step": 21134 + }, + { + "epoch": 2.6885892380104313, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.958852767944336, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.888938307762146, + "num_tokens": 820697789.0, + "step": 21135 + }, + { + "epoch": 2.688716448289022, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.914116859436035, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.89873206615448, + "num_tokens": 820736023.0, + "step": 21136 + }, + { + "epoch": 2.6888436585676123, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.876524925231934, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9118101596832275, + "num_tokens": 820770211.0, + "step": 21137 + }, + { + "epoch": 2.688970868846203, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.958897590637207, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.9055878520011902, + "num_tokens": 820806975.0, + "step": 21138 + }, + { + "epoch": 2.6890980791247934, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.919339179992676, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.9065173864364624, + "num_tokens": 820845796.0, + "step": 21139 + }, + { + "epoch": 2.689225289403384, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.023069381713867, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8999641537666321, + "num_tokens": 820883369.0, + "step": 21140 + }, + { + "epoch": 2.6893524996819744, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.837886810302734, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.9009106159210205, + "num_tokens": 820924928.0, + "step": 21141 + }, + { + "epoch": 2.689479709960565, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.030957221984863, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8977546691894531, + "num_tokens": 820959761.0, + "step": 21142 + }, + { + "epoch": 2.6896069202391555, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.891910552978516, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.898047685623169, + "num_tokens": 820999219.0, + "step": 21143 + }, + { + "epoch": 2.6897341305177456, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.884814262390137, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8955976963043213, + "num_tokens": 821043238.0, + "step": 21144 + }, + { + "epoch": 2.6898613407963365, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.969980239868164, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9040220975875854, + "num_tokens": 821076037.0, + "step": 21145 + }, + { + "epoch": 2.6899885510749266, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.859212875366211, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9056544303894043, + "num_tokens": 821116275.0, + "step": 21146 + }, + { + "epoch": 2.6901157613535176, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.91270923614502, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.9016639590263367, + "num_tokens": 821156128.0, + "step": 21147 + }, + { + "epoch": 2.6902429716321077, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.00336742401123, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8952046632766724, + "num_tokens": 821191657.0, + "step": 21148 + }, + { + "epoch": 2.6903701819106987, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.944889068603516, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8998222947120667, + "num_tokens": 821233298.0, + "step": 21149 + }, + { + "epoch": 2.6904973921892887, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.067516326904297, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8834161162376404, + "num_tokens": 821267897.0, + "step": 21150 + }, + { + "epoch": 2.6906246024678793, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.898543357849121, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8955683708190918, + "num_tokens": 821309164.0, + "step": 21151 + }, + { + "epoch": 2.69075181274647, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.039668083190918, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8830541372299194, + "num_tokens": 821347543.0, + "step": 21152 + }, + { + "epoch": 2.6908790230250603, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.116551399230957, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.895301342010498, + "num_tokens": 821377308.0, + "step": 21153 + }, + { + "epoch": 2.691006233303651, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.972786903381348, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8983508944511414, + "num_tokens": 821411802.0, + "step": 21154 + }, + { + "epoch": 2.6911334435822414, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.061084747314453, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8845345973968506, + "num_tokens": 821446246.0, + "step": 21155 + }, + { + "epoch": 2.691260653860832, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.885431289672852, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9091418981552124, + "num_tokens": 821481998.0, + "step": 21156 + }, + { + "epoch": 2.6913878641394224, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.885181427001953, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.9056987762451172, + "num_tokens": 821522433.0, + "step": 21157 + }, + { + "epoch": 2.691515074418013, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.947999000549316, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8896653056144714, + "num_tokens": 821565949.0, + "step": 21158 + }, + { + "epoch": 2.6916422846966035, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.903533935546875, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8935553431510925, + "num_tokens": 821610828.0, + "step": 21159 + }, + { + "epoch": 2.691769494975194, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.889771461486816, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.9017391204833984, + "num_tokens": 821650925.0, + "step": 21160 + }, + { + "epoch": 2.6918967052537845, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.955867767333984, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9042352437973022, + "num_tokens": 821687602.0, + "step": 21161 + }, + { + "epoch": 2.692023915532375, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.815165519714355, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.9053943157196045, + "num_tokens": 821733498.0, + "step": 21162 + }, + { + "epoch": 2.6921511258109656, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.980501174926758, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.9044706225395203, + "num_tokens": 821774180.0, + "step": 21163 + }, + { + "epoch": 2.692278336089556, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.932853698730469, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8999181389808655, + "num_tokens": 821818169.0, + "step": 21164 + }, + { + "epoch": 2.6924055463681467, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.0897798538208, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.9016724824905396, + "num_tokens": 821851214.0, + "step": 21165 + }, + { + "epoch": 2.692532756646737, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.001535415649414, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8916391134262085, + "num_tokens": 821891730.0, + "step": 21166 + }, + { + "epoch": 2.6926599669253277, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.051873207092285, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.9000381231307983, + "num_tokens": 821922896.0, + "step": 21167 + }, + { + "epoch": 2.6927871772039182, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.10145378112793, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8917708992958069, + "num_tokens": 821954212.0, + "step": 21168 + }, + { + "epoch": 2.6929143874825083, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.908304214477539, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8882886171340942, + "num_tokens": 821991145.0, + "step": 21169 + }, + { + "epoch": 2.6930415977610993, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.874722480773926, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9026369452476501, + "num_tokens": 822031247.0, + "step": 21170 + }, + { + "epoch": 2.6931688080396894, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.004429817199707, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8889739513397217, + "num_tokens": 822071090.0, + "step": 21171 + }, + { + "epoch": 2.6932960183182804, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.914494514465332, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8982999920845032, + "num_tokens": 822114727.0, + "step": 21172 + }, + { + "epoch": 2.6934232285968704, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.837056159973145, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.899410605430603, + "num_tokens": 822159072.0, + "step": 21173 + }, + { + "epoch": 2.693550438875461, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.974952697753906, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.913699746131897, + "num_tokens": 822195158.0, + "step": 21174 + }, + { + "epoch": 2.6936776491540515, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.888131141662598, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.9078865051269531, + "num_tokens": 822236982.0, + "step": 21175 + }, + { + "epoch": 2.693804859432642, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.976569175720215, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9064140319824219, + "num_tokens": 822272069.0, + "step": 21176 + }, + { + "epoch": 2.6939320697112326, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.880764961242676, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8931804895401001, + "num_tokens": 822315451.0, + "step": 21177 + }, + { + "epoch": 2.694059279989823, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.000482559204102, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8978955745697021, + "num_tokens": 822352988.0, + "step": 21178 + }, + { + "epoch": 2.6941864902684136, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.965288162231445, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8983027935028076, + "num_tokens": 822391483.0, + "step": 21179 + }, + { + "epoch": 2.694313700547004, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.0191068649292, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8884931802749634, + "num_tokens": 822425052.0, + "step": 21180 + }, + { + "epoch": 2.6944409108255947, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.441375732421875e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.989686012268066, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8969393372535706, + "num_tokens": 822464995.0, + "step": 21181 + }, + { + "epoch": 2.694568121104185, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.877824783325195, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8938542604446411, + "num_tokens": 822506763.0, + "step": 21182 + }, + { + "epoch": 2.6946953313827757, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.03034496307373, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.9020955562591553, + "num_tokens": 822542762.0, + "step": 21183 + }, + { + "epoch": 2.6948225416613663, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.84373950958252, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8932293653488159, + "num_tokens": 822585637.0, + "step": 21184 + }, + { + "epoch": 2.694949751939957, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.146869659423828, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8680788278579712, + "num_tokens": 822623941.0, + "step": 21185 + }, + { + "epoch": 2.6950769622185473, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.862951278686523, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.9055260419845581, + "num_tokens": 822661897.0, + "step": 21186 + }, + { + "epoch": 2.695204172497138, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.89371395111084, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.905036211013794, + "num_tokens": 822697993.0, + "step": 21187 + }, + { + "epoch": 2.6953313827757284, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.949934959411621, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9107711315155029, + "num_tokens": 822732016.0, + "step": 21188 + }, + { + "epoch": 2.695458593054319, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.078882217407227, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9062246680259705, + "num_tokens": 822764594.0, + "step": 21189 + }, + { + "epoch": 2.6955858033329094, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.937986373901367, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9099604487419128, + "num_tokens": 822799365.0, + "step": 21190 + }, + { + "epoch": 2.6957130136115, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.962288856506348, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9070853590965271, + "num_tokens": 822832265.0, + "step": 21191 + }, + { + "epoch": 2.6958402238900905, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.057429313659668, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8766586780548096, + "num_tokens": 822870630.0, + "step": 21192 + }, + { + "epoch": 2.695967434168681, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.976113319396973, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8938298225402832, + "num_tokens": 822911561.0, + "step": 21193 + }, + { + "epoch": 2.696094644447271, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.891865730285645, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9042673110961914, + "num_tokens": 822952237.0, + "step": 21194 + }, + { + "epoch": 2.696221854725862, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.04589557647705, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.887261152267456, + "num_tokens": 822989006.0, + "step": 21195 + }, + { + "epoch": 2.696349065004452, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.91800594329834, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.887122631072998, + "num_tokens": 823032322.0, + "step": 21196 + }, + { + "epoch": 2.696476275283043, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.992106437683105, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.890975296497345, + "num_tokens": 823067794.0, + "step": 21197 + }, + { + "epoch": 2.696603485561633, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.93631649017334, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8976292610168457, + "num_tokens": 823106903.0, + "step": 21198 + }, + { + "epoch": 2.6967306958402237, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.860002517700195, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.9057551622390747, + "num_tokens": 823145491.0, + "step": 21199 + }, + { + "epoch": 2.6968579061188143, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.898613929748535, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.9038851857185364, + "num_tokens": 823185269.0, + "step": 21200 + }, + { + "epoch": 2.696985116397405, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.815313339233398, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.9054380655288696, + "num_tokens": 823229559.0, + "step": 21201 + }, + { + "epoch": 2.6971123266759953, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.840287208557129, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8960744142532349, + "num_tokens": 823271455.0, + "step": 21202 + }, + { + "epoch": 2.697239536954586, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.960470199584961, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.903399646282196, + "num_tokens": 823308903.0, + "step": 21203 + }, + { + "epoch": 2.6973667472331764, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.8438081741333, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8988535404205322, + "num_tokens": 823354226.0, + "step": 21204 + }, + { + "epoch": 2.697493957511767, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.968018531799316, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9085090160369873, + "num_tokens": 823388052.0, + "step": 21205 + }, + { + "epoch": 2.6976211677903574, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.006202697753906, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8963018655776978, + "num_tokens": 823427386.0, + "step": 21206 + }, + { + "epoch": 2.697748378068948, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.92982006072998, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.9009202718734741, + "num_tokens": 823466018.0, + "step": 21207 + }, + { + "epoch": 2.6978755883475385, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.953810691833496, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8979077339172363, + "num_tokens": 823503510.0, + "step": 21208 + }, + { + "epoch": 2.698002798626129, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.962847709655762, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8993996977806091, + "num_tokens": 823536606.0, + "step": 21209 + }, + { + "epoch": 2.6981300089047195, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.911484718322754, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.9058279991149902, + "num_tokens": 823577025.0, + "step": 21210 + }, + { + "epoch": 2.69825721918331, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.98189640045166, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8852816820144653, + "num_tokens": 823617644.0, + "step": 21211 + }, + { + "epoch": 2.6983844294619006, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.755797386169434, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9185773730278015, + "num_tokens": 823657576.0, + "step": 21212 + }, + { + "epoch": 2.698511639740491, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.982649803161621, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9088079333305359, + "num_tokens": 823692101.0, + "step": 21213 + }, + { + "epoch": 2.6986388500190817, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.924966812133789, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9029989838600159, + "num_tokens": 823728531.0, + "step": 21214 + }, + { + "epoch": 2.698766060297672, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.924357414245605, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8976633548736572, + "num_tokens": 823765222.0, + "step": 21215 + }, + { + "epoch": 2.6988932705762627, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.94528865814209, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8952476978302002, + "num_tokens": 823806832.0, + "step": 21216 + }, + { + "epoch": 2.699020480854853, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.042140007019043, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8934652805328369, + "num_tokens": 823841937.0, + "step": 21217 + }, + { + "epoch": 2.6991476911334438, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.979976654052734, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.9012933969497681, + "num_tokens": 823882723.0, + "step": 21218 + }, + { + "epoch": 2.699274901412034, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.990736961364746, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8899250030517578, + "num_tokens": 823921176.0, + "step": 21219 + }, + { + "epoch": 2.699402111690625, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.878317832946777, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9120584726333618, + "num_tokens": 823964394.0, + "step": 21220 + }, + { + "epoch": 2.699529321969215, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.907797813415527, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9076735377311707, + "num_tokens": 824004897.0, + "step": 21221 + }, + { + "epoch": 2.699656532247806, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.067289352416992, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8985142707824707, + "num_tokens": 824043686.0, + "step": 21222 + }, + { + "epoch": 2.699783742526396, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.870626449584961, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9059008359909058, + "num_tokens": 824086402.0, + "step": 21223 + }, + { + "epoch": 2.6999109528049865, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.007322311401367, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.896653413772583, + "num_tokens": 824124407.0, + "step": 21224 + }, + { + "epoch": 2.700038163083577, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.939212799072266, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8980011343955994, + "num_tokens": 824164366.0, + "step": 21225 + }, + { + "epoch": 2.7001653733621676, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.03648567199707, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.9007434248924255, + "num_tokens": 824200034.0, + "step": 21226 + }, + { + "epoch": 2.700292583640758, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.980475425720215, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8924705982208252, + "num_tokens": 824240588.0, + "step": 21227 + }, + { + "epoch": 2.7004197939193486, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.95816707611084, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8928241729736328, + "num_tokens": 824280852.0, + "step": 21228 + }, + { + "epoch": 2.700547004197939, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.964472770690918, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8973870277404785, + "num_tokens": 824317642.0, + "step": 21229 + }, + { + "epoch": 2.7006742144765297, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.932196617126465, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9032671451568604, + "num_tokens": 824352159.0, + "step": 21230 + }, + { + "epoch": 2.70080142475512, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.9152250289917, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8901346921920776, + "num_tokens": 824391028.0, + "step": 21231 + }, + { + "epoch": 2.7009286350337107, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.926772117614746, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.9000270366668701, + "num_tokens": 824426208.0, + "step": 21232 + }, + { + "epoch": 2.7010558453123013, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.983109474182129, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8903206586837769, + "num_tokens": 824462267.0, + "step": 21233 + }, + { + "epoch": 2.701183055590892, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.897586822509766, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8934944272041321, + "num_tokens": 824500402.0, + "step": 21234 + }, + { + "epoch": 2.7013102658694823, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.902298927307129, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8891376852989197, + "num_tokens": 824545175.0, + "step": 21235 + }, + { + "epoch": 2.701437476148073, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.933014869689941, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.9023350477218628, + "num_tokens": 824590931.0, + "step": 21236 + }, + { + "epoch": 2.7015646864266634, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.919217109680176, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9073569178581238, + "num_tokens": 824629573.0, + "step": 21237 + }, + { + "epoch": 2.701691896705254, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.959940910339355, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8938418626785278, + "num_tokens": 824671029.0, + "step": 21238 + }, + { + "epoch": 2.7018191069838444, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.895040512084961, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8925520181655884, + "num_tokens": 824711535.0, + "step": 21239 + }, + { + "epoch": 2.701946317262435, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.883695602416992, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8935036659240723, + "num_tokens": 824756181.0, + "step": 21240 + }, + { + "epoch": 2.7020735275410255, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.981536865234375, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9072887301445007, + "num_tokens": 824788289.0, + "step": 21241 + }, + { + "epoch": 2.7022007378196156, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.91976547241211, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9020932912826538, + "num_tokens": 824824224.0, + "step": 21242 + }, + { + "epoch": 2.7023279480982065, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.852896690368652, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.9038510918617249, + "num_tokens": 824862020.0, + "step": 21243 + }, + { + "epoch": 2.7024551583767966, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.867986679077148, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9138675332069397, + "num_tokens": 824903470.0, + "step": 21244 + }, + { + "epoch": 2.7025823686553876, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.884997367858887, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8964030146598816, + "num_tokens": 824945838.0, + "step": 21245 + }, + { + "epoch": 2.7027095789339777, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.972784042358398, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.907414436340332, + "num_tokens": 824982398.0, + "step": 21246 + }, + { + "epoch": 2.7028367892125686, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.09988021850586, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8900694847106934, + "num_tokens": 825016419.0, + "step": 21247 + }, + { + "epoch": 2.7029639994911587, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.914876937866211, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.891830563545227, + "num_tokens": 825056689.0, + "step": 21248 + }, + { + "epoch": 2.7030912097697493, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.910691261291504, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.9013605117797852, + "num_tokens": 825096732.0, + "step": 21249 + }, + { + "epoch": 2.70321842004834, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.899410247802734, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9122104644775391, + "num_tokens": 825135114.0, + "step": 21250 + }, + { + "epoch": 2.7033456303269303, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.968548774719238, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8961982131004333, + "num_tokens": 825172441.0, + "step": 21251 + }, + { + "epoch": 2.703472840605521, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.991369247436523, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9011503458023071, + "num_tokens": 825209950.0, + "step": 21252 + }, + { + "epoch": 2.7036000508841114, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.879546165466309, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8960870504379272, + "num_tokens": 825252495.0, + "step": 21253 + }, + { + "epoch": 2.703727261162702, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.922861099243164, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.9009054899215698, + "num_tokens": 825291675.0, + "step": 21254 + }, + { + "epoch": 2.7038544714412924, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.941993713378906, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.9028327465057373, + "num_tokens": 825328857.0, + "step": 21255 + }, + { + "epoch": 2.703981681719883, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.126139640808105, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8899840712547302, + "num_tokens": 825364636.0, + "step": 21256 + }, + { + "epoch": 2.7041088919984735, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.878162384033203, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9098387956619263, + "num_tokens": 825408948.0, + "step": 21257 + }, + { + "epoch": 2.704236102277064, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.993282318115234, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.90021812915802, + "num_tokens": 825445850.0, + "step": 21258 + }, + { + "epoch": 2.7043633125556545, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.982999801635742, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.9022086262702942, + "num_tokens": 825480863.0, + "step": 21259 + }, + { + "epoch": 2.704490522834245, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.879146575927734, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.9034870266914368, + "num_tokens": 825522929.0, + "step": 21260 + }, + { + "epoch": 2.7046177331128356, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.940200805664062, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.9050872325897217, + "num_tokens": 825561642.0, + "step": 21261 + }, + { + "epoch": 2.704744943391426, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.061285972595215, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.891191840171814, + "num_tokens": 825596404.0, + "step": 21262 + }, + { + "epoch": 2.7048721536700167, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.940816879272461, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9090624451637268, + "num_tokens": 825631492.0, + "step": 21263 + }, + { + "epoch": 2.704999363948607, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.005578994750977, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8907286524772644, + "num_tokens": 825665464.0, + "step": 21264 + }, + { + "epoch": 2.7051265742271977, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.956931114196777, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.9071637392044067, + "num_tokens": 825704140.0, + "step": 21265 + }, + { + "epoch": 2.7052537845057882, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.897378921508789, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8880711793899536, + "num_tokens": 825745016.0, + "step": 21266 + }, + { + "epoch": 2.7053809947843783, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.863561630249023, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.909410834312439, + "num_tokens": 825785013.0, + "step": 21267 + }, + { + "epoch": 2.7055082050629693, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.85673999786377, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8944590091705322, + "num_tokens": 825827733.0, + "step": 21268 + }, + { + "epoch": 2.7056354153415594, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.005043983459473, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8904277086257935, + "num_tokens": 825870060.0, + "step": 21269 + }, + { + "epoch": 2.7057626256201504, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.109561920166016, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8804381489753723, + "num_tokens": 825902683.0, + "step": 21270 + }, + { + "epoch": 2.7058898358987404, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.960973739624023, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8997169733047485, + "num_tokens": 825938590.0, + "step": 21271 + }, + { + "epoch": 2.706017046177331, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.952893257141113, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8904630541801453, + "num_tokens": 825974709.0, + "step": 21272 + }, + { + "epoch": 2.7061442564559215, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.96183967590332, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.9009487628936768, + "num_tokens": 826010332.0, + "step": 21273 + }, + { + "epoch": 2.706271466734512, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.953572273254395, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8903442621231079, + "num_tokens": 826050365.0, + "step": 21274 + }, + { + "epoch": 2.7063986770131025, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.938676834106445, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.9046617746353149, + "num_tokens": 826086591.0, + "step": 21275 + }, + { + "epoch": 2.706525887291693, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.937042236328125, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.9040650725364685, + "num_tokens": 826120684.0, + "step": 21276 + }, + { + "epoch": 2.7066530975702836, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.953130722045898, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9112129211425781, + "num_tokens": 826152041.0, + "step": 21277 + }, + { + "epoch": 2.706780307848874, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.945425033569336, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8958431482315063, + "num_tokens": 826188280.0, + "step": 21278 + }, + { + "epoch": 2.7069075181274647, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.02125072479248, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8885475993156433, + "num_tokens": 826225529.0, + "step": 21279 + }, + { + "epoch": 2.707034728406055, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.080406188964844, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8944541215896606, + "num_tokens": 826255574.0, + "step": 21280 + }, + { + "epoch": 2.7071619386846457, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.928948402404785, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8947769999504089, + "num_tokens": 826295852.0, + "step": 21281 + }, + { + "epoch": 2.7072891489632362, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.947996139526367, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8982283473014832, + "num_tokens": 826336804.0, + "step": 21282 + }, + { + "epoch": 2.7074163592418268, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.949063301086426, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8977663516998291, + "num_tokens": 826373052.0, + "step": 21283 + }, + { + "epoch": 2.7075435695204173, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.979333877563477, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.9032766819000244, + "num_tokens": 826408947.0, + "step": 21284 + }, + { + "epoch": 2.707670779799008, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.996150970458984, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8961005210876465, + "num_tokens": 826445246.0, + "step": 21285 + }, + { + "epoch": 2.7077979900775984, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.898462295532227, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8966333866119385, + "num_tokens": 826483920.0, + "step": 21286 + }, + { + "epoch": 2.707925200356189, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.004446029663086, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9054035544395447, + "num_tokens": 826519402.0, + "step": 21287 + }, + { + "epoch": 2.7080524106347794, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.960573196411133, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8979929685592651, + "num_tokens": 826560938.0, + "step": 21288 + }, + { + "epoch": 2.70817962091337, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.910011291503906, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.904532790184021, + "num_tokens": 826601296.0, + "step": 21289 + }, + { + "epoch": 2.7083068311919605, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.939818382263184, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9074150323867798, + "num_tokens": 826641224.0, + "step": 21290 + }, + { + "epoch": 2.708434041470551, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.05472469329834, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8970271944999695, + "num_tokens": 826671718.0, + "step": 21291 + }, + { + "epoch": 2.708561251749141, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.037455558776855, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9076578617095947, + "num_tokens": 826703426.0, + "step": 21292 + }, + { + "epoch": 2.708688462027732, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.883946418762207, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.9006699323654175, + "num_tokens": 826748078.0, + "step": 21293 + }, + { + "epoch": 2.708815672306322, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.95338249206543, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8965991139411926, + "num_tokens": 826786064.0, + "step": 21294 + }, + { + "epoch": 2.708942882584913, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.945720672607422, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8971748352050781, + "num_tokens": 826827367.0, + "step": 21295 + }, + { + "epoch": 2.709070092863503, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.864377975463867, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.9075044393539429, + "num_tokens": 826866301.0, + "step": 21296 + }, + { + "epoch": 2.7091973031420937, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.965653419494629, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.887692928314209, + "num_tokens": 826910005.0, + "step": 21297 + }, + { + "epoch": 2.7093245134206843, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.974201202392578, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8970433473587036, + "num_tokens": 826947687.0, + "step": 21298 + }, + { + "epoch": 2.709451723699275, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.88913345336914, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8965047597885132, + "num_tokens": 826991089.0, + "step": 21299 + }, + { + "epoch": 2.7095789339778653, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.007262229919434, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9049715995788574, + "num_tokens": 827027996.0, + "step": 21300 + }, + { + "epoch": 2.709706144256456, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.958635330200195, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8933231830596924, + "num_tokens": 827068195.0, + "step": 21301 + }, + { + "epoch": 2.7098333545350464, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.939581871032715, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8973039984703064, + "num_tokens": 827107895.0, + "step": 21302 + }, + { + "epoch": 2.709960564813637, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.937042236328125, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8890011310577393, + "num_tokens": 827147320.0, + "step": 21303 + }, + { + "epoch": 2.7100877750922274, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.892111778259277, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9002224206924438, + "num_tokens": 827190238.0, + "step": 21304 + }, + { + "epoch": 2.710214985370818, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.004303932189941, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.889319896697998, + "num_tokens": 827226013.0, + "step": 21305 + }, + { + "epoch": 2.7103421956494085, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.956792831420898, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9150084257125854, + "num_tokens": 827256585.0, + "step": 21306 + }, + { + "epoch": 2.710469405927999, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.956648826599121, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8960594534873962, + "num_tokens": 827292934.0, + "step": 21307 + }, + { + "epoch": 2.7105966162065895, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.924275398254395, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.9035822749137878, + "num_tokens": 827334816.0, + "step": 21308 + }, + { + "epoch": 2.71072382648518, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.82870864868164, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9136797785758972, + "num_tokens": 827369820.0, + "step": 21309 + }, + { + "epoch": 2.7108510367637706, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.968832969665527, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9057426452636719, + "num_tokens": 827412224.0, + "step": 21310 + }, + { + "epoch": 2.710978247042361, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.903128623962402, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8997050523757935, + "num_tokens": 827461313.0, + "step": 21311 + }, + { + "epoch": 2.7111054573209517, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.911697387695312, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.9042251706123352, + "num_tokens": 827500957.0, + "step": 21312 + }, + { + "epoch": 2.711232667599542, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.00076675415039, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8901727199554443, + "num_tokens": 827538610.0, + "step": 21313 + }, + { + "epoch": 2.7113598778781327, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.83995246887207, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9091222286224365, + "num_tokens": 827578845.0, + "step": 21314 + }, + { + "epoch": 2.711487088156723, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.880928993225098, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.9052244424819946, + "num_tokens": 827614202.0, + "step": 21315 + }, + { + "epoch": 2.7116142984353138, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.88241195678711, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8982162475585938, + "num_tokens": 827655156.0, + "step": 21316 + }, + { + "epoch": 2.711741508713904, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.96070671081543, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.9022166728973389, + "num_tokens": 827691491.0, + "step": 21317 + }, + { + "epoch": 2.711868718992495, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.935210227966309, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8964589834213257, + "num_tokens": 827731533.0, + "step": 21318 + }, + { + "epoch": 2.711995929271085, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2040138244628906e-05, + "grad_norm": 9.888219833374023, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8941794633865356, + "num_tokens": 827769070.0, + "step": 21319 + }, + { + "epoch": 2.712123139549676, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.35903263092041, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8858681321144104, + "num_tokens": 827813956.0, + "step": 21320 + }, + { + "epoch": 2.712250349828266, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.859213829040527, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.9012371301651001, + "num_tokens": 827859847.0, + "step": 21321 + }, + { + "epoch": 2.7123775601068565, + "ewc_loss": 0.021484375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.892560958862305, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8957387804985046, + "num_tokens": 827897019.0, + "step": 21322 + }, + { + "epoch": 2.712504770385447, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.987397193908691, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8862641453742981, + "num_tokens": 827937432.0, + "step": 21323 + }, + { + "epoch": 2.7126319806640375, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.867291450500488, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8928216099739075, + "num_tokens": 827981971.0, + "step": 21324 + }, + { + "epoch": 2.712759190942628, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.983745574951172, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.9031132459640503, + "num_tokens": 828014678.0, + "step": 21325 + }, + { + "epoch": 2.7128864012212186, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.998282432556152, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.89996737241745, + "num_tokens": 828050179.0, + "step": 21326 + }, + { + "epoch": 2.713013611499809, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.975137710571289, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8876010179519653, + "num_tokens": 828088407.0, + "step": 21327 + }, + { + "epoch": 2.7131408217783997, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.031272888183594, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.902732789516449, + "num_tokens": 828124091.0, + "step": 21328 + }, + { + "epoch": 2.71326803205699, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.004888534545898, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8985798358917236, + "num_tokens": 828160180.0, + "step": 21329 + }, + { + "epoch": 2.7133952423355807, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.908123016357422, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8949481248855591, + "num_tokens": 828201290.0, + "step": 21330 + }, + { + "epoch": 2.7135224526141712, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.97929573059082, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8962968587875366, + "num_tokens": 828241827.0, + "step": 21331 + }, + { + "epoch": 2.7136496628927618, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.919772148132324, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.9023469686508179, + "num_tokens": 828281811.0, + "step": 21332 + }, + { + "epoch": 2.7137768731713523, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.911053657531738, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9136426448822021, + "num_tokens": 828320121.0, + "step": 21333 + }, + { + "epoch": 2.713904083449943, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.95978832244873, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8919283747673035, + "num_tokens": 828361201.0, + "step": 21334 + }, + { + "epoch": 2.7140312937285334, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.973759651184082, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9099377989768982, + "num_tokens": 828397349.0, + "step": 21335 + }, + { + "epoch": 2.714158504007124, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.029483795166016, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.9002999663352966, + "num_tokens": 828432820.0, + "step": 21336 + }, + { + "epoch": 2.7142857142857144, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.987972259521484, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8923547267913818, + "num_tokens": 828470653.0, + "step": 21337 + }, + { + "epoch": 2.714412924564305, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.993274688720703, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9025163054466248, + "num_tokens": 828505945.0, + "step": 21338 + }, + { + "epoch": 2.7145401348428955, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.940674781799316, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9044032096862793, + "num_tokens": 828542920.0, + "step": 21339 + }, + { + "epoch": 2.7146673451214856, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.879533767700195, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9061269164085388, + "num_tokens": 828580413.0, + "step": 21340 + }, + { + "epoch": 2.7147945554000765, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.064396858215332, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.9011560082435608, + "num_tokens": 828617907.0, + "step": 21341 + }, + { + "epoch": 2.7149217656786666, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.93154525756836, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8844529390335083, + "num_tokens": 828663346.0, + "step": 21342 + }, + { + "epoch": 2.7150489759572576, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.037266731262207, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8945508003234863, + "num_tokens": 828703988.0, + "step": 21343 + }, + { + "epoch": 2.7151761862358477, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.97624397277832, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8947166204452515, + "num_tokens": 828743276.0, + "step": 21344 + }, + { + "epoch": 2.7153033965144386, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.94682502746582, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8989686965942383, + "num_tokens": 828786147.0, + "step": 21345 + }, + { + "epoch": 2.7154306067930287, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.971827507019043, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8938577175140381, + "num_tokens": 828827845.0, + "step": 21346 + }, + { + "epoch": 2.7155578170716193, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.012282371520996, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9038630127906799, + "num_tokens": 828863107.0, + "step": 21347 + }, + { + "epoch": 2.71568502735021, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.011999130249023, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.9019691944122314, + "num_tokens": 828897508.0, + "step": 21348 + }, + { + "epoch": 2.7158122376288003, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.966485023498535, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8988813161849976, + "num_tokens": 828937022.0, + "step": 21349 + }, + { + "epoch": 2.715939447907391, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.983744621276855, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9077484607696533, + "num_tokens": 828973097.0, + "step": 21350 + }, + { + "epoch": 2.7160666581859814, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.965008735656738, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.9029563665390015, + "num_tokens": 829006545.0, + "step": 21351 + }, + { + "epoch": 2.716193868464572, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.163084030151367, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8999415040016174, + "num_tokens": 829037649.0, + "step": 21352 + }, + { + "epoch": 2.7163210787431624, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.992298126220703, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8819564580917358, + "num_tokens": 829078599.0, + "step": 21353 + }, + { + "epoch": 2.716448289021753, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.489059448242188e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.0037260055542, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.9032732248306274, + "num_tokens": 829119821.0, + "step": 21354 + }, + { + "epoch": 2.7165754993003435, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.958353042602539, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8911745548248291, + "num_tokens": 829156692.0, + "step": 21355 + }, + { + "epoch": 2.716702709578934, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.963451385498047, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9029884338378906, + "num_tokens": 829198266.0, + "step": 21356 + }, + { + "epoch": 2.7168299198575245, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.994392395019531, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8974069952964783, + "num_tokens": 829230492.0, + "step": 21357 + }, + { + "epoch": 2.716957130136115, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.882304191589355, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9039878845214844, + "num_tokens": 829268619.0, + "step": 21358 + }, + { + "epoch": 2.7170843404147056, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.925731658935547, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8976576328277588, + "num_tokens": 829308445.0, + "step": 21359 + }, + { + "epoch": 2.717211550693296, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.9153413772583, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9057208299636841, + "num_tokens": 829348340.0, + "step": 21360 + }, + { + "epoch": 2.7173387609718866, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.062765121459961, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.887014627456665, + "num_tokens": 829388217.0, + "step": 21361 + }, + { + "epoch": 2.717465971250477, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.939810752868652, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8990781307220459, + "num_tokens": 829424135.0, + "step": 21362 + }, + { + "epoch": 2.7175931815290677, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.927878379821777, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9143249988555908, + "num_tokens": 829459360.0, + "step": 21363 + }, + { + "epoch": 2.7177203918076582, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.972615242004395, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.902771532535553, + "num_tokens": 829497832.0, + "step": 21364 + }, + { + "epoch": 2.7178476020862483, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.92342758178711, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8981119394302368, + "num_tokens": 829539000.0, + "step": 21365 + }, + { + "epoch": 2.7179748123648393, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.067353248596191, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.9000621438026428, + "num_tokens": 829572035.0, + "step": 21366 + }, + { + "epoch": 2.7181020226434294, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.963470458984375, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9140044450759888, + "num_tokens": 829604087.0, + "step": 21367 + }, + { + "epoch": 2.7182292329220203, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.129730224609375, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9022139310836792, + "num_tokens": 829643012.0, + "step": 21368 + }, + { + "epoch": 2.7183564432006104, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.906760215759277, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8994561433792114, + "num_tokens": 829683488.0, + "step": 21369 + }, + { + "epoch": 2.718483653479201, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.0940523147583, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8765223026275635, + "num_tokens": 829722072.0, + "step": 21370 + }, + { + "epoch": 2.7186108637577915, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.955291748046875, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8897855281829834, + "num_tokens": 829759265.0, + "step": 21371 + }, + { + "epoch": 2.718738074036382, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.004443168640137, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.9002479314804077, + "num_tokens": 829797770.0, + "step": 21372 + }, + { + "epoch": 2.7188652843149725, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.028571128845215, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.9009342789649963, + "num_tokens": 829834484.0, + "step": 21373 + }, + { + "epoch": 2.718992494593563, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.966543197631836, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8835822343826294, + "num_tokens": 829873747.0, + "step": 21374 + }, + { + "epoch": 2.7191197048721536, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.004138946533203, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8846932053565979, + "num_tokens": 829912501.0, + "step": 21375 + }, + { + "epoch": 2.719246915150744, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.943552017211914, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8869742751121521, + "num_tokens": 829956989.0, + "step": 21376 + }, + { + "epoch": 2.7193741254293347, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.148911476135254, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.89207923412323, + "num_tokens": 829990206.0, + "step": 21377 + }, + { + "epoch": 2.719501335707925, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.918811798095703, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9115184545516968, + "num_tokens": 830026290.0, + "step": 21378 + }, + { + "epoch": 2.7196285459865157, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.864102363586426, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8883119225502014, + "num_tokens": 830070209.0, + "step": 21379 + }, + { + "epoch": 2.7197557562651062, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.096549034118652, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.9027858972549438, + "num_tokens": 830102767.0, + "step": 21380 + }, + { + "epoch": 2.7198829665436968, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.823427200317383, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8991163969039917, + "num_tokens": 830149449.0, + "step": 21381 + }, + { + "epoch": 2.7200101768222873, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.931453704833984, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.9004195928573608, + "num_tokens": 830186619.0, + "step": 21382 + }, + { + "epoch": 2.720137387100878, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.87194538116455, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.9008280634880066, + "num_tokens": 830230917.0, + "step": 21383 + }, + { + "epoch": 2.7202645973794684, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.00969123840332, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8958849906921387, + "num_tokens": 830269472.0, + "step": 21384 + }, + { + "epoch": 2.720391807658059, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.022637367248535, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8847641944885254, + "num_tokens": 830306324.0, + "step": 21385 + }, + { + "epoch": 2.7205190179366494, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.9219388961792, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.9058221578598022, + "num_tokens": 830341180.0, + "step": 21386 + }, + { + "epoch": 2.72064622821524, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.001070022583008, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9016134142875671, + "num_tokens": 830375919.0, + "step": 21387 + }, + { + "epoch": 2.7207734384938305, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.984036445617676, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.894318699836731, + "num_tokens": 830414048.0, + "step": 21388 + }, + { + "epoch": 2.720900648772421, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.91848373413086, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.9069323539733887, + "num_tokens": 830446661.0, + "step": 21389 + }, + { + "epoch": 2.721027859051011, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.255417823791504, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8826414346694946, + "num_tokens": 830475255.0, + "step": 21390 + }, + { + "epoch": 2.721155069329602, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.806741714477539, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9095685482025146, + "num_tokens": 830518173.0, + "step": 21391 + }, + { + "epoch": 2.721282279608192, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.050795555114746, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8962693214416504, + "num_tokens": 830553523.0, + "step": 21392 + }, + { + "epoch": 2.721409489886783, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.952178001403809, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9092670679092407, + "num_tokens": 830590218.0, + "step": 21393 + }, + { + "epoch": 2.721536700165373, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.917319297790527, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.9026681184768677, + "num_tokens": 830629011.0, + "step": 21394 + }, + { + "epoch": 2.7216639104439637, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.031980514526367, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8958879709243774, + "num_tokens": 830665601.0, + "step": 21395 + }, + { + "epoch": 2.7217911207225542, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.053315162658691, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.902769923210144, + "num_tokens": 830698010.0, + "step": 21396 + }, + { + "epoch": 2.7219183310011448, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.928284645080566, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9110997915267944, + "num_tokens": 830733174.0, + "step": 21397 + }, + { + "epoch": 2.7220455412797353, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.96464729309082, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8891313672065735, + "num_tokens": 830774307.0, + "step": 21398 + }, + { + "epoch": 2.722172751558326, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.88137149810791, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9051113128662109, + "num_tokens": 830812248.0, + "step": 21399 + }, + { + "epoch": 2.7222999618369164, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.981046676635742, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.9059741497039795, + "num_tokens": 830849502.0, + "step": 21400 + }, + { + "epoch": 2.722427172115507, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.905282020568848, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8999738693237305, + "num_tokens": 830889941.0, + "step": 21401 + }, + { + "epoch": 2.7225543823940974, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.015233039855957, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.9019428491592407, + "num_tokens": 830922929.0, + "step": 21402 + }, + { + "epoch": 2.722681592672688, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.863795280456543, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.9047049283981323, + "num_tokens": 830963203.0, + "step": 21403 + }, + { + "epoch": 2.7228088029512785, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.96642780303955, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.900694727897644, + "num_tokens": 831001670.0, + "step": 21404 + }, + { + "epoch": 2.722936013229869, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.893229484558105, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9102692008018494, + "num_tokens": 831044120.0, + "step": 21405 + }, + { + "epoch": 2.7230632235084595, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.9386625289917, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8949390649795532, + "num_tokens": 831086833.0, + "step": 21406 + }, + { + "epoch": 2.72319043378705, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.924763679504395, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8916736841201782, + "num_tokens": 831132645.0, + "step": 21407 + }, + { + "epoch": 2.7233176440656406, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.990850448608398, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9030828475952148, + "num_tokens": 831163059.0, + "step": 21408 + }, + { + "epoch": 2.723444854344231, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.025171279907227, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8877987265586853, + "num_tokens": 831203399.0, + "step": 21409 + }, + { + "epoch": 2.7235720646228216, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.948081970214844, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.893412709236145, + "num_tokens": 831245527.0, + "step": 21410 + }, + { + "epoch": 2.723699274901412, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.894950866699219, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.885521650314331, + "num_tokens": 831293430.0, + "step": 21411 + }, + { + "epoch": 2.7238264851800027, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.970446586608887, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8975673913955688, + "num_tokens": 831330692.0, + "step": 21412 + }, + { + "epoch": 2.723953695458593, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.060945510864258, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8906733989715576, + "num_tokens": 831364602.0, + "step": 21413 + }, + { + "epoch": 2.7240809057371838, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.862907409667969, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8884991407394409, + "num_tokens": 831405684.0, + "step": 21414 + }, + { + "epoch": 2.724208116015774, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.870990753173828, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8983628749847412, + "num_tokens": 831447398.0, + "step": 21415 + }, + { + "epoch": 2.724335326294365, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.009513854980469, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8927462697029114, + "num_tokens": 831480708.0, + "step": 21416 + }, + { + "epoch": 2.724462536572955, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.94060230255127, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8898529410362244, + "num_tokens": 831520892.0, + "step": 21417 + }, + { + "epoch": 2.724589746851546, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.894761085510254, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.916290283203125, + "num_tokens": 831561112.0, + "step": 21418 + }, + { + "epoch": 2.724716957130136, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.065500259399414, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8938161730766296, + "num_tokens": 831592293.0, + "step": 21419 + }, + { + "epoch": 2.7248441674087265, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.956204414367676, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8930385112762451, + "num_tokens": 831636469.0, + "step": 21420 + }, + { + "epoch": 2.724971377687317, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.940591812133789, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8965985774993896, + "num_tokens": 831672174.0, + "step": 21421 + }, + { + "epoch": 2.7250985879659075, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.931276321411133, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8995348215103149, + "num_tokens": 831714172.0, + "step": 21422 + }, + { + "epoch": 2.725225798244498, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.998444557189941, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.903420627117157, + "num_tokens": 831749040.0, + "step": 21423 + }, + { + "epoch": 2.7253530085230886, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.848974227905273, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9158302545547485, + "num_tokens": 831790323.0, + "step": 21424 + }, + { + "epoch": 2.725480218801679, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.038737297058105, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.9041333198547363, + "num_tokens": 831823468.0, + "step": 21425 + }, + { + "epoch": 2.7256074290802697, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.949515342712402, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.9025312662124634, + "num_tokens": 831865315.0, + "step": 21426 + }, + { + "epoch": 2.72573463935886, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.948104858398438, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9058488011360168, + "num_tokens": 831903884.0, + "step": 21427 + }, + { + "epoch": 2.7258618496374507, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.089821815490723, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8937364816665649, + "num_tokens": 831937529.0, + "step": 21428 + }, + { + "epoch": 2.7259890599160412, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.04223918914795, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8869068622589111, + "num_tokens": 831972816.0, + "step": 21429 + }, + { + "epoch": 2.7261162701946318, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.806356430053711, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.905394434928894, + "num_tokens": 832013063.0, + "step": 21430 + }, + { + "epoch": 2.7262434804732223, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.93026065826416, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8902803659439087, + "num_tokens": 832055658.0, + "step": 21431 + }, + { + "epoch": 2.726370690751813, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.885626792907715, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.9020923376083374, + "num_tokens": 832093198.0, + "step": 21432 + }, + { + "epoch": 2.7264979010304033, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.882664680480957, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.9000298380851746, + "num_tokens": 832136133.0, + "step": 21433 + }, + { + "epoch": 2.726625111308994, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.071109771728516, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8978400230407715, + "num_tokens": 832170030.0, + "step": 21434 + }, + { + "epoch": 2.7267523215875844, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.879339218139648, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.9044042229652405, + "num_tokens": 832211609.0, + "step": 21435 + }, + { + "epoch": 2.726879531866175, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.87342357635498, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9089006185531616, + "num_tokens": 832249674.0, + "step": 21436 + }, + { + "epoch": 2.7270067421447655, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.915104866027832, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9046317934989929, + "num_tokens": 832285131.0, + "step": 21437 + }, + { + "epoch": 2.7271339524233555, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.950488090515137, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.9009047746658325, + "num_tokens": 832323963.0, + "step": 21438 + }, + { + "epoch": 2.7272611627019465, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.021921157836914, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8977658748626709, + "num_tokens": 832356114.0, + "step": 21439 + }, + { + "epoch": 2.7273883729805366, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.875929832458496, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.9000869989395142, + "num_tokens": 832398246.0, + "step": 21440 + }, + { + "epoch": 2.7275155832591276, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.8836030960083, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.902271032333374, + "num_tokens": 832443745.0, + "step": 21441 + }, + { + "epoch": 2.7276427935377177, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.001716613769531, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8936710953712463, + "num_tokens": 832478919.0, + "step": 21442 + }, + { + "epoch": 2.7277700038163086, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.949811935424805, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.9002282619476318, + "num_tokens": 832515187.0, + "step": 21443 + }, + { + "epoch": 2.7278972140948987, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.878430366516113, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.9010770916938782, + "num_tokens": 832560315.0, + "step": 21444 + }, + { + "epoch": 2.7280244243734892, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.036511421203613, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.9000763297080994, + "num_tokens": 832593907.0, + "step": 21445 + }, + { + "epoch": 2.7281516346520798, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.895674705505371, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.9028623700141907, + "num_tokens": 832634115.0, + "step": 21446 + }, + { + "epoch": 2.7282788449306703, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.997771263122559, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8975197076797485, + "num_tokens": 832669706.0, + "step": 21447 + }, + { + "epoch": 2.728406055209261, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.002557754516602, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8859394788742065, + "num_tokens": 832710764.0, + "step": 21448 + }, + { + "epoch": 2.7285332654878514, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.027652740478516, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.9010777473449707, + "num_tokens": 832745451.0, + "step": 21449 + }, + { + "epoch": 2.728660475766442, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.009636878967285, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.9023236632347107, + "num_tokens": 832783443.0, + "step": 21450 + }, + { + "epoch": 2.7287876860450324, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.93707275390625, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8970947861671448, + "num_tokens": 832825523.0, + "step": 21451 + }, + { + "epoch": 2.728914896323623, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.972199440002441, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9031463861465454, + "num_tokens": 832859329.0, + "step": 21452 + }, + { + "epoch": 2.7290421066022135, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.90236759185791, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8942849040031433, + "num_tokens": 832904942.0, + "step": 21453 + }, + { + "epoch": 2.729169316880804, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.909918785095215, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.904798686504364, + "num_tokens": 832947134.0, + "step": 21454 + }, + { + "epoch": 2.7292965271593945, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.025490760803223, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8982290029525757, + "num_tokens": 832984124.0, + "step": 21455 + }, + { + "epoch": 2.729423737437985, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.031438827514648, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8968282341957092, + "num_tokens": 833017850.0, + "step": 21456 + }, + { + "epoch": 2.7295509477165756, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.920047760009766, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9039803743362427, + "num_tokens": 833056462.0, + "step": 21457 + }, + { + "epoch": 2.729678157995166, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.90001106262207, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9030828475952148, + "num_tokens": 833097212.0, + "step": 21458 + }, + { + "epoch": 2.7298053682737566, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.967069625854492, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8996912837028503, + "num_tokens": 833138075.0, + "step": 21459 + }, + { + "epoch": 2.729932578552347, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.940667152404785, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8997692465782166, + "num_tokens": 833176843.0, + "step": 21460 + }, + { + "epoch": 2.7300597888309377, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.950343132019043, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8967342376708984, + "num_tokens": 833212089.0, + "step": 21461 + }, + { + "epoch": 2.7301869991095282, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.9061861038208, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.9047499299049377, + "num_tokens": 833252189.0, + "step": 21462 + }, + { + "epoch": 2.7303142093881183, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.903922080993652, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9031655192375183, + "num_tokens": 833292645.0, + "step": 21463 + }, + { + "epoch": 2.7304414196667093, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.009961128234863, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.9029527902603149, + "num_tokens": 833330796.0, + "step": 21464 + }, + { + "epoch": 2.7305686299452994, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.855263710021973, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.9020784497261047, + "num_tokens": 833369475.0, + "step": 21465 + }, + { + "epoch": 2.7306958402238903, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.899564743041992, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8989526629447937, + "num_tokens": 833411577.0, + "step": 21466 + }, + { + "epoch": 2.7308230505024804, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.927825927734375, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8925372362136841, + "num_tokens": 833455387.0, + "step": 21467 + }, + { + "epoch": 2.730950260781071, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.93242073059082, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8994982838630676, + "num_tokens": 833494362.0, + "step": 21468 + }, + { + "epoch": 2.7310774710596615, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.877504348754883, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9071946740150452, + "num_tokens": 833535255.0, + "step": 21469 + }, + { + "epoch": 2.731204681338252, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.94876766204834, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8987043499946594, + "num_tokens": 833575323.0, + "step": 21470 + }, + { + "epoch": 2.7313318916168425, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.96043872833252, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.898358941078186, + "num_tokens": 833618736.0, + "step": 21471 + }, + { + "epoch": 2.731459101895433, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.933937072753906, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8952475786209106, + "num_tokens": 833657238.0, + "step": 21472 + }, + { + "epoch": 2.7315863121740236, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.92670726776123, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8984339833259583, + "num_tokens": 833698138.0, + "step": 21473 + }, + { + "epoch": 2.731713522452614, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.002310752868652, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9049049019813538, + "num_tokens": 833732236.0, + "step": 21474 + }, + { + "epoch": 2.7318407327312046, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.940817832946777, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8992195129394531, + "num_tokens": 833773944.0, + "step": 21475 + }, + { + "epoch": 2.731967943009795, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.889019966125488, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.903575599193573, + "num_tokens": 833809527.0, + "step": 21476 + }, + { + "epoch": 2.7320951532883857, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.999116897583008, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8939422965049744, + "num_tokens": 833850200.0, + "step": 21477 + }, + { + "epoch": 2.7322223635669762, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.912446022033691, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9061187505722046, + "num_tokens": 833886633.0, + "step": 21478 + }, + { + "epoch": 2.7323495738455668, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.925481796264648, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.890843391418457, + "num_tokens": 833930607.0, + "step": 21479 + }, + { + "epoch": 2.7324767841241573, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.960410118103027, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8954711556434631, + "num_tokens": 833970790.0, + "step": 21480 + }, + { + "epoch": 2.732603994402748, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.956825256347656, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8919271230697632, + "num_tokens": 834009190.0, + "step": 21481 + }, + { + "epoch": 2.7327312046813383, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.9260892868042, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8935157060623169, + "num_tokens": 834052894.0, + "step": 21482 + }, + { + "epoch": 2.732858414959929, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.03139591217041, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8941856026649475, + "num_tokens": 834088334.0, + "step": 21483 + }, + { + "epoch": 2.7329856252385194, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.886739730834961, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8974212408065796, + "num_tokens": 834130822.0, + "step": 21484 + }, + { + "epoch": 2.73311283551711, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.91762924194336, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8939065337181091, + "num_tokens": 834167410.0, + "step": 21485 + }, + { + "epoch": 2.7332400457957005, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.97783088684082, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.9040656089782715, + "num_tokens": 834201083.0, + "step": 21486 + }, + { + "epoch": 2.733367256074291, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.87365436553955, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9046114683151245, + "num_tokens": 834238764.0, + "step": 21487 + }, + { + "epoch": 2.733494466352881, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.929784774780273, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8923035860061646, + "num_tokens": 834279500.0, + "step": 21488 + }, + { + "epoch": 2.733621676631472, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.99862289428711, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.9010823369026184, + "num_tokens": 834315568.0, + "step": 21489 + }, + { + "epoch": 2.733748886910062, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.969253540039062, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8851177096366882, + "num_tokens": 834357685.0, + "step": 21490 + }, + { + "epoch": 2.733876097188653, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.884573936462402, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9031798839569092, + "num_tokens": 834395187.0, + "step": 21491 + }, + { + "epoch": 2.734003307467243, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.028308868408203, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8967150449752808, + "num_tokens": 834436468.0, + "step": 21492 + }, + { + "epoch": 2.7341305177458337, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.037535667419434, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8793051838874817, + "num_tokens": 834473018.0, + "step": 21493 + }, + { + "epoch": 2.7342577280244242, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.017596244812012, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8908215761184692, + "num_tokens": 834509275.0, + "step": 21494 + }, + { + "epoch": 2.7343849383030148, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.881247520446777, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8935967683792114, + "num_tokens": 834549696.0, + "step": 21495 + }, + { + "epoch": 2.7345121485816053, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.866926193237305, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.9025517702102661, + "num_tokens": 834587674.0, + "step": 21496 + }, + { + "epoch": 2.734639358860196, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.826911926269531, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8996818661689758, + "num_tokens": 834631157.0, + "step": 21497 + }, + { + "epoch": 2.7347665691387864, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.007280349731445, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8979485630989075, + "num_tokens": 834668366.0, + "step": 21498 + }, + { + "epoch": 2.734893779417377, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.872294425964355, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9156279563903809, + "num_tokens": 834703886.0, + "step": 21499 + }, + { + "epoch": 2.7350209896959674, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.883309364318848, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8956458568572998, + "num_tokens": 834750462.0, + "step": 21500 + }, + { + "epoch": 2.735148199974558, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.034229278564453, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8877021670341492, + "num_tokens": 834787387.0, + "step": 21501 + }, + { + "epoch": 2.7352754102531485, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.014657020568848, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8957401514053345, + "num_tokens": 834823806.0, + "step": 21502 + }, + { + "epoch": 2.735402620531739, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.014352798461914, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8947337865829468, + "num_tokens": 834857598.0, + "step": 21503 + }, + { + "epoch": 2.7355298308103295, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.945182800292969, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8937815427780151, + "num_tokens": 834896002.0, + "step": 21504 + }, + { + "epoch": 2.73565704108892, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.922629356384277, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.9050498008728027, + "num_tokens": 834932701.0, + "step": 21505 + }, + { + "epoch": 2.7357842513675106, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.063591957092285, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8916381001472473, + "num_tokens": 834968471.0, + "step": 21506 + }, + { + "epoch": 2.735911461646101, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.963817596435547, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8987066149711609, + "num_tokens": 835007057.0, + "step": 21507 + }, + { + "epoch": 2.7360386719246916, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.947094917297363, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.888360857963562, + "num_tokens": 835045748.0, + "step": 21508 + }, + { + "epoch": 2.736165882203282, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.921233177185059, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9083144664764404, + "num_tokens": 835083425.0, + "step": 21509 + }, + { + "epoch": 2.7362930924818727, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.98694896697998, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.905095100402832, + "num_tokens": 835124032.0, + "step": 21510 + }, + { + "epoch": 2.7364203027604628, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.944124221801758, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9074438810348511, + "num_tokens": 835162802.0, + "step": 21511 + }, + { + "epoch": 2.7365475130390537, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.968567848205566, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.9048199653625488, + "num_tokens": 835200564.0, + "step": 21512 + }, + { + "epoch": 2.736674723317644, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.021596908569336, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8869161605834961, + "num_tokens": 835239925.0, + "step": 21513 + }, + { + "epoch": 2.736801933596235, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.007688522338867, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8956544399261475, + "num_tokens": 835280611.0, + "step": 21514 + }, + { + "epoch": 2.736929143874825, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.967641830444336, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9048280119895935, + "num_tokens": 835321654.0, + "step": 21515 + }, + { + "epoch": 2.737056354153416, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.057154655456543, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.900497317314148, + "num_tokens": 835358339.0, + "step": 21516 + }, + { + "epoch": 2.737183564432006, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.983283042907715, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8952295780181885, + "num_tokens": 835391844.0, + "step": 21517 + }, + { + "epoch": 2.7373107747105965, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.075520515441895, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.901585578918457, + "num_tokens": 835429282.0, + "step": 21518 + }, + { + "epoch": 2.737437984989187, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.922348976135254, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8923328518867493, + "num_tokens": 835468963.0, + "step": 21519 + }, + { + "epoch": 2.7375651952677775, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.91164779663086, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8948542475700378, + "num_tokens": 835511165.0, + "step": 21520 + }, + { + "epoch": 2.737692405546368, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.89882755279541, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.9044920802116394, + "num_tokens": 835555986.0, + "step": 21521 + }, + { + "epoch": 2.7378196158249586, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.957916259765625, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9046093225479126, + "num_tokens": 835596763.0, + "step": 21522 + }, + { + "epoch": 2.737946826103549, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.98276424407959, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9036109447479248, + "num_tokens": 835631626.0, + "step": 21523 + }, + { + "epoch": 2.7380740363821396, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.001727104187012, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8980758786201477, + "num_tokens": 835670854.0, + "step": 21524 + }, + { + "epoch": 2.73820124666073, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.5367431640625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.968978881835938, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.9026458263397217, + "num_tokens": 835710760.0, + "step": 21525 + }, + { + "epoch": 2.7383284569393207, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.96044921875, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8885409832000732, + "num_tokens": 835751589.0, + "step": 21526 + }, + { + "epoch": 2.7384556672179112, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.974564552307129, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8969136476516724, + "num_tokens": 835787446.0, + "step": 21527 + }, + { + "epoch": 2.7385828774965018, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.902445793151855, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8973174095153809, + "num_tokens": 835826386.0, + "step": 21528 + }, + { + "epoch": 2.7387100877750923, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.85423469543457, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9081375598907471, + "num_tokens": 835866257.0, + "step": 21529 + }, + { + "epoch": 2.738837298053683, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.927940368652344, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.895606279373169, + "num_tokens": 835913187.0, + "step": 21530 + }, + { + "epoch": 2.7389645083322733, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.886872291564941, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8964625597000122, + "num_tokens": 835959329.0, + "step": 21531 + }, + { + "epoch": 2.739091718610864, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.91275691986084, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.891550600528717, + "num_tokens": 836008667.0, + "step": 21532 + }, + { + "epoch": 2.7392189288894544, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.946759223937988, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8910958766937256, + "num_tokens": 836048903.0, + "step": 21533 + }, + { + "epoch": 2.739346139168045, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.905078887939453, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.9068356156349182, + "num_tokens": 836089863.0, + "step": 21534 + }, + { + "epoch": 2.7394733494466355, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.925042152404785, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8821525573730469, + "num_tokens": 836137011.0, + "step": 21535 + }, + { + "epoch": 2.7396005597252255, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.074861526489258, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8988023996353149, + "num_tokens": 836167366.0, + "step": 21536 + }, + { + "epoch": 2.7397277700038165, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.917908668518066, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8910778760910034, + "num_tokens": 836207451.0, + "step": 21537 + }, + { + "epoch": 2.7398549802824066, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.96861457824707, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.9028858542442322, + "num_tokens": 836241680.0, + "step": 21538 + }, + { + "epoch": 2.7399821905609976, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.947057723999023, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8975498676300049, + "num_tokens": 836279836.0, + "step": 21539 + }, + { + "epoch": 2.7401094008395877, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.966273307800293, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8892505168914795, + "num_tokens": 836320041.0, + "step": 21540 + }, + { + "epoch": 2.7402366111181786, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.927129745483398, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.9057348966598511, + "num_tokens": 836360032.0, + "step": 21541 + }, + { + "epoch": 2.7403638213967687, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 10.061145782470703, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8896677494049072, + "num_tokens": 836399579.0, + "step": 21542 + }, + { + "epoch": 2.7404910316753592, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.868645668029785, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.9071553349494934, + "num_tokens": 836445171.0, + "step": 21543 + }, + { + "epoch": 2.7406182419539498, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.032200813293457, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.9001227617263794, + "num_tokens": 836481825.0, + "step": 21544 + }, + { + "epoch": 2.7407454522325403, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.008170127868652, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8872839212417603, + "num_tokens": 836524713.0, + "step": 21545 + }, + { + "epoch": 2.740872662511131, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.978492736816406, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8939658403396606, + "num_tokens": 836560819.0, + "step": 21546 + }, + { + "epoch": 2.7409998727897213, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.959637641906738, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8925233483314514, + "num_tokens": 836601408.0, + "step": 21547 + }, + { + "epoch": 2.741127083068312, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.94234848022461, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.895289957523346, + "num_tokens": 836639527.0, + "step": 21548 + }, + { + "epoch": 2.7412542933469024, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.985054969787598, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.9072171449661255, + "num_tokens": 836674732.0, + "step": 21549 + }, + { + "epoch": 2.741381503625493, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.93264102935791, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8904036283493042, + "num_tokens": 836718984.0, + "step": 21550 + }, + { + "epoch": 2.7415087139040835, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.9926118850708, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8922885060310364, + "num_tokens": 836755413.0, + "step": 21551 + }, + { + "epoch": 2.741635924182674, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.955744743347168, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.9016326665878296, + "num_tokens": 836796857.0, + "step": 21552 + }, + { + "epoch": 2.7417631344612645, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.943347930908203, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.9054659008979797, + "num_tokens": 836838030.0, + "step": 21553 + }, + { + "epoch": 2.741890344739855, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.937532424926758, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8995050191879272, + "num_tokens": 836882633.0, + "step": 21554 + }, + { + "epoch": 2.7420175550184456, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.000417709350586, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8940550088882446, + "num_tokens": 836919514.0, + "step": 21555 + }, + { + "epoch": 2.742144765297036, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.946699142456055, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9100320339202881, + "num_tokens": 836955845.0, + "step": 21556 + }, + { + "epoch": 2.7422719755756266, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.960007667541504, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.882331132888794, + "num_tokens": 836998450.0, + "step": 21557 + }, + { + "epoch": 2.742399185854217, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.959160804748535, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9039833545684814, + "num_tokens": 837032470.0, + "step": 21558 + }, + { + "epoch": 2.7425263961328077, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.965033531188965, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.9069850444793701, + "num_tokens": 837073342.0, + "step": 21559 + }, + { + "epoch": 2.742653606411398, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.970852851867676, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8982197046279907, + "num_tokens": 837113268.0, + "step": 21560 + }, + { + "epoch": 2.7427808166899883, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.983060836791992, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8930507898330688, + "num_tokens": 837151364.0, + "step": 21561 + }, + { + "epoch": 2.7429080269685793, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.02346420288086, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8990080952644348, + "num_tokens": 837185482.0, + "step": 21562 + }, + { + "epoch": 2.7430352372471694, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.993255615234375, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8906234502792358, + "num_tokens": 837226050.0, + "step": 21563 + }, + { + "epoch": 2.7431624475257603, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.902081489562988, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.9025604724884033, + "num_tokens": 837266407.0, + "step": 21564 + }, + { + "epoch": 2.7432896578043504, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.944706916809082, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.9002562761306763, + "num_tokens": 837307219.0, + "step": 21565 + }, + { + "epoch": 2.743416868082941, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.97565746307373, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8960621356964111, + "num_tokens": 837342317.0, + "step": 21566 + }, + { + "epoch": 2.7435440783615315, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.121838569641113, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8970360159873962, + "num_tokens": 837373964.0, + "step": 21567 + }, + { + "epoch": 2.743671288640122, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.957954406738281, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8951581120491028, + "num_tokens": 837411037.0, + "step": 21568 + }, + { + "epoch": 2.7437984989187125, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.890802383422852, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8940959572792053, + "num_tokens": 837453494.0, + "step": 21569 + }, + { + "epoch": 2.743925709197303, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.982847213745117, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8934469223022461, + "num_tokens": 837493759.0, + "step": 21570 + }, + { + "epoch": 2.7440529194758936, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.91628360748291, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8962249755859375, + "num_tokens": 837534935.0, + "step": 21571 + }, + { + "epoch": 2.744180129754484, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.87404727935791, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8988603353500366, + "num_tokens": 837580261.0, + "step": 21572 + }, + { + "epoch": 2.7443073400330746, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.98373031616211, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8821198344230652, + "num_tokens": 837622898.0, + "step": 21573 + }, + { + "epoch": 2.744434550311665, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.00869369506836, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9123969078063965, + "num_tokens": 837658612.0, + "step": 21574 + }, + { + "epoch": 2.7445617605902557, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.946329116821289, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.9001803398132324, + "num_tokens": 837704813.0, + "step": 21575 + }, + { + "epoch": 2.7446889708688462, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.967378616333008, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.89662766456604, + "num_tokens": 837747070.0, + "step": 21576 + }, + { + "epoch": 2.7448161811474368, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.944457054138184, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9064953923225403, + "num_tokens": 837788762.0, + "step": 21577 + }, + { + "epoch": 2.7449433914260273, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.933698654174805, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8962101340293884, + "num_tokens": 837825946.0, + "step": 21578 + }, + { + "epoch": 2.745070601704618, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.126835823059082, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8900792598724365, + "num_tokens": 837859268.0, + "step": 21579 + }, + { + "epoch": 2.7451978119832083, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.929784774780273, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8904104828834534, + "num_tokens": 837904610.0, + "step": 21580 + }, + { + "epoch": 2.745325022261799, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.011908531188965, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8884114623069763, + "num_tokens": 837943112.0, + "step": 21581 + }, + { + "epoch": 2.7454522325403894, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.009199142456055, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.9034401774406433, + "num_tokens": 837982748.0, + "step": 21582 + }, + { + "epoch": 2.74557944281898, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.922347068786621, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.897796094417572, + "num_tokens": 838021294.0, + "step": 21583 + }, + { + "epoch": 2.7457066530975704, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.937822341918945, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9059865474700928, + "num_tokens": 838056168.0, + "step": 21584 + }, + { + "epoch": 2.745833863376161, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.993535995483398, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9044238328933716, + "num_tokens": 838094900.0, + "step": 21585 + }, + { + "epoch": 2.745961073654751, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.986757278442383, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8989928960800171, + "num_tokens": 838129159.0, + "step": 21586 + }, + { + "epoch": 2.746088283933342, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.969937324523926, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.9014880061149597, + "num_tokens": 838167798.0, + "step": 21587 + }, + { + "epoch": 2.746215494211932, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.842782974243164, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.9046915769577026, + "num_tokens": 838208821.0, + "step": 21588 + }, + { + "epoch": 2.746342704490523, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.948970794677734, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8957381248474121, + "num_tokens": 838247654.0, + "step": 21589 + }, + { + "epoch": 2.746469914769113, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.033978462219238, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8880875110626221, + "num_tokens": 838288101.0, + "step": 21590 + }, + { + "epoch": 2.7465971250477037, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.844523429870605, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9093875885009766, + "num_tokens": 838332136.0, + "step": 21591 + }, + { + "epoch": 2.7467243353262942, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.937054634094238, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8915673494338989, + "num_tokens": 838369464.0, + "step": 21592 + }, + { + "epoch": 2.7468515456048848, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.915024757385254, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.9004995822906494, + "num_tokens": 838413833.0, + "step": 21593 + }, + { + "epoch": 2.7469787558834753, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.946547508239746, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8963037729263306, + "num_tokens": 838447896.0, + "step": 21594 + }, + { + "epoch": 2.747105966162066, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.980436325073242, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8903807401657104, + "num_tokens": 838489913.0, + "step": 21595 + }, + { + "epoch": 2.7472331764406563, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.864374160766602, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9054858088493347, + "num_tokens": 838527432.0, + "step": 21596 + }, + { + "epoch": 2.747360386719247, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.059264183044434, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.9076278209686279, + "num_tokens": 838558404.0, + "step": 21597 + }, + { + "epoch": 2.7474875969978374, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.913649559020996, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8932399749755859, + "num_tokens": 838602561.0, + "step": 21598 + }, + { + "epoch": 2.747614807276428, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.925636291503906, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9075672030448914, + "num_tokens": 838639956.0, + "step": 21599 + }, + { + "epoch": 2.7477420175550185, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.060256958007812, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8943725824356079, + "num_tokens": 838677869.0, + "step": 21600 + }, + { + "epoch": 2.747869227833609, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.958268165588379, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8969751596450806, + "num_tokens": 838715856.0, + "step": 21601 + }, + { + "epoch": 2.7479964381121995, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.890739440917969, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8949460983276367, + "num_tokens": 838757641.0, + "step": 21602 + }, + { + "epoch": 2.74812364839079, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.973934173583984, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9026505947113037, + "num_tokens": 838795699.0, + "step": 21603 + }, + { + "epoch": 2.7482508586693806, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.934263229370117, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.9042292237281799, + "num_tokens": 838839406.0, + "step": 21604 + }, + { + "epoch": 2.748378068947971, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.875362396240234, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.9003663063049316, + "num_tokens": 838881375.0, + "step": 21605 + }, + { + "epoch": 2.7485052792265616, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.88315200805664, + "learning_rate": 1e-06, + "loss": 0.2343, + "mean_token_accuracy": 0.919205367565155, + "num_tokens": 838921417.0, + "step": 21606 + }, + { + "epoch": 2.748632489505152, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.858119010925293, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9128784537315369, + "num_tokens": 838965671.0, + "step": 21607 + }, + { + "epoch": 2.7487596997837427, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.90096378326416, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.9050724506378174, + "num_tokens": 839009812.0, + "step": 21608 + }, + { + "epoch": 2.7488869100623328, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.893453598022461, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9063119888305664, + "num_tokens": 839053180.0, + "step": 21609 + }, + { + "epoch": 2.7490141203409237, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.99661922454834, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8944098353385925, + "num_tokens": 839095058.0, + "step": 21610 + }, + { + "epoch": 2.749141330619514, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.977228164672852, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8926422595977783, + "num_tokens": 839133939.0, + "step": 21611 + }, + { + "epoch": 2.749268540898105, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.933427810668945, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.9021378755569458, + "num_tokens": 839174722.0, + "step": 21612 + }, + { + "epoch": 2.749395751176695, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.962565422058105, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8919100761413574, + "num_tokens": 839214105.0, + "step": 21613 + }, + { + "epoch": 2.749522961455286, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.934510231018066, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9062793254852295, + "num_tokens": 839256490.0, + "step": 21614 + }, + { + "epoch": 2.749650171733876, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.931303024291992, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8797820210456848, + "num_tokens": 839298557.0, + "step": 21615 + }, + { + "epoch": 2.7497773820124665, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.983829498291016, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8930469751358032, + "num_tokens": 839334805.0, + "step": 21616 + }, + { + "epoch": 2.749904592291057, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.952110290527344, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8922271132469177, + "num_tokens": 839375327.0, + "step": 21617 + }, + { + "epoch": 2.7500318025696475, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.801645278930664, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9083467721939087, + "num_tokens": 839417395.0, + "step": 21618 + }, + { + "epoch": 2.750159012848238, + "ewc_loss": 0.0216064453125, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.1980533599853516e-05, + "grad_norm": 9.9048490524292, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8831682205200195, + "num_tokens": 839461930.0, + "step": 21619 + }, + { + "epoch": 2.7502862231268286, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.922475814819336, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.9037445783615112, + "num_tokens": 839498892.0, + "step": 21620 + }, + { + "epoch": 2.750413433405419, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.974452018737793, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9073607325553894, + "num_tokens": 839532125.0, + "step": 21621 + }, + { + "epoch": 2.7505406436840096, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.010309219360352, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8932414054870605, + "num_tokens": 839568229.0, + "step": 21622 + }, + { + "epoch": 2.7506678539626, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.996461868286133, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8910810351371765, + "num_tokens": 839604261.0, + "step": 21623 + }, + { + "epoch": 2.7507950642411907, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.119746208190918, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.897365927696228, + "num_tokens": 839632330.0, + "step": 21624 + }, + { + "epoch": 2.750922274519781, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.852060317993164, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8883829116821289, + "num_tokens": 839679053.0, + "step": 21625 + }, + { + "epoch": 2.7510494847983717, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.868413925170898, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8991166353225708, + "num_tokens": 839723487.0, + "step": 21626 + }, + { + "epoch": 2.7511766950769623, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.9490385055542, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.9035882949829102, + "num_tokens": 839757657.0, + "step": 21627 + }, + { + "epoch": 2.751303905355553, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.95553207397461, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8974862694740295, + "num_tokens": 839794172.0, + "step": 21628 + }, + { + "epoch": 2.7514311156341433, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.92702865600586, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8962720036506653, + "num_tokens": 839831054.0, + "step": 21629 + }, + { + "epoch": 2.751558325912734, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.913284301757812, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8987506031990051, + "num_tokens": 839870893.0, + "step": 21630 + }, + { + "epoch": 2.7516855361913244, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.9473876953125, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.912195086479187, + "num_tokens": 839907509.0, + "step": 21631 + }, + { + "epoch": 2.751812746469915, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.943608283996582, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9041926860809326, + "num_tokens": 839945985.0, + "step": 21632 + }, + { + "epoch": 2.7519399567485054, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.900771141052246, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8979328870773315, + "num_tokens": 839990369.0, + "step": 21633 + }, + { + "epoch": 2.7520671670270955, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.003768920898438, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.9017868041992188, + "num_tokens": 840024595.0, + "step": 21634 + }, + { + "epoch": 2.7521943773056865, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.949958801269531, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8865314722061157, + "num_tokens": 840063340.0, + "step": 21635 + }, + { + "epoch": 2.7523215875842766, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.949424743652344, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.9047349095344543, + "num_tokens": 840102224.0, + "step": 21636 + }, + { + "epoch": 2.7524487978628676, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.875713348388672, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.9168132543563843, + "num_tokens": 840140608.0, + "step": 21637 + }, + { + "epoch": 2.7525760081414576, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.859362602233887, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8939813375473022, + "num_tokens": 840184593.0, + "step": 21638 + }, + { + "epoch": 2.7527032184200486, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.964043617248535, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8860169053077698, + "num_tokens": 840227501.0, + "step": 21639 + }, + { + "epoch": 2.7528304286986387, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.947901725769043, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.9009832143783569, + "num_tokens": 840265583.0, + "step": 21640 + }, + { + "epoch": 2.7529576389772292, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.898368835449219, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.903550922870636, + "num_tokens": 840303595.0, + "step": 21641 + }, + { + "epoch": 2.7530848492558198, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.999394416809082, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8994060754776001, + "num_tokens": 840335322.0, + "step": 21642 + }, + { + "epoch": 2.7532120595344103, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.986092567443848, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8963220715522766, + "num_tokens": 840373953.0, + "step": 21643 + }, + { + "epoch": 2.753339269813001, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.884952545166016, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9073181748390198, + "num_tokens": 840411081.0, + "step": 21644 + }, + { + "epoch": 2.7534664800915913, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.053339958190918, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8816043138504028, + "num_tokens": 840452380.0, + "step": 21645 + }, + { + "epoch": 2.753593690370182, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.966435432434082, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8993623852729797, + "num_tokens": 840495715.0, + "step": 21646 + }, + { + "epoch": 2.7537209006487724, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.921867370605469, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.901136577129364, + "num_tokens": 840533808.0, + "step": 21647 + }, + { + "epoch": 2.753848110927363, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.98493480682373, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8976852893829346, + "num_tokens": 840572620.0, + "step": 21648 + }, + { + "epoch": 2.7539753212059535, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.957905769348145, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8923935294151306, + "num_tokens": 840612356.0, + "step": 21649 + }, + { + "epoch": 2.754102531484544, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.979490280151367, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8939229249954224, + "num_tokens": 840649200.0, + "step": 21650 + }, + { + "epoch": 2.7542297417631345, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.934901237487793, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8984723687171936, + "num_tokens": 840689100.0, + "step": 21651 + }, + { + "epoch": 2.754356952041725, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.882115364074707, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8959652185440063, + "num_tokens": 840728210.0, + "step": 21652 + }, + { + "epoch": 2.7544841623203156, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.992502212524414, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8949679136276245, + "num_tokens": 840768354.0, + "step": 21653 + }, + { + "epoch": 2.754611372598906, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.899136543273926, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9045991897583008, + "num_tokens": 840805862.0, + "step": 21654 + }, + { + "epoch": 2.7547385828774966, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.983821868896484, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.9048657417297363, + "num_tokens": 840840348.0, + "step": 21655 + }, + { + "epoch": 2.754865793156087, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.922906875610352, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8979336023330688, + "num_tokens": 840880984.0, + "step": 21656 + }, + { + "epoch": 2.7549930034346777, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.896062850952148, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8958095908164978, + "num_tokens": 840923618.0, + "step": 21657 + }, + { + "epoch": 2.755120213713268, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.889870643615723, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8946476578712463, + "num_tokens": 840967606.0, + "step": 21658 + }, + { + "epoch": 2.7552474239918583, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.033967971801758, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8960078358650208, + "num_tokens": 841007078.0, + "step": 21659 + }, + { + "epoch": 2.7553746342704493, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.984848022460938, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9037585258483887, + "num_tokens": 841045228.0, + "step": 21660 + }, + { + "epoch": 2.7555018445490393, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.896818161010742, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8988326787948608, + "num_tokens": 841085913.0, + "step": 21661 + }, + { + "epoch": 2.7556290548276303, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.866835594177246, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.903451144695282, + "num_tokens": 841125258.0, + "step": 21662 + }, + { + "epoch": 2.7557562651062204, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.900229454040527, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8984177112579346, + "num_tokens": 841168280.0, + "step": 21663 + }, + { + "epoch": 2.755883475384811, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.004999160766602, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.9057764410972595, + "num_tokens": 841204244.0, + "step": 21664 + }, + { + "epoch": 2.7560106856634015, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.005084037780762, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8871521949768066, + "num_tokens": 841247253.0, + "step": 21665 + }, + { + "epoch": 2.756137895941992, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.923178672790527, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8933591842651367, + "num_tokens": 841286529.0, + "step": 21666 + }, + { + "epoch": 2.7562651062205825, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.971400260925293, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8887816071510315, + "num_tokens": 841327664.0, + "step": 21667 + }, + { + "epoch": 2.756392316499173, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.945640563964844, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8938900232315063, + "num_tokens": 841370075.0, + "step": 21668 + }, + { + "epoch": 2.7565195267777636, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.986967086791992, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.884670615196228, + "num_tokens": 841409581.0, + "step": 21669 + }, + { + "epoch": 2.756646737056354, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.931432723999023, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8998396992683411, + "num_tokens": 841451246.0, + "step": 21670 + }, + { + "epoch": 2.7567739473349446, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.907267570495605, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.9001408815383911, + "num_tokens": 841489199.0, + "step": 21671 + }, + { + "epoch": 2.756901157613535, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.047744750976562, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.9010348320007324, + "num_tokens": 841529062.0, + "step": 21672 + }, + { + "epoch": 2.7570283678921257, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.99631404876709, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8928260803222656, + "num_tokens": 841566008.0, + "step": 21673 + }, + { + "epoch": 2.757155578170716, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.052592277526855, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8862931728363037, + "num_tokens": 841603319.0, + "step": 21674 + }, + { + "epoch": 2.7572827884493067, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.931127548217773, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8929662704467773, + "num_tokens": 841643946.0, + "step": 21675 + }, + { + "epoch": 2.7574099987278973, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.07456111907959, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.894147515296936, + "num_tokens": 841675458.0, + "step": 21676 + }, + { + "epoch": 2.757537209006488, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.973384857177734, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8948752880096436, + "num_tokens": 841712798.0, + "step": 21677 + }, + { + "epoch": 2.7576644192850783, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.001805305480957, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9003725051879883, + "num_tokens": 841750546.0, + "step": 21678 + }, + { + "epoch": 2.757791629563669, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.999961853027344, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8922361731529236, + "num_tokens": 841786473.0, + "step": 21679 + }, + { + "epoch": 2.7579188398422594, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.952509880065918, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8986072540283203, + "num_tokens": 841827101.0, + "step": 21680 + }, + { + "epoch": 2.75804605012085, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.02055549621582, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9108703136444092, + "num_tokens": 841859347.0, + "step": 21681 + }, + { + "epoch": 2.7581732603994404, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.952783584594727, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8937386274337769, + "num_tokens": 841900581.0, + "step": 21682 + }, + { + "epoch": 2.758300470678031, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.969157218933105, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9118182063102722, + "num_tokens": 841937774.0, + "step": 21683 + }, + { + "epoch": 2.758427680956621, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.880303382873535, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9111678600311279, + "num_tokens": 841980072.0, + "step": 21684 + }, + { + "epoch": 2.758554891235212, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.89908504486084, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.9036701917648315, + "num_tokens": 842022661.0, + "step": 21685 + }, + { + "epoch": 2.758682101513802, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.107123374938965, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.9035601019859314, + "num_tokens": 842061697.0, + "step": 21686 + }, + { + "epoch": 2.758809311792393, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.936139106750488, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.9056819081306458, + "num_tokens": 842094506.0, + "step": 21687 + }, + { + "epoch": 2.758936522070983, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.904422760009766, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9117113351821899, + "num_tokens": 842129622.0, + "step": 21688 + }, + { + "epoch": 2.7590637323495737, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.910096168518066, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8929945826530457, + "num_tokens": 842171762.0, + "step": 21689 + }, + { + "epoch": 2.7591909426281642, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.958806037902832, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8964325189590454, + "num_tokens": 842209752.0, + "step": 21690 + }, + { + "epoch": 2.7593181529067548, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.921553611755371, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8914128541946411, + "num_tokens": 842252007.0, + "step": 21691 + }, + { + "epoch": 2.7594453631853453, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.022562026977539, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8946518898010254, + "num_tokens": 842290700.0, + "step": 21692 + }, + { + "epoch": 2.759572573463936, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.045551300048828, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.883280873298645, + "num_tokens": 842328284.0, + "step": 21693 + }, + { + "epoch": 2.7596997837425263, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.947636604309082, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8992394804954529, + "num_tokens": 842372559.0, + "step": 21694 + }, + { + "epoch": 2.759826994021117, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.91592788696289, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8951648473739624, + "num_tokens": 842414083.0, + "step": 21695 + }, + { + "epoch": 2.7599542042997074, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.002032279968262, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8944898247718811, + "num_tokens": 842451756.0, + "step": 21696 + }, + { + "epoch": 2.760081414578298, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.919075012207031, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9089856147766113, + "num_tokens": 842490246.0, + "step": 21697 + }, + { + "epoch": 2.7602086248568884, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.846522331237793, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.9071263074874878, + "num_tokens": 842540356.0, + "step": 21698 + }, + { + "epoch": 2.760335835135479, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.964109420776367, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8914417028427124, + "num_tokens": 842580923.0, + "step": 21699 + }, + { + "epoch": 2.7604630454140695, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.0191011428833, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8947373628616333, + "num_tokens": 842621435.0, + "step": 21700 + }, + { + "epoch": 2.76059025569266, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.042859077453613, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8861767649650574, + "num_tokens": 842665169.0, + "step": 21701 + }, + { + "epoch": 2.7607174659712506, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.00545883178711, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8921459317207336, + "num_tokens": 842704566.0, + "step": 21702 + }, + { + "epoch": 2.760844676249841, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.865724563598633, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.897768497467041, + "num_tokens": 842752609.0, + "step": 21703 + }, + { + "epoch": 2.7609718865284316, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.984816551208496, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8977523446083069, + "num_tokens": 842794003.0, + "step": 21704 + }, + { + "epoch": 2.761099096807022, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.94483757019043, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8982251882553101, + "num_tokens": 842837449.0, + "step": 21705 + }, + { + "epoch": 2.7612263070856127, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.98556900024414, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8918262720108032, + "num_tokens": 842875911.0, + "step": 21706 + }, + { + "epoch": 2.7613535173642028, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.086508750915527, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8958625793457031, + "num_tokens": 842909997.0, + "step": 21707 + }, + { + "epoch": 2.7614807276427937, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.05766773223877, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.889197587966919, + "num_tokens": 842943558.0, + "step": 21708 + }, + { + "epoch": 2.761607937921384, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.075728416442871, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8924859762191772, + "num_tokens": 842979639.0, + "step": 21709 + }, + { + "epoch": 2.761735148199975, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.906018257141113, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8993330001831055, + "num_tokens": 843019378.0, + "step": 21710 + }, + { + "epoch": 2.761862358478565, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.894116401672363, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9074909090995789, + "num_tokens": 843061228.0, + "step": 21711 + }, + { + "epoch": 2.761989568757156, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.097982406616211, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.9019390344619751, + "num_tokens": 843093573.0, + "step": 21712 + }, + { + "epoch": 2.762116779035746, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.902599334716797, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8977453708648682, + "num_tokens": 843130984.0, + "step": 21713 + }, + { + "epoch": 2.7622439893143365, + "ewc_loss": 0.021728515625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.97555160522461, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.912773847579956, + "num_tokens": 843160340.0, + "step": 21714 + }, + { + "epoch": 2.762371199592927, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.975759506225586, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8893839120864868, + "num_tokens": 843198595.0, + "step": 21715 + }, + { + "epoch": 2.7624984098715175, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.82809066772461, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.9062368869781494, + "num_tokens": 843242033.0, + "step": 21716 + }, + { + "epoch": 2.762625620150108, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.048093795776367, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8836789131164551, + "num_tokens": 843280046.0, + "step": 21717 + }, + { + "epoch": 2.7627528304286986, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.977654457092285, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9050726890563965, + "num_tokens": 843323289.0, + "step": 21718 + }, + { + "epoch": 2.762880040707289, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.889266967773438, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8887842893600464, + "num_tokens": 843363888.0, + "step": 21719 + }, + { + "epoch": 2.7630072509858796, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.120570182800293, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8920078277587891, + "num_tokens": 843396010.0, + "step": 21720 + }, + { + "epoch": 2.76313446126447, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.040328025817871, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8943007588386536, + "num_tokens": 843431454.0, + "step": 21721 + }, + { + "epoch": 2.7632616715430607, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.86590576171875, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9104336500167847, + "num_tokens": 843475085.0, + "step": 21722 + }, + { + "epoch": 2.763388881821651, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.092083930969238, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9015228748321533, + "num_tokens": 843509593.0, + "step": 21723 + }, + { + "epoch": 2.7635160921002417, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.965998649597168, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9146285057067871, + "num_tokens": 843545355.0, + "step": 21724 + }, + { + "epoch": 2.7636433023788323, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.068682670593262, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8817921876907349, + "num_tokens": 843582634.0, + "step": 21725 + }, + { + "epoch": 2.763770512657423, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.975228309631348, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8900030255317688, + "num_tokens": 843622792.0, + "step": 21726 + }, + { + "epoch": 2.7638977229360133, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.91796588897705, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.9111267924308777, + "num_tokens": 843662667.0, + "step": 21727 + }, + { + "epoch": 2.764024933214604, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.03126049041748, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8819651007652283, + "num_tokens": 843699570.0, + "step": 21728 + }, + { + "epoch": 2.7641521434931944, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.946871757507324, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9053103923797607, + "num_tokens": 843739466.0, + "step": 21729 + }, + { + "epoch": 2.764279353771785, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.93531322479248, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8866029977798462, + "num_tokens": 843785732.0, + "step": 21730 + }, + { + "epoch": 2.7644065640503754, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.997270584106445, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.9052455425262451, + "num_tokens": 843821765.0, + "step": 21731 + }, + { + "epoch": 2.7645337743289655, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.932941436767578, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9097095131874084, + "num_tokens": 843858720.0, + "step": 21732 + }, + { + "epoch": 2.7646609846075565, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.953530311584473, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8949602842330933, + "num_tokens": 843901084.0, + "step": 21733 + }, + { + "epoch": 2.7647881948861466, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.965632438659668, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9071661233901978, + "num_tokens": 843939650.0, + "step": 21734 + }, + { + "epoch": 2.7649154051647375, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.926848411560059, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8992512226104736, + "num_tokens": 843981592.0, + "step": 21735 + }, + { + "epoch": 2.7650426154433276, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.023655891418457, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8898823261260986, + "num_tokens": 844023196.0, + "step": 21736 + }, + { + "epoch": 2.7651698257219186, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.032245635986328, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8903716206550598, + "num_tokens": 844062526.0, + "step": 21737 + }, + { + "epoch": 2.7652970360005087, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.93389892578125, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.9021790027618408, + "num_tokens": 844099744.0, + "step": 21738 + }, + { + "epoch": 2.765424246279099, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.910295486450195, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.898063063621521, + "num_tokens": 844144307.0, + "step": 21739 + }, + { + "epoch": 2.7655514565576897, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.944024085998535, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8985158205032349, + "num_tokens": 844186891.0, + "step": 21740 + }, + { + "epoch": 2.7656786668362803, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.025036811828613, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9047749042510986, + "num_tokens": 844222454.0, + "step": 21741 + }, + { + "epoch": 2.765805877114871, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.971475601196289, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8958865404129028, + "num_tokens": 844259628.0, + "step": 21742 + }, + { + "epoch": 2.7659330873934613, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.968148231506348, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9008742570877075, + "num_tokens": 844300338.0, + "step": 21743 + }, + { + "epoch": 2.766060297672052, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.879545211791992, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.9007371664047241, + "num_tokens": 844345253.0, + "step": 21744 + }, + { + "epoch": 2.7661875079506424, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.938764572143555, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8965623378753662, + "num_tokens": 844387292.0, + "step": 21745 + }, + { + "epoch": 2.766314718229233, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.936304092407227, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8981831669807434, + "num_tokens": 844426177.0, + "step": 21746 + }, + { + "epoch": 2.7664419285078234, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.875663757324219, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8964453339576721, + "num_tokens": 844463932.0, + "step": 21747 + }, + { + "epoch": 2.766569138786414, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.086856842041016, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.896531879901886, + "num_tokens": 844501597.0, + "step": 21748 + }, + { + "epoch": 2.7666963490650045, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.910152435302734, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8902828097343445, + "num_tokens": 844542280.0, + "step": 21749 + }, + { + "epoch": 2.766823559343595, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.866300582885742, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8992602825164795, + "num_tokens": 844588759.0, + "step": 21750 + }, + { + "epoch": 2.7669507696221856, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.889886856079102, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.9045143723487854, + "num_tokens": 844630052.0, + "step": 21751 + }, + { + "epoch": 2.767077979900776, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.92478084564209, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8981713056564331, + "num_tokens": 844670513.0, + "step": 21752 + }, + { + "epoch": 2.7672051901793666, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.015986442565918, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8967084288597107, + "num_tokens": 844709855.0, + "step": 21753 + }, + { + "epoch": 2.767332400457957, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.900529861450195, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8926981091499329, + "num_tokens": 844749932.0, + "step": 21754 + }, + { + "epoch": 2.7674596107365477, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.970192909240723, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8908088207244873, + "num_tokens": 844791194.0, + "step": 21755 + }, + { + "epoch": 2.767586821015138, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.986546516418457, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.9024784564971924, + "num_tokens": 844827952.0, + "step": 21756 + }, + { + "epoch": 2.7677140312937283, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.938920021057129, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9071990251541138, + "num_tokens": 844863544.0, + "step": 21757 + }, + { + "epoch": 2.7678412415723193, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.963724136352539, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.9016953706741333, + "num_tokens": 844908873.0, + "step": 21758 + }, + { + "epoch": 2.7679684518509093, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.002483367919922, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8977124094963074, + "num_tokens": 844943870.0, + "step": 21759 + }, + { + "epoch": 2.7680956621295003, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.993002891540527, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.9000749588012695, + "num_tokens": 844984698.0, + "step": 21760 + }, + { + "epoch": 2.7682228724080904, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.107831001281738, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8824490904808044, + "num_tokens": 845019540.0, + "step": 21761 + }, + { + "epoch": 2.768350082686681, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.91051959991455, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.900398850440979, + "num_tokens": 845062201.0, + "step": 21762 + }, + { + "epoch": 2.7684772929652715, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.963065147399902, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8970756530761719, + "num_tokens": 845100746.0, + "step": 21763 + }, + { + "epoch": 2.768604503243862, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.094937324523926, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8915733098983765, + "num_tokens": 845140716.0, + "step": 21764 + }, + { + "epoch": 2.7687317135224525, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.960488319396973, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8967344164848328, + "num_tokens": 845177103.0, + "step": 21765 + }, + { + "epoch": 2.768858923801043, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.051292419433594, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8805657029151917, + "num_tokens": 845219075.0, + "step": 21766 + }, + { + "epoch": 2.7689861340796336, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.961944580078125, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8966858386993408, + "num_tokens": 845255384.0, + "step": 21767 + }, + { + "epoch": 2.769113344358224, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.191362380981445, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.90174400806427, + "num_tokens": 845289801.0, + "step": 21768 + }, + { + "epoch": 2.7692405546368146, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.204691886901855, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8939318656921387, + "num_tokens": 845322126.0, + "step": 21769 + }, + { + "epoch": 2.769367764915405, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.98855972290039, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8899421691894531, + "num_tokens": 845357251.0, + "step": 21770 + }, + { + "epoch": 2.7694949751939957, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.943304061889648, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9043540954589844, + "num_tokens": 845400649.0, + "step": 21771 + }, + { + "epoch": 2.769622185472586, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.186688423156738, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8947782516479492, + "num_tokens": 845438521.0, + "step": 21772 + }, + { + "epoch": 2.7697493957511767, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.925971984863281, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.9002910852432251, + "num_tokens": 845480130.0, + "step": 21773 + }, + { + "epoch": 2.7698766060297673, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.946614265441895, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.896419882774353, + "num_tokens": 845524663.0, + "step": 21774 + }, + { + "epoch": 2.770003816308358, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.938875198364258, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8977296352386475, + "num_tokens": 845561824.0, + "step": 21775 + }, + { + "epoch": 2.7701310265869483, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.934917449951172, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9025465250015259, + "num_tokens": 845602470.0, + "step": 21776 + }, + { + "epoch": 2.770258236865539, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.050244331359863, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8911749720573425, + "num_tokens": 845638896.0, + "step": 21777 + }, + { + "epoch": 2.7703854471441294, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.033842086791992, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.9002819061279297, + "num_tokens": 845676388.0, + "step": 21778 + }, + { + "epoch": 2.77051265742272, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.95971965789795, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8992617726325989, + "num_tokens": 845716941.0, + "step": 21779 + }, + { + "epoch": 2.77063986770131, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.007453918457031, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9044570326805115, + "num_tokens": 845749002.0, + "step": 21780 + }, + { + "epoch": 2.770767077979901, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.072833061218262, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.889462947845459, + "num_tokens": 845785490.0, + "step": 21781 + }, + { + "epoch": 2.770894288258491, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.919879913330078, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8895471096038818, + "num_tokens": 845829256.0, + "step": 21782 + }, + { + "epoch": 2.771021498537082, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.03713607788086, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8923963308334351, + "num_tokens": 845867851.0, + "step": 21783 + }, + { + "epoch": 2.771148708815672, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.02795124053955, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9068589806556702, + "num_tokens": 845903864.0, + "step": 21784 + }, + { + "epoch": 2.771275919094263, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.970760345458984, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8996678590774536, + "num_tokens": 845939294.0, + "step": 21785 + }, + { + "epoch": 2.771403129372853, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.085070610046387, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8919070363044739, + "num_tokens": 845979194.0, + "step": 21786 + }, + { + "epoch": 2.7715303396514437, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.094823837280273, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8823825120925903, + "num_tokens": 846015149.0, + "step": 21787 + }, + { + "epoch": 2.771657549930034, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.90172290802002, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8934211730957031, + "num_tokens": 846054501.0, + "step": 21788 + }, + { + "epoch": 2.7717847602086247, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.934884071350098, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9044525623321533, + "num_tokens": 846096190.0, + "step": 21789 + }, + { + "epoch": 2.7719119704872153, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.028401374816895, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8970296382904053, + "num_tokens": 846131053.0, + "step": 21790 + }, + { + "epoch": 2.772039180765806, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.960936546325684, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8944718241691589, + "num_tokens": 846173149.0, + "step": 21791 + }, + { + "epoch": 2.7721663910443963, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.880287170410156, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.883621871471405, + "num_tokens": 846219470.0, + "step": 21792 + }, + { + "epoch": 2.772293601322987, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.993886947631836, + "learning_rate": 1e-06, + "loss": 0.2499, + "mean_token_accuracy": 0.9137157201766968, + "num_tokens": 846252179.0, + "step": 21793 + }, + { + "epoch": 2.7724208116015774, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.932893753051758, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.9007701873779297, + "num_tokens": 846289413.0, + "step": 21794 + }, + { + "epoch": 2.772548021880168, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.857887268066406, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.9024263620376587, + "num_tokens": 846333833.0, + "step": 21795 + }, + { + "epoch": 2.7726752321587584, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.95868968963623, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.9005730748176575, + "num_tokens": 846370680.0, + "step": 21796 + }, + { + "epoch": 2.772802442437349, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.943534851074219, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8904275298118591, + "num_tokens": 846414437.0, + "step": 21797 + }, + { + "epoch": 2.7729296527159395, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.992940902709961, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9041846990585327, + "num_tokens": 846452194.0, + "step": 21798 + }, + { + "epoch": 2.77305686299453, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.959158897399902, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9021768569946289, + "num_tokens": 846487492.0, + "step": 21799 + }, + { + "epoch": 2.7731840732731206, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.93419361114502, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9045708775520325, + "num_tokens": 846528165.0, + "step": 21800 + }, + { + "epoch": 2.773311283551711, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.93192195892334, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9120699763298035, + "num_tokens": 846564369.0, + "step": 21801 + }, + { + "epoch": 2.7734384938303016, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.123069763183594, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8734236359596252, + "num_tokens": 846603116.0, + "step": 21802 + }, + { + "epoch": 2.773565704108892, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.996816635131836, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8872970342636108, + "num_tokens": 846644735.0, + "step": 21803 + }, + { + "epoch": 2.7736929143874827, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.054023742675781, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9024234414100647, + "num_tokens": 846676467.0, + "step": 21804 + }, + { + "epoch": 2.7738201246660728, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.974332809448242, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.902450680732727, + "num_tokens": 846712029.0, + "step": 21805 + }, + { + "epoch": 2.7739473349446637, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.914047241210938, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9065571427345276, + "num_tokens": 846749119.0, + "step": 21806 + }, + { + "epoch": 2.774074545223254, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.026010513305664, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8989720344543457, + "num_tokens": 846784415.0, + "step": 21807 + }, + { + "epoch": 2.774201755501845, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.888252258300781, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9112676382064819, + "num_tokens": 846825500.0, + "step": 21808 + }, + { + "epoch": 2.774328965780435, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.076330184936523, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.9007716178894043, + "num_tokens": 846861866.0, + "step": 21809 + }, + { + "epoch": 2.774456176059026, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.032928466796875, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8969402313232422, + "num_tokens": 846899046.0, + "step": 21810 + }, + { + "epoch": 2.774583386337616, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.949734687805176, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.899090588092804, + "num_tokens": 846937037.0, + "step": 21811 + }, + { + "epoch": 2.7747105966162064, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.04238224029541, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.902129054069519, + "num_tokens": 846969900.0, + "step": 21812 + }, + { + "epoch": 2.774837806894797, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.017495155334473, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.9024993777275085, + "num_tokens": 847002183.0, + "step": 21813 + }, + { + "epoch": 2.7749650171733875, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.936891555786133, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9037867784500122, + "num_tokens": 847044104.0, + "step": 21814 + }, + { + "epoch": 2.775092227451978, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.937904357910156, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8995370864868164, + "num_tokens": 847087511.0, + "step": 21815 + }, + { + "epoch": 2.7752194377305686, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.019308090209961, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8906044363975525, + "num_tokens": 847132719.0, + "step": 21816 + }, + { + "epoch": 2.775346648009159, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.954151153564453, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9018768072128296, + "num_tokens": 847167114.0, + "step": 21817 + }, + { + "epoch": 2.7754738582877496, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.049844741821289, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8937269449234009, + "num_tokens": 847203572.0, + "step": 21818 + }, + { + "epoch": 2.77560106856634, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.974529266357422, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8886436223983765, + "num_tokens": 847247765.0, + "step": 21819 + }, + { + "epoch": 2.7757282788449307, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.024656295776367, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8951506614685059, + "num_tokens": 847283658.0, + "step": 21820 + }, + { + "epoch": 2.775855489123521, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.859148979187012, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9082520604133606, + "num_tokens": 847325848.0, + "step": 21821 + }, + { + "epoch": 2.7759826994021117, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.952757835388184, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.9004116654396057, + "num_tokens": 847363835.0, + "step": 21822 + }, + { + "epoch": 2.7761099096807023, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.023626327514648, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8897889852523804, + "num_tokens": 847404392.0, + "step": 21823 + }, + { + "epoch": 2.776237119959293, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.908738136291504, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8964759111404419, + "num_tokens": 847449031.0, + "step": 21824 + }, + { + "epoch": 2.7763643302378833, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.018682479858398, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.894336462020874, + "num_tokens": 847488225.0, + "step": 21825 + }, + { + "epoch": 2.776491540516474, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.844550132751465, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.897849977016449, + "num_tokens": 847531331.0, + "step": 21826 + }, + { + "epoch": 2.7766187507950644, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.01759147644043, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.889397382736206, + "num_tokens": 847567382.0, + "step": 21827 + }, + { + "epoch": 2.776745961073655, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.952518463134766, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8924461603164673, + "num_tokens": 847609961.0, + "step": 21828 + }, + { + "epoch": 2.7768731713522454, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.889544486999512, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.9022437334060669, + "num_tokens": 847653289.0, + "step": 21829 + }, + { + "epoch": 2.7770003816308355, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.940261840820312, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8738263845443726, + "num_tokens": 847693593.0, + "step": 21830 + }, + { + "epoch": 2.7771275919094265, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.02835464477539, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8954272866249084, + "num_tokens": 847729818.0, + "step": 21831 + }, + { + "epoch": 2.7772548021880166, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.966071128845215, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8886420726776123, + "num_tokens": 847769927.0, + "step": 21832 + }, + { + "epoch": 2.7773820124666075, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.061466217041016, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8960931301116943, + "num_tokens": 847806918.0, + "step": 21833 + }, + { + "epoch": 2.7775092227451976, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.030292510986328, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8930840492248535, + "num_tokens": 847845364.0, + "step": 21834 + }, + { + "epoch": 2.7776364330237886, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.950424194335938, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.9045900702476501, + "num_tokens": 847879984.0, + "step": 21835 + }, + { + "epoch": 2.7777636433023787, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.868711471557617, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9085546731948853, + "num_tokens": 847920767.0, + "step": 21836 + }, + { + "epoch": 2.777890853580969, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.003451347351074, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.905794620513916, + "num_tokens": 847957078.0, + "step": 21837 + }, + { + "epoch": 2.7780180638595597, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.016203880310059, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8984206914901733, + "num_tokens": 847993102.0, + "step": 21838 + }, + { + "epoch": 2.7781452741381503, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.991280555725098, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.894500195980072, + "num_tokens": 848032992.0, + "step": 21839 + }, + { + "epoch": 2.778272484416741, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.930020332336426, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9106563925743103, + "num_tokens": 848070250.0, + "step": 21840 + }, + { + "epoch": 2.7783996946953313, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.98017692565918, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8930904865264893, + "num_tokens": 848107207.0, + "step": 21841 + }, + { + "epoch": 2.778526904973922, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.03597354888916, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8949968814849854, + "num_tokens": 848145719.0, + "step": 21842 + }, + { + "epoch": 2.7786541152525124, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.047808647155762, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8921817541122437, + "num_tokens": 848178873.0, + "step": 21843 + }, + { + "epoch": 2.778781325531103, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.027368545532227, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.9007956385612488, + "num_tokens": 848214051.0, + "step": 21844 + }, + { + "epoch": 2.7789085358096934, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.938143730163574, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.906425952911377, + "num_tokens": 848252775.0, + "step": 21845 + }, + { + "epoch": 2.779035746088284, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.01055908203125, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8914515972137451, + "num_tokens": 848288339.0, + "step": 21846 + }, + { + "epoch": 2.7791629563668745, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.040908813476562, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8931922316551208, + "num_tokens": 848325972.0, + "step": 21847 + }, + { + "epoch": 2.779290166645465, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.991110801696777, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8914721012115479, + "num_tokens": 848364388.0, + "step": 21848 + }, + { + "epoch": 2.7794173769240555, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.896811485290527, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8981608152389526, + "num_tokens": 848403177.0, + "step": 21849 + }, + { + "epoch": 2.779544587202646, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.013116836547852, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9090534448623657, + "num_tokens": 848438005.0, + "step": 21850 + }, + { + "epoch": 2.7796717974812366, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.070330619812012, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8967523574829102, + "num_tokens": 848475919.0, + "step": 21851 + }, + { + "epoch": 2.779799007759827, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.945167541503906, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9005120396614075, + "num_tokens": 848518033.0, + "step": 21852 + }, + { + "epoch": 2.7799262180384177, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.978371620178223, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.9035799503326416, + "num_tokens": 848555693.0, + "step": 21853 + }, + { + "epoch": 2.780053428317008, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.963501930236816, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.9007174372673035, + "num_tokens": 848594030.0, + "step": 21854 + }, + { + "epoch": 2.7801806385955983, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.083239555358887, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8903026580810547, + "num_tokens": 848632012.0, + "step": 21855 + }, + { + "epoch": 2.7803078488741892, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.206814765930176, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8789798021316528, + "num_tokens": 848665052.0, + "step": 21856 + }, + { + "epoch": 2.7804350591527793, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.905121803283691, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8951661586761475, + "num_tokens": 848710506.0, + "step": 21857 + }, + { + "epoch": 2.7805622694313703, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.081344604492188, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8935925364494324, + "num_tokens": 848744434.0, + "step": 21858 + }, + { + "epoch": 2.7806894797099604, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.967180252075195, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8976537585258484, + "num_tokens": 848782013.0, + "step": 21859 + }, + { + "epoch": 2.780816689988551, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.965958595275879, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8889570236206055, + "num_tokens": 848820063.0, + "step": 21860 + }, + { + "epoch": 2.7809439002671414, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.996736526489258, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8919832706451416, + "num_tokens": 848862432.0, + "step": 21861 + }, + { + "epoch": 2.781071110545732, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.868457794189453, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9078785181045532, + "num_tokens": 848903519.0, + "step": 21862 + }, + { + "epoch": 2.7811983208243225, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.957633018493652, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9061272144317627, + "num_tokens": 848942618.0, + "step": 21863 + }, + { + "epoch": 2.781325531102913, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.947460174560547, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8870418667793274, + "num_tokens": 848984806.0, + "step": 21864 + }, + { + "epoch": 2.7814527413815036, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.923934936523438, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.9058132171630859, + "num_tokens": 849023462.0, + "step": 21865 + }, + { + "epoch": 2.781579951660094, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.949013710021973, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9045150279998779, + "num_tokens": 849062335.0, + "step": 21866 + }, + { + "epoch": 2.7817071619386846, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.998096466064453, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.894533097743988, + "num_tokens": 849102302.0, + "step": 21867 + }, + { + "epoch": 2.781834372217275, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.901801109313965, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.9040981531143188, + "num_tokens": 849142580.0, + "step": 21868 + }, + { + "epoch": 2.7819615824958657, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.108416557312012, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8785855174064636, + "num_tokens": 849180593.0, + "step": 21869 + }, + { + "epoch": 2.782088792774456, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.00203800201416, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8986494541168213, + "num_tokens": 849218263.0, + "step": 21870 + }, + { + "epoch": 2.7822160030530467, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.925549507141113, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8896632790565491, + "num_tokens": 849262970.0, + "step": 21871 + }, + { + "epoch": 2.7823432133316373, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.00461196899414, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8951374292373657, + "num_tokens": 849300511.0, + "step": 21872 + }, + { + "epoch": 2.782470423610228, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.972068786621094, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8902899026870728, + "num_tokens": 849339009.0, + "step": 21873 + }, + { + "epoch": 2.7825976338888183, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.93076229095459, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9061440229415894, + "num_tokens": 849375386.0, + "step": 21874 + }, + { + "epoch": 2.782724844167409, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.940187454223633, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8971444964408875, + "num_tokens": 849418633.0, + "step": 21875 + }, + { + "epoch": 2.7828520544459994, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.933550834655762, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9098745584487915, + "num_tokens": 849458803.0, + "step": 21876 + }, + { + "epoch": 2.78297926472459, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.945935249328613, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8920025825500488, + "num_tokens": 849499994.0, + "step": 21877 + }, + { + "epoch": 2.78310647500318, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.973612785339355, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8945016264915466, + "num_tokens": 849542124.0, + "step": 21878 + }, + { + "epoch": 2.783233685281771, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.923200607299805, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9099275469779968, + "num_tokens": 849582044.0, + "step": 21879 + }, + { + "epoch": 2.783360895560361, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.076494216918945, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.9017612338066101, + "num_tokens": 849614612.0, + "step": 21880 + }, + { + "epoch": 2.783488105838952, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.068840980529785, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8950022459030151, + "num_tokens": 849654440.0, + "step": 21881 + }, + { + "epoch": 2.783615316117542, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.01734733581543, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.9018610715866089, + "num_tokens": 849693015.0, + "step": 21882 + }, + { + "epoch": 2.783742526396133, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.019947052001953, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.905854344367981, + "num_tokens": 849731592.0, + "step": 21883 + }, + { + "epoch": 2.783869736674723, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.98652172088623, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8995323181152344, + "num_tokens": 849765864.0, + "step": 21884 + }, + { + "epoch": 2.7839969469533137, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.022351264953613, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8881808519363403, + "num_tokens": 849804771.0, + "step": 21885 + }, + { + "epoch": 2.784124157231904, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.937376976013184, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8980586528778076, + "num_tokens": 849843059.0, + "step": 21886 + }, + { + "epoch": 2.7842513675104947, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.082024574279785, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8962389826774597, + "num_tokens": 849877272.0, + "step": 21887 + }, + { + "epoch": 2.7843785777890853, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.026468276977539, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8910460472106934, + "num_tokens": 849916311.0, + "step": 21888 + }, + { + "epoch": 2.784505788067676, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.955309867858887, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8903186321258545, + "num_tokens": 849957566.0, + "step": 21889 + }, + { + "epoch": 2.7846329983462663, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.89574909210205, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9071111679077148, + "num_tokens": 849998953.0, + "step": 21890 + }, + { + "epoch": 2.784760208624857, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.983095169067383, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.9090262055397034, + "num_tokens": 850038101.0, + "step": 21891 + }, + { + "epoch": 2.7848874189034474, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.948296546936035, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8940213322639465, + "num_tokens": 850078313.0, + "step": 21892 + }, + { + "epoch": 2.785014629182038, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.916308403015137, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8922051191329956, + "num_tokens": 850120809.0, + "step": 21893 + }, + { + "epoch": 2.7851418394606284, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.015188217163086, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.877568244934082, + "num_tokens": 850162815.0, + "step": 21894 + }, + { + "epoch": 2.785269049739219, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.992912292480469, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9166405200958252, + "num_tokens": 850194603.0, + "step": 21895 + }, + { + "epoch": 2.7853962600178095, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.991595268249512, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8833163380622864, + "num_tokens": 850240577.0, + "step": 21896 + }, + { + "epoch": 2.7855234702964, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.94188117980957, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.898987889289856, + "num_tokens": 850283530.0, + "step": 21897 + }, + { + "epoch": 2.7856506805749905, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.915291786193848, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8910202980041504, + "num_tokens": 850328194.0, + "step": 21898 + }, + { + "epoch": 2.785777890853581, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.026474952697754, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.9008568525314331, + "num_tokens": 850360894.0, + "step": 21899 + }, + { + "epoch": 2.7859051011321716, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.001938819885254, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.9006467461585999, + "num_tokens": 850395326.0, + "step": 21900 + }, + { + "epoch": 2.786032311410762, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.959392547607422, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8953614234924316, + "num_tokens": 850433881.0, + "step": 21901 + }, + { + "epoch": 2.7861595216893527, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.918380737304688, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.9045649766921997, + "num_tokens": 850475067.0, + "step": 21902 + }, + { + "epoch": 2.7862867319679427, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.961877822875977, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8880606293678284, + "num_tokens": 850514896.0, + "step": 21903 + }, + { + "epoch": 2.7864139422465337, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.992213249206543, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.9007275104522705, + "num_tokens": 850548056.0, + "step": 21904 + }, + { + "epoch": 2.786541152525124, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.995140075683594, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8925766944885254, + "num_tokens": 850583904.0, + "step": 21905 + }, + { + "epoch": 2.7866683628037148, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.940048217773438, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8882337212562561, + "num_tokens": 850629040.0, + "step": 21906 + }, + { + "epoch": 2.786795573082305, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.072860717773438, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8908557891845703, + "num_tokens": 850661194.0, + "step": 21907 + }, + { + "epoch": 2.786922783360896, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.095846176147461, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8999046087265015, + "num_tokens": 850694015.0, + "step": 21908 + }, + { + "epoch": 2.787049993639486, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.935827255249023, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.893405020236969, + "num_tokens": 850734167.0, + "step": 21909 + }, + { + "epoch": 2.7871772039180764, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.913320541381836, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9068875908851624, + "num_tokens": 850773562.0, + "step": 21910 + }, + { + "epoch": 2.787304414196667, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.953011512756348, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8966119289398193, + "num_tokens": 850811123.0, + "step": 21911 + }, + { + "epoch": 2.7874316244752575, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.247950553894043, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8928414583206177, + "num_tokens": 850848812.0, + "step": 21912 + }, + { + "epoch": 2.787558834753848, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.99513053894043, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9077919721603394, + "num_tokens": 850885163.0, + "step": 21913 + }, + { + "epoch": 2.7876860450324386, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.028133392333984, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8996803164482117, + "num_tokens": 850915928.0, + "step": 21914 + }, + { + "epoch": 2.787813255311029, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.9213228225708, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.9032022356987, + "num_tokens": 850953017.0, + "step": 21915 + }, + { + "epoch": 2.7879404655896196, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.999188423156738, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8895518779754639, + "num_tokens": 850991423.0, + "step": 21916 + }, + { + "epoch": 2.78806767586821, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.938836097717285, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8961207866668701, + "num_tokens": 851030763.0, + "step": 21917 + }, + { + "epoch": 2.7881948861468007, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.981175422668457, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9033341407775879, + "num_tokens": 851068755.0, + "step": 21918 + }, + { + "epoch": 2.788322096425391, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.993568420410156, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8918719291687012, + "num_tokens": 851110292.0, + "step": 21919 + }, + { + "epoch": 2.7884493067039817, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.021432876586914, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.9022030830383301, + "num_tokens": 851148369.0, + "step": 21920 + }, + { + "epoch": 2.7885765169825723, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.022521018981934, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8909116983413696, + "num_tokens": 851191392.0, + "step": 21921 + }, + { + "epoch": 2.788703727261163, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.864361763000488, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9025296568870544, + "num_tokens": 851232598.0, + "step": 21922 + }, + { + "epoch": 2.7888309375397533, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.998579025268555, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8980039954185486, + "num_tokens": 851272114.0, + "step": 21923 + }, + { + "epoch": 2.788958147818344, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.92331314086914, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.905656099319458, + "num_tokens": 851313906.0, + "step": 21924 + }, + { + "epoch": 2.7890853580969344, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.99485969543457, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.88910973072052, + "num_tokens": 851353898.0, + "step": 21925 + }, + { + "epoch": 2.789212568375525, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.904780387878418, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9100859761238098, + "num_tokens": 851394625.0, + "step": 21926 + }, + { + "epoch": 2.7893397786541154, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.949962615966797, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.9008558988571167, + "num_tokens": 851434997.0, + "step": 21927 + }, + { + "epoch": 2.7894669889327055, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.995290756225586, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9041147232055664, + "num_tokens": 851470545.0, + "step": 21928 + }, + { + "epoch": 2.7895941992112965, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.932759284973145, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8986591100692749, + "num_tokens": 851514402.0, + "step": 21929 + }, + { + "epoch": 2.7897214094898866, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.016993522644043, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.90838623046875, + "num_tokens": 851549335.0, + "step": 21930 + }, + { + "epoch": 2.7898486197684775, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.119330406188965, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8863402605056763, + "num_tokens": 851584555.0, + "step": 21931 + }, + { + "epoch": 2.7899758300470676, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.092902183532715, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.9053943753242493, + "num_tokens": 851615128.0, + "step": 21932 + }, + { + "epoch": 2.7901030403256586, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.867264747619629, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8966740369796753, + "num_tokens": 851662832.0, + "step": 21933 + }, + { + "epoch": 2.7902302506042487, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.935552597045898, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9064651131629944, + "num_tokens": 851701101.0, + "step": 21934 + }, + { + "epoch": 2.790357460882839, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.062365531921387, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8998044729232788, + "num_tokens": 851736262.0, + "step": 21935 + }, + { + "epoch": 2.7904846711614297, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.931015968322754, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9065161943435669, + "num_tokens": 851774532.0, + "step": 21936 + }, + { + "epoch": 2.7906118814400203, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.154751777648926, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8911701440811157, + "num_tokens": 851805538.0, + "step": 21937 + }, + { + "epoch": 2.790739091718611, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.911384582519531, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9057132601737976, + "num_tokens": 851849015.0, + "step": 21938 + }, + { + "epoch": 2.7908663019972013, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.996028900146484, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9065576791763306, + "num_tokens": 851886206.0, + "step": 21939 + }, + { + "epoch": 2.790993512275792, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.927604675292969, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8991854190826416, + "num_tokens": 851929993.0, + "step": 21940 + }, + { + "epoch": 2.7911207225543824, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.033695220947266, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8887056708335876, + "num_tokens": 851968370.0, + "step": 21941 + }, + { + "epoch": 2.791247932832973, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.979415893554688, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8932694792747498, + "num_tokens": 852005749.0, + "step": 21942 + }, + { + "epoch": 2.7913751431115634, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.975015640258789, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8990626335144043, + "num_tokens": 852047239.0, + "step": 21943 + }, + { + "epoch": 2.791502353390154, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.88995361328125, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.9013957977294922, + "num_tokens": 852088322.0, + "step": 21944 + }, + { + "epoch": 2.7916295636687445, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.903279304504395, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9056308269500732, + "num_tokens": 852133256.0, + "step": 21945 + }, + { + "epoch": 2.791756773947335, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.00697135925293, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8940953016281128, + "num_tokens": 852172439.0, + "step": 21946 + }, + { + "epoch": 2.7918839842259255, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.973550796508789, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9082238078117371, + "num_tokens": 852207957.0, + "step": 21947 + }, + { + "epoch": 2.792011194504516, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.006591796875, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8901635408401489, + "num_tokens": 852247524.0, + "step": 21948 + }, + { + "epoch": 2.7921384047831066, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.96529769897461, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8970106244087219, + "num_tokens": 852285393.0, + "step": 21949 + }, + { + "epoch": 2.792265615061697, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.999759674072266, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.904288113117218, + "num_tokens": 852322788.0, + "step": 21950 + }, + { + "epoch": 2.7923928253402877, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.901267051696777, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9113233089447021, + "num_tokens": 852364338.0, + "step": 21951 + }, + { + "epoch": 2.792520035618878, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.008358001708984, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.9054220914840698, + "num_tokens": 852401505.0, + "step": 21952 + }, + { + "epoch": 2.7926472458974683, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.017308235168457, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.9018259048461914, + "num_tokens": 852438045.0, + "step": 21953 + }, + { + "epoch": 2.7927744561760592, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.847399711608887, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8902902603149414, + "num_tokens": 852486346.0, + "step": 21954 + }, + { + "epoch": 2.7929016664546493, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.006491661071777, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8941489458084106, + "num_tokens": 852527739.0, + "step": 21955 + }, + { + "epoch": 2.7930288767332403, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.099474906921387, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9037202000617981, + "num_tokens": 852554892.0, + "step": 21956 + }, + { + "epoch": 2.7931560870118304, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.951991081237793, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9044694900512695, + "num_tokens": 852596172.0, + "step": 21957 + }, + { + "epoch": 2.793283297290421, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.936236381530762, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8947857022285461, + "num_tokens": 852641268.0, + "step": 21958 + }, + { + "epoch": 2.7934105075690114, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.98509693145752, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8989555239677429, + "num_tokens": 852679063.0, + "step": 21959 + }, + { + "epoch": 2.793537717847602, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.980819702148438, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.9027328491210938, + "num_tokens": 852716414.0, + "step": 21960 + }, + { + "epoch": 2.7936649281261925, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.899667739868164, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9076615571975708, + "num_tokens": 852756866.0, + "step": 21961 + }, + { + "epoch": 2.793792138404783, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.989928245544434, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9100072979927063, + "num_tokens": 852790283.0, + "step": 21962 + }, + { + "epoch": 2.7939193486833735, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.006866455078125, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.9048192501068115, + "num_tokens": 852824675.0, + "step": 21963 + }, + { + "epoch": 2.794046558961964, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.069507598876953, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8958355188369751, + "num_tokens": 852862816.0, + "step": 21964 + }, + { + "epoch": 2.7941737692405546, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.915233612060547, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8963980674743652, + "num_tokens": 852908232.0, + "step": 21965 + }, + { + "epoch": 2.794300979519145, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.975364685058594, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.9027165174484253, + "num_tokens": 852946550.0, + "step": 21966 + }, + { + "epoch": 2.7944281897977357, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.010462760925293, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.9017035961151123, + "num_tokens": 852983729.0, + "step": 21967 + }, + { + "epoch": 2.794555400076326, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.978168487548828, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.9048206806182861, + "num_tokens": 853019310.0, + "step": 21968 + }, + { + "epoch": 2.7946826103549167, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.109060287475586, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8964145183563232, + "num_tokens": 853053999.0, + "step": 21969 + }, + { + "epoch": 2.7948098206335072, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.864105224609375, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.9008176922798157, + "num_tokens": 853096816.0, + "step": 21970 + }, + { + "epoch": 2.7949370309120978, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.841878890991211, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.907005250453949, + "num_tokens": 853139269.0, + "step": 21971 + }, + { + "epoch": 2.7950642411906883, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.954700469970703, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8995251655578613, + "num_tokens": 853182105.0, + "step": 21972 + }, + { + "epoch": 2.795191451469279, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.995253562927246, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.9015735387802124, + "num_tokens": 853212445.0, + "step": 21973 + }, + { + "epoch": 2.7953186617478694, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.899385452270508, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8971313238143921, + "num_tokens": 853260560.0, + "step": 21974 + }, + { + "epoch": 2.79544587202646, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.894262313842773, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9071360230445862, + "num_tokens": 853302311.0, + "step": 21975 + }, + { + "epoch": 2.79557308230505, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.046252250671387, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8971824645996094, + "num_tokens": 853338228.0, + "step": 21976 + }, + { + "epoch": 2.795700292583641, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.864572525024414, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8997896909713745, + "num_tokens": 853380782.0, + "step": 21977 + }, + { + "epoch": 2.795827502862231, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.90223217010498, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8931801915168762, + "num_tokens": 853429147.0, + "step": 21978 + }, + { + "epoch": 2.795954713140822, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.032899856567383, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8986390829086304, + "num_tokens": 853469317.0, + "step": 21979 + }, + { + "epoch": 2.796081923419412, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.086669921875, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8888449668884277, + "num_tokens": 853503500.0, + "step": 21980 + }, + { + "epoch": 2.796209133698003, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.981433868408203, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8792047500610352, + "num_tokens": 853548682.0, + "step": 21981 + }, + { + "epoch": 2.796336343976593, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.997727394104004, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8924158811569214, + "num_tokens": 853583625.0, + "step": 21982 + }, + { + "epoch": 2.7964635542551837, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.979934692382812, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8907474279403687, + "num_tokens": 853628022.0, + "step": 21983 + }, + { + "epoch": 2.796590764533774, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.947460174560547, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.9056158661842346, + "num_tokens": 853660478.0, + "step": 21984 + }, + { + "epoch": 2.7967179748123647, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.919918060302734, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8945112228393555, + "num_tokens": 853704961.0, + "step": 21985 + }, + { + "epoch": 2.7968451850909553, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.939043045043945, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8954452276229858, + "num_tokens": 853743220.0, + "step": 21986 + }, + { + "epoch": 2.796972395369546, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.98049545288086, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8952211737632751, + "num_tokens": 853784079.0, + "step": 21987 + }, + { + "epoch": 2.7970996056481363, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.913254737854004, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.9032765626907349, + "num_tokens": 853826580.0, + "step": 21988 + }, + { + "epoch": 2.797226815926727, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.959583282470703, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9029356837272644, + "num_tokens": 853867896.0, + "step": 21989 + }, + { + "epoch": 2.7973540262053174, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.041054725646973, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.9051205515861511, + "num_tokens": 853900670.0, + "step": 21990 + }, + { + "epoch": 2.797481236483908, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.977485656738281, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8902127742767334, + "num_tokens": 853940651.0, + "step": 21991 + }, + { + "epoch": 2.7976084467624984, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.08914852142334, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8872871994972229, + "num_tokens": 853977829.0, + "step": 21992 + }, + { + "epoch": 2.797735657041089, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.901089668273926, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.897597074508667, + "num_tokens": 854021989.0, + "step": 21993 + }, + { + "epoch": 2.7978628673196795, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.959168434143066, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.9018864035606384, + "num_tokens": 854056893.0, + "step": 21994 + }, + { + "epoch": 2.79799007759827, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.066055297851562, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8991678953170776, + "num_tokens": 854090773.0, + "step": 21995 + }, + { + "epoch": 2.7981172878768605, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.098018646240234, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.897682785987854, + "num_tokens": 854120102.0, + "step": 21996 + }, + { + "epoch": 2.798244498155451, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.892524719238281, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.9059501886367798, + "num_tokens": 854162544.0, + "step": 21997 + }, + { + "epoch": 2.7983717084340416, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.982333183288574, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8990461826324463, + "num_tokens": 854199383.0, + "step": 21998 + }, + { + "epoch": 2.798498918712632, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.968326568603516, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9066048264503479, + "num_tokens": 854233276.0, + "step": 21999 + }, + { + "epoch": 2.7986261289912227, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.912920951843262, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.9013533592224121, + "num_tokens": 854274585.0, + "step": 22000 + }, + { + "epoch": 2.7987533392698127, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.967570304870605, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8907904624938965, + "num_tokens": 854313874.0, + "step": 22001 + }, + { + "epoch": 2.7988805495484037, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.011305809020996, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8912752866744995, + "num_tokens": 854353236.0, + "step": 22002 + }, + { + "epoch": 2.799007759826994, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.869321823120117, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9092403054237366, + "num_tokens": 854391468.0, + "step": 22003 + }, + { + "epoch": 2.7991349701055848, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.008770942687988, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8989309072494507, + "num_tokens": 854429853.0, + "step": 22004 + }, + { + "epoch": 2.799262180384175, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.025732040405273, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8819258213043213, + "num_tokens": 854472871.0, + "step": 22005 + }, + { + "epoch": 2.799389390662766, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.893566131591797, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9034252166748047, + "num_tokens": 854508390.0, + "step": 22006 + }, + { + "epoch": 2.799516600941356, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.872782707214355, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9073558449745178, + "num_tokens": 854549241.0, + "step": 22007 + }, + { + "epoch": 2.7996438112199464, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.962784767150879, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8900467753410339, + "num_tokens": 854590956.0, + "step": 22008 + }, + { + "epoch": 2.799771021498537, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 9.99865436553955, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8956394195556641, + "num_tokens": 854631102.0, + "step": 22009 + }, + { + "epoch": 2.7998982317771275, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.031449317932129, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8831309080123901, + "num_tokens": 854666683.0, + "step": 22010 + }, + { + "epoch": 2.800025442055718, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.00674057006836, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8997306823730469, + "num_tokens": 854698537.0, + "step": 22011 + }, + { + "epoch": 2.8001526523343085, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.973837852478027, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9134838581085205, + "num_tokens": 854735318.0, + "step": 22012 + }, + { + "epoch": 2.800279862612899, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.921342849731445, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8924152851104736, + "num_tokens": 854777155.0, + "step": 22013 + }, + { + "epoch": 2.8004070728914896, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.0067720413208, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8951171040534973, + "num_tokens": 854817333.0, + "step": 22014 + }, + { + "epoch": 2.80053428317008, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.880534172058105, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.9049403667449951, + "num_tokens": 854860443.0, + "step": 22015 + }, + { + "epoch": 2.8006614934486707, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.018288612365723, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.893949031829834, + "num_tokens": 854901775.0, + "step": 22016 + }, + { + "epoch": 2.800788703727261, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.89917278289795, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9040271639823914, + "num_tokens": 854941013.0, + "step": 22017 + }, + { + "epoch": 2.8009159140058517, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.123477935791016, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8969839215278625, + "num_tokens": 854973531.0, + "step": 22018 + }, + { + "epoch": 2.8010431242844422, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.02987003326416, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9143257141113281, + "num_tokens": 855004522.0, + "step": 22019 + }, + { + "epoch": 2.8011703345630328, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.973381996154785, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.898804783821106, + "num_tokens": 855048083.0, + "step": 22020 + }, + { + "epoch": 2.8012975448416233, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.104881286621094, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8936536312103271, + "num_tokens": 855083722.0, + "step": 22021 + }, + { + "epoch": 2.801424755120214, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.031615257263184, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8984960913658142, + "num_tokens": 855125121.0, + "step": 22022 + }, + { + "epoch": 2.8015519653988044, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.861186027526855, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8971290588378906, + "num_tokens": 855164673.0, + "step": 22023 + }, + { + "epoch": 2.801679175677395, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.015726089477539, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8924202919006348, + "num_tokens": 855207225.0, + "step": 22024 + }, + { + "epoch": 2.8018063859559854, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.98958969116211, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8931801319122314, + "num_tokens": 855241661.0, + "step": 22025 + }, + { + "epoch": 2.8019335962345755, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.965651512145996, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.9005495309829712, + "num_tokens": 855280798.0, + "step": 22026 + }, + { + "epoch": 2.8020608065131665, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.950608253479004, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9037047028541565, + "num_tokens": 855320188.0, + "step": 22027 + }, + { + "epoch": 2.8021880167917566, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.155672073364258, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8888753652572632, + "num_tokens": 855354193.0, + "step": 22028 + }, + { + "epoch": 2.8023152270703475, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.010905265808105, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8971322774887085, + "num_tokens": 855391136.0, + "step": 22029 + }, + { + "epoch": 2.8024424373489376, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.947954177856445, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8848971128463745, + "num_tokens": 855434786.0, + "step": 22030 + }, + { + "epoch": 2.8025696476275286, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.05846118927002, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8941818475723267, + "num_tokens": 855472442.0, + "step": 22031 + }, + { + "epoch": 2.8026968579061187, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.011311531066895, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8877575397491455, + "num_tokens": 855510341.0, + "step": 22032 + }, + { + "epoch": 2.802824068184709, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.941228866577148, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9105759263038635, + "num_tokens": 855551374.0, + "step": 22033 + }, + { + "epoch": 2.8029512784632997, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.082231521606445, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8864081501960754, + "num_tokens": 855589115.0, + "step": 22034 + }, + { + "epoch": 2.8030784887418903, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.971782684326172, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9018562436103821, + "num_tokens": 855627983.0, + "step": 22035 + }, + { + "epoch": 2.803205699020481, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.965371131896973, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8967260122299194, + "num_tokens": 855666010.0, + "step": 22036 + }, + { + "epoch": 2.8033329092990713, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.95787239074707, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8913311958312988, + "num_tokens": 855708506.0, + "step": 22037 + }, + { + "epoch": 2.803460119577662, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.863581657409668, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9034445285797119, + "num_tokens": 855749009.0, + "step": 22038 + }, + { + "epoch": 2.8035873298562524, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.012670516967773, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.899084746837616, + "num_tokens": 855790877.0, + "step": 22039 + }, + { + "epoch": 2.803714540134843, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.980647087097168, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8900629878044128, + "num_tokens": 855827204.0, + "step": 22040 + }, + { + "epoch": 2.8038417504134334, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.000195503234863, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8972474336624146, + "num_tokens": 855860744.0, + "step": 22041 + }, + { + "epoch": 2.803968960692024, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.965479850769043, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8914934992790222, + "num_tokens": 855900061.0, + "step": 22042 + }, + { + "epoch": 2.8040961709706145, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.971979141235352, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9063340425491333, + "num_tokens": 855935685.0, + "step": 22043 + }, + { + "epoch": 2.804223381249205, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.985381126403809, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9126445055007935, + "num_tokens": 855968079.0, + "step": 22044 + }, + { + "epoch": 2.8043505915277955, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.89366340637207, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9082447290420532, + "num_tokens": 856011679.0, + "step": 22045 + }, + { + "epoch": 2.804477801806386, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.908913612365723, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8967458009719849, + "num_tokens": 856052570.0, + "step": 22046 + }, + { + "epoch": 2.8046050120849766, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.113125801086426, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8987837433815002, + "num_tokens": 856085734.0, + "step": 22047 + }, + { + "epoch": 2.804732222363567, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.022892951965332, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8918190002441406, + "num_tokens": 856125383.0, + "step": 22048 + }, + { + "epoch": 2.8048594326421576, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.89163589477539, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9115735292434692, + "num_tokens": 856165656.0, + "step": 22049 + }, + { + "epoch": 2.804986642920748, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.01848030090332, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.9059057235717773, + "num_tokens": 856197891.0, + "step": 22050 + }, + { + "epoch": 2.8051138531993383, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.006194114685059, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8960025906562805, + "num_tokens": 856235789.0, + "step": 22051 + }, + { + "epoch": 2.8052410634779292, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.99071979522705, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8958166837692261, + "num_tokens": 856274206.0, + "step": 22052 + }, + { + "epoch": 2.8053682737565193, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.965795516967773, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8979215025901794, + "num_tokens": 856313042.0, + "step": 22053 + }, + { + "epoch": 2.8054954840351103, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.118169784545898, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8958518505096436, + "num_tokens": 856344353.0, + "step": 22054 + }, + { + "epoch": 2.8056226943137004, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.87075424194336, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.9050244092941284, + "num_tokens": 856384678.0, + "step": 22055 + }, + { + "epoch": 2.805749904592291, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.94970417022705, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8963683843612671, + "num_tokens": 856426990.0, + "step": 22056 + }, + { + "epoch": 2.8058771148708814, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.973641395568848, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.9035230875015259, + "num_tokens": 856465411.0, + "step": 22057 + }, + { + "epoch": 2.806004325149472, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.00194263458252, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8986762762069702, + "num_tokens": 856510446.0, + "step": 22058 + }, + { + "epoch": 2.8061315354280625, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.913787841796875, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8949698209762573, + "num_tokens": 856555251.0, + "step": 22059 + }, + { + "epoch": 2.806258745706653, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.928282737731934, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9127428531646729, + "num_tokens": 856594977.0, + "step": 22060 + }, + { + "epoch": 2.8063859559852435, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.039505958557129, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8990262150764465, + "num_tokens": 856631953.0, + "step": 22061 + }, + { + "epoch": 2.806513166263834, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.950708389282227, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9056093692779541, + "num_tokens": 856665312.0, + "step": 22062 + }, + { + "epoch": 2.8066403765424246, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.000555038452148, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8868861198425293, + "num_tokens": 856708113.0, + "step": 22063 + }, + { + "epoch": 2.806767586821015, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.987547874450684, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.9015109539031982, + "num_tokens": 856744969.0, + "step": 22064 + }, + { + "epoch": 2.8068947970996057, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.632110595703125e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.993886947631836, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8874894380569458, + "num_tokens": 856788151.0, + "step": 22065 + }, + { + "epoch": 2.807022007378196, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.103316307067871, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.9002910852432251, + "num_tokens": 856823124.0, + "step": 22066 + }, + { + "epoch": 2.8071492176567867, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.906872749328613, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9178762435913086, + "num_tokens": 856864044.0, + "step": 22067 + }, + { + "epoch": 2.8072764279353772, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.957182884216309, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8974242210388184, + "num_tokens": 856901379.0, + "step": 22068 + }, + { + "epoch": 2.8074036382139678, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.968659400939941, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8983747363090515, + "num_tokens": 856942776.0, + "step": 22069 + }, + { + "epoch": 2.8075308484925583, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.988893508911133, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8817875385284424, + "num_tokens": 856986645.0, + "step": 22070 + }, + { + "epoch": 2.807658058771149, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.905454635620117, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8875007629394531, + "num_tokens": 857028439.0, + "step": 22071 + }, + { + "epoch": 2.8077852690497394, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.988435745239258, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.894566535949707, + "num_tokens": 857071469.0, + "step": 22072 + }, + { + "epoch": 2.80791247932833, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.967731475830078, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8970257639884949, + "num_tokens": 857107571.0, + "step": 22073 + }, + { + "epoch": 2.80803968960692, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.013042449951172, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8949114084243774, + "num_tokens": 857148606.0, + "step": 22074 + }, + { + "epoch": 2.808166899885511, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.057290077209473, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9037594199180603, + "num_tokens": 857180295.0, + "step": 22075 + }, + { + "epoch": 2.808294110164101, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.9682035446167, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8864631056785583, + "num_tokens": 857221014.0, + "step": 22076 + }, + { + "epoch": 2.808421320442692, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.024931907653809, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8979744911193848, + "num_tokens": 857259283.0, + "step": 22077 + }, + { + "epoch": 2.808548530721282, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.971588134765625, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8876588344573975, + "num_tokens": 857298792.0, + "step": 22078 + }, + { + "epoch": 2.808675740999873, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.070089340209961, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.885234534740448, + "num_tokens": 857336952.0, + "step": 22079 + }, + { + "epoch": 2.808802951278463, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.886134147644043, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8966622352600098, + "num_tokens": 857379407.0, + "step": 22080 + }, + { + "epoch": 2.8089301615570537, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.941407203674316, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8871840238571167, + "num_tokens": 857418754.0, + "step": 22081 + }, + { + "epoch": 2.809057371835644, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.949542999267578, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8916024565696716, + "num_tokens": 857458364.0, + "step": 22082 + }, + { + "epoch": 2.8091845821142347, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.949477195739746, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8997313976287842, + "num_tokens": 857499213.0, + "step": 22083 + }, + { + "epoch": 2.8093117923928252, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.994101524353027, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8941277265548706, + "num_tokens": 857539403.0, + "step": 22084 + }, + { + "epoch": 2.8094390026714158, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.996084213256836, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8950870633125305, + "num_tokens": 857576907.0, + "step": 22085 + }, + { + "epoch": 2.8095662129500063, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.021808624267578, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8937178254127502, + "num_tokens": 857613377.0, + "step": 22086 + }, + { + "epoch": 2.809693423228597, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.880087852478027, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.9008595943450928, + "num_tokens": 857653245.0, + "step": 22087 + }, + { + "epoch": 2.8098206335071874, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.074167251586914, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8944343328475952, + "num_tokens": 857688066.0, + "step": 22088 + }, + { + "epoch": 2.809947843785778, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.922794342041016, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9140311479568481, + "num_tokens": 857725414.0, + "step": 22089 + }, + { + "epoch": 2.8100750540643684, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.911646842956543, + "learning_rate": 1e-06, + "loss": 0.2453, + "mean_token_accuracy": 0.9158987998962402, + "num_tokens": 857763198.0, + "step": 22090 + }, + { + "epoch": 2.810202264342959, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.051484107971191, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8986800312995911, + "num_tokens": 857796768.0, + "step": 22091 + }, + { + "epoch": 2.8103294746215495, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.044715881347656, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8926424980163574, + "num_tokens": 857832920.0, + "step": 22092 + }, + { + "epoch": 2.81045668490014, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.954233169555664, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9017647504806519, + "num_tokens": 857873691.0, + "step": 22093 + }, + { + "epoch": 2.8105838951787305, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.040928840637207, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8971815705299377, + "num_tokens": 857908614.0, + "step": 22094 + }, + { + "epoch": 2.810711105457321, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.030254364013672, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8979502320289612, + "num_tokens": 857947550.0, + "step": 22095 + }, + { + "epoch": 2.8108383157359116, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.96098804473877, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9041399955749512, + "num_tokens": 857985321.0, + "step": 22096 + }, + { + "epoch": 2.810965526014502, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.058977127075195, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8972957134246826, + "num_tokens": 858022697.0, + "step": 22097 + }, + { + "epoch": 2.8110927362930926, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.938810348510742, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8937922716140747, + "num_tokens": 858061345.0, + "step": 22098 + }, + { + "epoch": 2.8112199465716827, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.118112564086914, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8913033604621887, + "num_tokens": 858094386.0, + "step": 22099 + }, + { + "epoch": 2.8113471568502737, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.933497428894043, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.896337628364563, + "num_tokens": 858137987.0, + "step": 22100 + }, + { + "epoch": 2.811474367128864, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.951080322265625, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9100826382637024, + "num_tokens": 858173119.0, + "step": 22101 + }, + { + "epoch": 2.8116015774074548, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.911165237426758, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.908571720123291, + "num_tokens": 858214895.0, + "step": 22102 + }, + { + "epoch": 2.811728787686045, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.00704288482666, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8927180767059326, + "num_tokens": 858250920.0, + "step": 22103 + }, + { + "epoch": 2.811855997964636, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.893376350402832, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9118335247039795, + "num_tokens": 858292257.0, + "step": 22104 + }, + { + "epoch": 2.811983208243226, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.882983207702637, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8998916149139404, + "num_tokens": 858337345.0, + "step": 22105 + }, + { + "epoch": 2.8121104185218164, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.001216888427734, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8900840282440186, + "num_tokens": 858378968.0, + "step": 22106 + }, + { + "epoch": 2.812237628800407, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.089655876159668, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8982744216918945, + "num_tokens": 858413667.0, + "step": 22107 + }, + { + "epoch": 2.8123648390789975, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.070411682128906, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8864585161209106, + "num_tokens": 858451862.0, + "step": 22108 + }, + { + "epoch": 2.812492049357588, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.906513214111328, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9066349864006042, + "num_tokens": 858485920.0, + "step": 22109 + }, + { + "epoch": 2.8126192596361785, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.006717681884766, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.9010821580886841, + "num_tokens": 858525958.0, + "step": 22110 + }, + { + "epoch": 2.812746469914769, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.017724990844727, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.889677882194519, + "num_tokens": 858566141.0, + "step": 22111 + }, + { + "epoch": 2.8128736801933596, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.924260139465332, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.9016948938369751, + "num_tokens": 858608736.0, + "step": 22112 + }, + { + "epoch": 2.81300089047195, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.981537818908691, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8915932774543762, + "num_tokens": 858651668.0, + "step": 22113 + }, + { + "epoch": 2.8131281007505406, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.905393600463867, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8975090384483337, + "num_tokens": 858696714.0, + "step": 22114 + }, + { + "epoch": 2.813255311029131, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.079035758972168, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8950752019882202, + "num_tokens": 858734638.0, + "step": 22115 + }, + { + "epoch": 2.8133825213077217, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.017380714416504, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9056077599525452, + "num_tokens": 858768764.0, + "step": 22116 + }, + { + "epoch": 2.8135097315863122, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.961618423461914, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8936035633087158, + "num_tokens": 858807955.0, + "step": 22117 + }, + { + "epoch": 2.8136369418649028, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.949131965637207, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8977676033973694, + "num_tokens": 858854906.0, + "step": 22118 + }, + { + "epoch": 2.8137641521434933, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.051704406738281, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9089407920837402, + "num_tokens": 858883179.0, + "step": 22119 + }, + { + "epoch": 2.813891362422084, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.919820785522461, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8982868790626526, + "num_tokens": 858920980.0, + "step": 22120 + }, + { + "epoch": 2.8140185727006743, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.098018646240234, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8945697546005249, + "num_tokens": 858963490.0, + "step": 22121 + }, + { + "epoch": 2.814145782979265, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.886677742004395, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.9019114971160889, + "num_tokens": 859004535.0, + "step": 22122 + }, + { + "epoch": 2.8142729932578554, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.988138198852539, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9103666543960571, + "num_tokens": 859036316.0, + "step": 22123 + }, + { + "epoch": 2.8144002035364455, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.98082447052002, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8860164880752563, + "num_tokens": 859072969.0, + "step": 22124 + }, + { + "epoch": 2.8145274138150365, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.989988327026367, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9030653238296509, + "num_tokens": 859107946.0, + "step": 22125 + }, + { + "epoch": 2.8146546240936265, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.98849868774414, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8946411609649658, + "num_tokens": 859146158.0, + "step": 22126 + }, + { + "epoch": 2.8147818343722175, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.034399032592773, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8767440319061279, + "num_tokens": 859185195.0, + "step": 22127 + }, + { + "epoch": 2.8149090446508076, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.884507179260254, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9056037068367004, + "num_tokens": 859229869.0, + "step": 22128 + }, + { + "epoch": 2.815036254929398, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.920950889587402, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8976037502288818, + "num_tokens": 859270227.0, + "step": 22129 + }, + { + "epoch": 2.8151634652079887, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.948686599731445, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8994826674461365, + "num_tokens": 859311006.0, + "step": 22130 + }, + { + "epoch": 2.815290675486579, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.92736530303955, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9059795141220093, + "num_tokens": 859352256.0, + "step": 22131 + }, + { + "epoch": 2.8154178857651697, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.008500099182129, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8986034393310547, + "num_tokens": 859390102.0, + "step": 22132 + }, + { + "epoch": 2.8155450960437602, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.007381439208984, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8978964686393738, + "num_tokens": 859428843.0, + "step": 22133 + }, + { + "epoch": 2.8156723063223508, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.896502494812012, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.904499888420105, + "num_tokens": 859473932.0, + "step": 22134 + }, + { + "epoch": 2.8157995166009413, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.033751487731934, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8984060287475586, + "num_tokens": 859510406.0, + "step": 22135 + }, + { + "epoch": 2.815926726879532, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.96049690246582, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8999018669128418, + "num_tokens": 859549903.0, + "step": 22136 + }, + { + "epoch": 2.8160539371581224, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.976572036743164, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8998656272888184, + "num_tokens": 859586001.0, + "step": 22137 + }, + { + "epoch": 2.816181147436713, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.035161972045898, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8997955918312073, + "num_tokens": 859623361.0, + "step": 22138 + }, + { + "epoch": 2.8163083577153034, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.966357231140137, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.9011139869689941, + "num_tokens": 859662128.0, + "step": 22139 + }, + { + "epoch": 2.816435567993894, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.068598747253418, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8923341035842896, + "num_tokens": 859696678.0, + "step": 22140 + }, + { + "epoch": 2.8165627782724845, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.9677095413208, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9007081985473633, + "num_tokens": 859736792.0, + "step": 22141 + }, + { + "epoch": 2.816689988551075, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.985193252563477, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.889549970626831, + "num_tokens": 859777328.0, + "step": 22142 + }, + { + "epoch": 2.8168171988296655, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.006095886230469, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8912578225135803, + "num_tokens": 859814627.0, + "step": 22143 + }, + { + "epoch": 2.816944409108256, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.986209869384766, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9113472104072571, + "num_tokens": 859850689.0, + "step": 22144 + }, + { + "epoch": 2.8170716193868466, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.980441093444824, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.9016932249069214, + "num_tokens": 859888645.0, + "step": 22145 + }, + { + "epoch": 2.817198829665437, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.064547538757324, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8886198401451111, + "num_tokens": 859925182.0, + "step": 22146 + }, + { + "epoch": 2.8173260399440276, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.05159854888916, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8876180648803711, + "num_tokens": 859966077.0, + "step": 22147 + }, + { + "epoch": 2.817453250222618, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.912164688110352, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9092215895652771, + "num_tokens": 860004705.0, + "step": 22148 + }, + { + "epoch": 2.8175804605012083, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.004122734069824, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9031782150268555, + "num_tokens": 860041864.0, + "step": 22149 + }, + { + "epoch": 2.8177076707797992, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.072892189025879, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.897780179977417, + "num_tokens": 860080486.0, + "step": 22150 + }, + { + "epoch": 2.8178348810583893, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.94554615020752, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.9017770886421204, + "num_tokens": 860123084.0, + "step": 22151 + }, + { + "epoch": 2.8179620913369803, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.01661491394043, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.9016345143318176, + "num_tokens": 860161885.0, + "step": 22152 + }, + { + "epoch": 2.8180893016155704, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.045862197875977, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9092431664466858, + "num_tokens": 860200666.0, + "step": 22153 + }, + { + "epoch": 2.818216511894161, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.928011894226074, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9079323410987854, + "num_tokens": 860242599.0, + "step": 22154 + }, + { + "epoch": 2.8183437221727514, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.153816223144531, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.888964831829071, + "num_tokens": 860276548.0, + "step": 22155 + }, + { + "epoch": 2.818470932451342, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.065555572509766, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8929864168167114, + "num_tokens": 860311923.0, + "step": 22156 + }, + { + "epoch": 2.8185981427299325, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.890054702758789, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.9030192494392395, + "num_tokens": 860352677.0, + "step": 22157 + }, + { + "epoch": 2.818725353008523, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.007343292236328, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8941676020622253, + "num_tokens": 860393635.0, + "step": 22158 + }, + { + "epoch": 2.8188525632871135, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.988499641418457, + "learning_rate": 1e-06, + "loss": 0.2638, + "mean_token_accuracy": 0.9102463126182556, + "num_tokens": 860429523.0, + "step": 22159 + }, + { + "epoch": 2.818979773565704, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.911476135253906, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9030025601387024, + "num_tokens": 860469726.0, + "step": 22160 + }, + { + "epoch": 2.8191069838442946, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.990013122558594, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.9037361741065979, + "num_tokens": 860507450.0, + "step": 22161 + }, + { + "epoch": 2.819234194122885, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.135930061340332, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8943341970443726, + "num_tokens": 860543757.0, + "step": 22162 + }, + { + "epoch": 2.8193614044014756, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.03133487701416, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8917713165283203, + "num_tokens": 860581236.0, + "step": 22163 + }, + { + "epoch": 2.819488614680066, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.978569984436035, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8988136053085327, + "num_tokens": 860623882.0, + "step": 22164 + }, + { + "epoch": 2.8196158249586567, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.025415420532227, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8979179859161377, + "num_tokens": 860658466.0, + "step": 22165 + }, + { + "epoch": 2.8197430352372472, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.014752388000488, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8926987648010254, + "num_tokens": 860699642.0, + "step": 22166 + }, + { + "epoch": 2.8198702455158378, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.967201232910156, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.9004068374633789, + "num_tokens": 860736060.0, + "step": 22167 + }, + { + "epoch": 2.8199974557944283, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.99915885925293, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.9069141149520874, + "num_tokens": 860777196.0, + "step": 22168 + }, + { + "epoch": 2.820124666073019, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.023966789245605, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8921549320220947, + "num_tokens": 860813895.0, + "step": 22169 + }, + { + "epoch": 2.8202518763516093, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.9966402053833, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8965147137641907, + "num_tokens": 860849180.0, + "step": 22170 + }, + { + "epoch": 2.8203790866302, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.990700721740723, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9112096428871155, + "num_tokens": 860885324.0, + "step": 22171 + }, + { + "epoch": 2.82050629690879, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.90443229675293, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8971618413925171, + "num_tokens": 860925182.0, + "step": 22172 + }, + { + "epoch": 2.820633507187381, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.961430549621582, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8872207999229431, + "num_tokens": 860966465.0, + "step": 22173 + }, + { + "epoch": 2.820760717465971, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.010676383972168, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.90053391456604, + "num_tokens": 861001531.0, + "step": 22174 + }, + { + "epoch": 2.820887927744562, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.92194652557373, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8975645899772644, + "num_tokens": 861039708.0, + "step": 22175 + }, + { + "epoch": 2.821015138023152, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.932491302490234, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9121404886245728, + "num_tokens": 861081208.0, + "step": 22176 + }, + { + "epoch": 2.821142348301743, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.912679672241211, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9092189073562622, + "num_tokens": 861120436.0, + "step": 22177 + }, + { + "epoch": 2.821269558580333, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.926458358764648, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8952745795249939, + "num_tokens": 861157673.0, + "step": 22178 + }, + { + "epoch": 2.8213967688589237, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.989591598510742, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.9002183675765991, + "num_tokens": 861197191.0, + "step": 22179 + }, + { + "epoch": 2.821523979137514, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.136838912963867, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8860500454902649, + "num_tokens": 861229200.0, + "step": 22180 + }, + { + "epoch": 2.8216511894161047, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.931568145751953, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8878777623176575, + "num_tokens": 861275236.0, + "step": 22181 + }, + { + "epoch": 2.8217783996946952, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.957331657409668, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8867225646972656, + "num_tokens": 861317087.0, + "step": 22182 + }, + { + "epoch": 2.8219056099732858, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.091514587402344, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8886363506317139, + "num_tokens": 861357369.0, + "step": 22183 + }, + { + "epoch": 2.8220328202518763, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.079110145568848, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8922724723815918, + "num_tokens": 861392903.0, + "step": 22184 + }, + { + "epoch": 2.822160030530467, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.957152366638184, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8974977731704712, + "num_tokens": 861429063.0, + "step": 22185 + }, + { + "epoch": 2.8222872408090574, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.965644836425781, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8950436115264893, + "num_tokens": 861466977.0, + "step": 22186 + }, + { + "epoch": 2.822414451087648, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.875482559204102, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.90262371301651, + "num_tokens": 861509359.0, + "step": 22187 + }, + { + "epoch": 2.8225416613662384, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.951313018798828, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9168509244918823, + "num_tokens": 861546314.0, + "step": 22188 + }, + { + "epoch": 2.822668871644829, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.022518157958984, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.890764594078064, + "num_tokens": 861588102.0, + "step": 22189 + }, + { + "epoch": 2.8227960819234195, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.09010124206543, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8821668028831482, + "num_tokens": 861626609.0, + "step": 22190 + }, + { + "epoch": 2.82292329220201, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.10834789276123, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8877779245376587, + "num_tokens": 861671456.0, + "step": 22191 + }, + { + "epoch": 2.8230505024806005, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.911192893981934, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9128979444503784, + "num_tokens": 861708815.0, + "step": 22192 + }, + { + "epoch": 2.823177712759191, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.093880653381348, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8920553922653198, + "num_tokens": 861747965.0, + "step": 22193 + }, + { + "epoch": 2.8233049230377816, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.033751487731934, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8898194432258606, + "num_tokens": 861786849.0, + "step": 22194 + }, + { + "epoch": 2.823432133316372, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.987208366394043, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.9011099338531494, + "num_tokens": 861829059.0, + "step": 22195 + }, + { + "epoch": 2.8235593435949626, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.989466667175293, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8854684829711914, + "num_tokens": 861871071.0, + "step": 22196 + }, + { + "epoch": 2.8236865538735527, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.985180854797363, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.9010683298110962, + "num_tokens": 861907560.0, + "step": 22197 + }, + { + "epoch": 2.8238137641521437, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.994571685791016, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.887294590473175, + "num_tokens": 861947163.0, + "step": 22198 + }, + { + "epoch": 2.8239409744307338, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.081194877624512, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.89261394739151, + "num_tokens": 861988582.0, + "step": 22199 + }, + { + "epoch": 2.8240681847093247, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.079299926757812, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8773990869522095, + "num_tokens": 862030192.0, + "step": 22200 + }, + { + "epoch": 2.824195394987915, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.967169761657715, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8950152397155762, + "num_tokens": 862071032.0, + "step": 22201 + }, + { + "epoch": 2.824322605266506, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.994202613830566, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8980445265769958, + "num_tokens": 862110282.0, + "step": 22202 + }, + { + "epoch": 2.824449815545096, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.969005584716797, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8896206021308899, + "num_tokens": 862149636.0, + "step": 22203 + }, + { + "epoch": 2.8245770258236864, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.975552558898926, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8943080902099609, + "num_tokens": 862186750.0, + "step": 22204 + }, + { + "epoch": 2.824704236102277, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.921337127685547, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.9004138708114624, + "num_tokens": 862228926.0, + "step": 22205 + }, + { + "epoch": 2.8248314463808675, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.015941619873047, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.9019690155982971, + "num_tokens": 862263267.0, + "step": 22206 + }, + { + "epoch": 2.824958656659458, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.875039100646973, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.9025999307632446, + "num_tokens": 862302021.0, + "step": 22207 + }, + { + "epoch": 2.8250858669380485, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.861961364746094, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9099213480949402, + "num_tokens": 862349408.0, + "step": 22208 + }, + { + "epoch": 2.825213077216639, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.179582595825195, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8893880844116211, + "num_tokens": 862385371.0, + "step": 22209 + }, + { + "epoch": 2.8253402874952296, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.000202178955078, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8933556079864502, + "num_tokens": 862424039.0, + "step": 22210 + }, + { + "epoch": 2.82546749777382, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.989988327026367, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8968169093132019, + "num_tokens": 862462199.0, + "step": 22211 + }, + { + "epoch": 2.8255947080524106, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.048195838928223, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8884303569793701, + "num_tokens": 862501158.0, + "step": 22212 + }, + { + "epoch": 2.825721918331001, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.93322467803955, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8898574709892273, + "num_tokens": 862544124.0, + "step": 22213 + }, + { + "epoch": 2.8258491286095917, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.95285701751709, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9103755950927734, + "num_tokens": 862580605.0, + "step": 22214 + }, + { + "epoch": 2.8259763388881822, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.99142837524414, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8933819532394409, + "num_tokens": 862615394.0, + "step": 22215 + }, + { + "epoch": 2.8261035491667728, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.854265213012695, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9070619344711304, + "num_tokens": 862657971.0, + "step": 22216 + }, + { + "epoch": 2.8262307594453633, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.03735637664795, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.9001856446266174, + "num_tokens": 862695950.0, + "step": 22217 + }, + { + "epoch": 2.826357969723954, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.934682846069336, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9099802374839783, + "num_tokens": 862729737.0, + "step": 22218 + }, + { + "epoch": 2.8264851800025443, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.679794311523438e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.133431434631348, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.89197838306427, + "num_tokens": 862771198.0, + "step": 22219 + }, + { + "epoch": 2.826612390281135, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.995000839233398, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8978347778320312, + "num_tokens": 862804842.0, + "step": 22220 + }, + { + "epoch": 2.8267396005597254, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.919694900512695, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8950974941253662, + "num_tokens": 862842955.0, + "step": 22221 + }, + { + "epoch": 2.8268668108383155, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.958474159240723, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.900364100933075, + "num_tokens": 862884668.0, + "step": 22222 + }, + { + "epoch": 2.8269940211169065, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.927928924560547, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.894599199295044, + "num_tokens": 862927807.0, + "step": 22223 + }, + { + "epoch": 2.8271212313954965, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.978850364685059, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.9058173894882202, + "num_tokens": 862971121.0, + "step": 22224 + }, + { + "epoch": 2.8272484416740875, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.953167915344238, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8890801668167114, + "num_tokens": 863015771.0, + "step": 22225 + }, + { + "epoch": 2.8273756519526776, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.156216621398926, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8914502859115601, + "num_tokens": 863048705.0, + "step": 22226 + }, + { + "epoch": 2.827502862231268, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.959806442260742, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.892612636089325, + "num_tokens": 863091422.0, + "step": 22227 + }, + { + "epoch": 2.8276300725098586, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.887039184570312, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9104835987091064, + "num_tokens": 863127790.0, + "step": 22228 + }, + { + "epoch": 2.827757282788449, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.249083518981934, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8948044776916504, + "num_tokens": 863171406.0, + "step": 22229 + }, + { + "epoch": 2.8278844930670397, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.896800994873047, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8983184099197388, + "num_tokens": 863221352.0, + "step": 22230 + }, + { + "epoch": 2.8280117033456302, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 10.04209041595459, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8990325927734375, + "num_tokens": 863258905.0, + "step": 22231 + }, + { + "epoch": 2.8281389136242208, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.972193717956543, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.9008611440658569, + "num_tokens": 863301125.0, + "step": 22232 + }, + { + "epoch": 2.8282661239028113, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.896844863891602, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8974649906158447, + "num_tokens": 863344261.0, + "step": 22233 + }, + { + "epoch": 2.828393334181402, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.111238479614258, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.9005655646324158, + "num_tokens": 863377598.0, + "step": 22234 + }, + { + "epoch": 2.8285205444599923, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.989495277404785, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8927298784255981, + "num_tokens": 863411311.0, + "step": 22235 + }, + { + "epoch": 2.828647754738583, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.01220417022705, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8957420587539673, + "num_tokens": 863445403.0, + "step": 22236 + }, + { + "epoch": 2.8287749650171734, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.93907642364502, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.9015892148017883, + "num_tokens": 863482410.0, + "step": 22237 + }, + { + "epoch": 2.828902175295764, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.919036865234375, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8864536285400391, + "num_tokens": 863524144.0, + "step": 22238 + }, + { + "epoch": 2.8290293855743545, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.909297943115234, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8976128101348877, + "num_tokens": 863564983.0, + "step": 22239 + }, + { + "epoch": 2.829156595852945, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.919517517089844, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8979558944702148, + "num_tokens": 863607761.0, + "step": 22240 + }, + { + "epoch": 2.8292838061315355, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.005915641784668, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8993535041809082, + "num_tokens": 863641558.0, + "step": 22241 + }, + { + "epoch": 2.829411016410126, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.051435470581055, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8881795406341553, + "num_tokens": 863678953.0, + "step": 22242 + }, + { + "epoch": 2.8295382266887166, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.967988967895508, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8932523727416992, + "num_tokens": 863716426.0, + "step": 22243 + }, + { + "epoch": 2.829665436967307, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.904070854187012, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8987241387367249, + "num_tokens": 863756468.0, + "step": 22244 + }, + { + "epoch": 2.8297926472458976, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.904037475585938, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8983745574951172, + "num_tokens": 863802890.0, + "step": 22245 + }, + { + "epoch": 2.829919857524488, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.867610931396484, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9108425378799438, + "num_tokens": 863844438.0, + "step": 22246 + }, + { + "epoch": 2.8300470678030782, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.000283241271973, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8959134817123413, + "num_tokens": 863890433.0, + "step": 22247 + }, + { + "epoch": 2.830174278081669, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.097745895385742, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8895272016525269, + "num_tokens": 863924110.0, + "step": 22248 + }, + { + "epoch": 2.8303014883602593, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.986332893371582, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8861976861953735, + "num_tokens": 863963333.0, + "step": 22249 + }, + { + "epoch": 2.8304286986388503, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.074564933776855, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.911026656627655, + "num_tokens": 863991715.0, + "step": 22250 + }, + { + "epoch": 2.8305559089174404, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.022527694702148, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8948702812194824, + "num_tokens": 864026714.0, + "step": 22251 + }, + { + "epoch": 2.830683119196031, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.003192901611328, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8989486694335938, + "num_tokens": 864061635.0, + "step": 22252 + }, + { + "epoch": 2.8308103294746214, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.987194061279297, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8886755704879761, + "num_tokens": 864099714.0, + "step": 22253 + }, + { + "epoch": 2.830937539753212, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.948067665100098, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.9009853005409241, + "num_tokens": 864136519.0, + "step": 22254 + }, + { + "epoch": 2.8310647500318025, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.052105903625488, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.9000062942504883, + "num_tokens": 864175908.0, + "step": 22255 + }, + { + "epoch": 2.831191960310393, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.956950187683105, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9022558927536011, + "num_tokens": 864213942.0, + "step": 22256 + }, + { + "epoch": 2.8313191705889835, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.563963890075684, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8976564407348633, + "num_tokens": 864251419.0, + "step": 22257 + }, + { + "epoch": 2.831446380867574, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 9.960015296936035, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.910774827003479, + "num_tokens": 864287020.0, + "step": 22258 + }, + { + "epoch": 2.8315735911461646, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.919862747192383, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9047290086746216, + "num_tokens": 864329211.0, + "step": 22259 + }, + { + "epoch": 2.831700801424755, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.911513328552246, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8997273445129395, + "num_tokens": 864369944.0, + "step": 22260 + }, + { + "epoch": 2.8318280117033456, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.987553596496582, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.9016450047492981, + "num_tokens": 864408710.0, + "step": 22261 + }, + { + "epoch": 2.831955221981936, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.049823760986328, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8984488248825073, + "num_tokens": 864440389.0, + "step": 22262 + }, + { + "epoch": 2.8320824322605267, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.9727783203125, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.9000846147537231, + "num_tokens": 864480131.0, + "step": 22263 + }, + { + "epoch": 2.8322096425391172, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.029934883117676, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.9063456058502197, + "num_tokens": 864517801.0, + "step": 22264 + }, + { + "epoch": 2.8323368528177078, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.016070365905762, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9103248119354248, + "num_tokens": 864553764.0, + "step": 22265 + }, + { + "epoch": 2.8324640630962983, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.10101318359375, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.889475405216217, + "num_tokens": 864592427.0, + "step": 22266 + }, + { + "epoch": 2.832591273374889, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.926830291748047, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.908362865447998, + "num_tokens": 864631406.0, + "step": 22267 + }, + { + "epoch": 2.8327184836534793, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.949116706848145, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.902694582939148, + "num_tokens": 864671832.0, + "step": 22268 + }, + { + "epoch": 2.83284569393207, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.057147979736328, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9067823886871338, + "num_tokens": 864711107.0, + "step": 22269 + }, + { + "epoch": 2.83297290421066, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.035811424255371, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9052786231040955, + "num_tokens": 864743992.0, + "step": 22270 + }, + { + "epoch": 2.833100114489251, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.058110237121582, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8983349800109863, + "num_tokens": 864781151.0, + "step": 22271 + }, + { + "epoch": 2.833227324767841, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.14085865020752, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.9068681001663208, + "num_tokens": 864813888.0, + "step": 22272 + }, + { + "epoch": 2.833354535046432, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.022808074951172, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9111063480377197, + "num_tokens": 864854627.0, + "step": 22273 + }, + { + "epoch": 2.833481745325022, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.961697578430176, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8805035352706909, + "num_tokens": 864898692.0, + "step": 22274 + }, + { + "epoch": 2.833608955603613, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.075355529785156, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8930720090866089, + "num_tokens": 864938654.0, + "step": 22275 + }, + { + "epoch": 2.833736165882203, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.989845275878906, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8999556303024292, + "num_tokens": 864978328.0, + "step": 22276 + }, + { + "epoch": 2.8338633761607936, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.07115650177002, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8825750350952148, + "num_tokens": 865017936.0, + "step": 22277 + }, + { + "epoch": 2.833990586439384, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.968927383422852, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8989314436912537, + "num_tokens": 865054870.0, + "step": 22278 + }, + { + "epoch": 2.8341177967179747, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.007589340209961, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8948366641998291, + "num_tokens": 865094744.0, + "step": 22279 + }, + { + "epoch": 2.8342450069965652, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.060197830200195, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8893272876739502, + "num_tokens": 865134033.0, + "step": 22280 + }, + { + "epoch": 2.8343722172751558, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.042611122131348, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8978139162063599, + "num_tokens": 865171133.0, + "step": 22281 + }, + { + "epoch": 2.8344994275537463, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.019996643066406, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8933559656143188, + "num_tokens": 865213211.0, + "step": 22282 + }, + { + "epoch": 2.834626637832337, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.024941444396973, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8934464454650879, + "num_tokens": 865254600.0, + "step": 22283 + }, + { + "epoch": 2.8347538481109273, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.023796081542969, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.902710497379303, + "num_tokens": 865291569.0, + "step": 22284 + }, + { + "epoch": 2.834881058389518, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.905613899230957, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9075717926025391, + "num_tokens": 865332909.0, + "step": 22285 + }, + { + "epoch": 2.8350082686681084, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.07288646697998, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.9009579420089722, + "num_tokens": 865366738.0, + "step": 22286 + }, + { + "epoch": 2.835135478946699, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.109960556030273, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8905044198036194, + "num_tokens": 865402261.0, + "step": 22287 + }, + { + "epoch": 2.8352626892252895, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.023957252502441, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8961402177810669, + "num_tokens": 865437107.0, + "step": 22288 + }, + { + "epoch": 2.83538989950388, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.024914741516113, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8971024751663208, + "num_tokens": 865473895.0, + "step": 22289 + }, + { + "epoch": 2.8355171097824705, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.081307411193848, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8860204219818115, + "num_tokens": 865518396.0, + "step": 22290 + }, + { + "epoch": 2.835644320061061, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.069750785827637, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.9064663052558899, + "num_tokens": 865553860.0, + "step": 22291 + }, + { + "epoch": 2.8357715303396516, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.918984413146973, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9031753540039062, + "num_tokens": 865595165.0, + "step": 22292 + }, + { + "epoch": 2.835898740618242, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.950063705444336, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9069031476974487, + "num_tokens": 865634154.0, + "step": 22293 + }, + { + "epoch": 2.8360259508968326, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.022122383117676, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.903423011302948, + "num_tokens": 865675931.0, + "step": 22294 + }, + { + "epoch": 2.8361531611754227, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.97359561920166, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8911013603210449, + "num_tokens": 865716774.0, + "step": 22295 + }, + { + "epoch": 2.8362803714540137, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.148870468139648, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.9044249057769775, + "num_tokens": 865756168.0, + "step": 22296 + }, + { + "epoch": 2.8364075817326038, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.009109497070312, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9032679796218872, + "num_tokens": 865791170.0, + "step": 22297 + }, + { + "epoch": 2.8365347920111947, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.079856872558594, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8929592370986938, + "num_tokens": 865830375.0, + "step": 22298 + }, + { + "epoch": 2.836662002289785, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.146129608154297, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8950409889221191, + "num_tokens": 865869326.0, + "step": 22299 + }, + { + "epoch": 2.836789212568376, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.995135307312012, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.9004594683647156, + "num_tokens": 865906233.0, + "step": 22300 + }, + { + "epoch": 2.836916422846966, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.996272087097168, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8934980630874634, + "num_tokens": 865945061.0, + "step": 22301 + }, + { + "epoch": 2.8370436331255564, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.015947341918945, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.9010900259017944, + "num_tokens": 865984708.0, + "step": 22302 + }, + { + "epoch": 2.837170843404147, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.976496696472168, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8987194895744324, + "num_tokens": 866021191.0, + "step": 22303 + }, + { + "epoch": 2.8372980536827375, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.064726829528809, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.9051529169082642, + "num_tokens": 866057902.0, + "step": 22304 + }, + { + "epoch": 2.837425263961328, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.921277046203613, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8994258046150208, + "num_tokens": 866103352.0, + "step": 22305 + }, + { + "epoch": 2.8375524742399185, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.983400344848633, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9041801691055298, + "num_tokens": 866142301.0, + "step": 22306 + }, + { + "epoch": 2.837679684518509, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.030973434448242, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9039376378059387, + "num_tokens": 866182085.0, + "step": 22307 + }, + { + "epoch": 2.8378068947970996, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.094178199768066, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.898712158203125, + "num_tokens": 866222012.0, + "step": 22308 + }, + { + "epoch": 2.83793410507569, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.012030601501465, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.890603244304657, + "num_tokens": 866259987.0, + "step": 22309 + }, + { + "epoch": 2.8380613153542806, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.02819538116455, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8983651399612427, + "num_tokens": 866295573.0, + "step": 22310 + }, + { + "epoch": 2.838188525632871, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.941375732421875, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8965822458267212, + "num_tokens": 866338033.0, + "step": 22311 + }, + { + "epoch": 2.8383157359114617, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.029953956604004, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8880972862243652, + "num_tokens": 866378326.0, + "step": 22312 + }, + { + "epoch": 2.838442946190052, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.046611785888672, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8955975770950317, + "num_tokens": 866414583.0, + "step": 22313 + }, + { + "epoch": 2.8385701564686427, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.904873847961426, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8909246325492859, + "num_tokens": 866458155.0, + "step": 22314 + }, + { + "epoch": 2.8386973667472333, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.967962265014648, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9096822738647461, + "num_tokens": 866495137.0, + "step": 22315 + }, + { + "epoch": 2.838824577025824, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.15228271484375, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8997745513916016, + "num_tokens": 866532425.0, + "step": 22316 + }, + { + "epoch": 2.8389517873044143, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.120208740234375, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.9005352258682251, + "num_tokens": 866573659.0, + "step": 22317 + }, + { + "epoch": 2.839078997583005, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.903569221496582, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.9066104888916016, + "num_tokens": 866616938.0, + "step": 22318 + }, + { + "epoch": 2.8392062078615954, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 10.037090301513672, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8873789310455322, + "num_tokens": 866660805.0, + "step": 22319 + }, + { + "epoch": 2.8393334181401855, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.980027198791504, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8940597772598267, + "num_tokens": 866699073.0, + "step": 22320 + }, + { + "epoch": 2.8394606284187764, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.075145721435547, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8928883075714111, + "num_tokens": 866738590.0, + "step": 22321 + }, + { + "epoch": 2.8395878386973665, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.006685256958008, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.9015272259712219, + "num_tokens": 866777513.0, + "step": 22322 + }, + { + "epoch": 2.8397150489759575, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.025083541870117, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8951772451400757, + "num_tokens": 866816292.0, + "step": 22323 + }, + { + "epoch": 2.8398422592545476, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.980648040771484, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9044177532196045, + "num_tokens": 866856583.0, + "step": 22324 + }, + { + "epoch": 2.839969469533138, + "ewc_loss": 0.0218505859375, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2159347534179688e-05, + "grad_norm": 9.945191383361816, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8914333581924438, + "num_tokens": 866901343.0, + "step": 22325 + }, + { + "epoch": 2.8400966798117286, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.117959976196289, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8991292715072632, + "num_tokens": 866941145.0, + "step": 22326 + }, + { + "epoch": 2.840223890090319, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.958048820495605, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8908727169036865, + "num_tokens": 866987525.0, + "step": 22327 + }, + { + "epoch": 2.8403511003689097, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.960267066955566, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8983644843101501, + "num_tokens": 867026296.0, + "step": 22328 + }, + { + "epoch": 2.8404783106475002, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.029243469238281, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.903908371925354, + "num_tokens": 867059091.0, + "step": 22329 + }, + { + "epoch": 2.8406055209260908, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.058143615722656, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.894855260848999, + "num_tokens": 867098981.0, + "step": 22330 + }, + { + "epoch": 2.8407327312046813, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.968745231628418, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8979811668395996, + "num_tokens": 867138846.0, + "step": 22331 + }, + { + "epoch": 2.840859941483272, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.071703910827637, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.897560715675354, + "num_tokens": 867175151.0, + "step": 22332 + }, + { + "epoch": 2.8409871517618623, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.063831329345703, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8890342712402344, + "num_tokens": 867210436.0, + "step": 22333 + }, + { + "epoch": 2.841114362040453, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.056554794311523, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.9008705615997314, + "num_tokens": 867247962.0, + "step": 22334 + }, + { + "epoch": 2.8412415723190434, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.94286823272705, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8969389200210571, + "num_tokens": 867286931.0, + "step": 22335 + }, + { + "epoch": 2.841368782597634, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.972431182861328, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.9037869572639465, + "num_tokens": 867325135.0, + "step": 22336 + }, + { + "epoch": 2.8414959928762245, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.026208877563477, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.900269091129303, + "num_tokens": 867363315.0, + "step": 22337 + }, + { + "epoch": 2.841623203154815, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.051434516906738, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8951457738876343, + "num_tokens": 867402729.0, + "step": 22338 + }, + { + "epoch": 2.8417504134334055, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.087811470031738, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8996514081954956, + "num_tokens": 867435645.0, + "step": 22339 + }, + { + "epoch": 2.841877623711996, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.939180374145508, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9114018678665161, + "num_tokens": 867471284.0, + "step": 22340 + }, + { + "epoch": 2.8420048339905866, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.080055236816406, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8933074474334717, + "num_tokens": 867509428.0, + "step": 22341 + }, + { + "epoch": 2.842132044269177, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.942282676696777, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.9014332294464111, + "num_tokens": 867547544.0, + "step": 22342 + }, + { + "epoch": 2.8422592545477676, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.097698211669922, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8900666832923889, + "num_tokens": 867585934.0, + "step": 22343 + }, + { + "epoch": 2.842386464826358, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.991155624389648, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.9085423946380615, + "num_tokens": 867626742.0, + "step": 22344 + }, + { + "epoch": 2.8425136751049482, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.000381469726562, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.901685357093811, + "num_tokens": 867661772.0, + "step": 22345 + }, + { + "epoch": 2.842640885383539, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.098281860351562, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8903353214263916, + "num_tokens": 867698660.0, + "step": 22346 + }, + { + "epoch": 2.8427680956621293, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.97281265258789, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.9021604061126709, + "num_tokens": 867740479.0, + "step": 22347 + }, + { + "epoch": 2.8428953059407203, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.025609016418457, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8958173394203186, + "num_tokens": 867777694.0, + "step": 22348 + }, + { + "epoch": 2.8430225162193103, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.06646728515625, + "learning_rate": 1e-06, + "loss": 0.3788, + "mean_token_accuracy": 0.8722511529922485, + "num_tokens": 867821253.0, + "step": 22349 + }, + { + "epoch": 2.843149726497901, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.097291946411133, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.9076025485992432, + "num_tokens": 867854535.0, + "step": 22350 + }, + { + "epoch": 2.8432769367764914, + "ewc_loss": 0.02197265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.047647476196289, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.894113302230835, + "num_tokens": 867888235.0, + "step": 22351 + }, + { + "epoch": 2.843404147055082, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.86847972869873, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.9067232012748718, + "num_tokens": 867934365.0, + "step": 22352 + }, + { + "epoch": 2.8435313573336725, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.775161743164062e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.89418888092041, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9108271598815918, + "num_tokens": 867973315.0, + "step": 22353 + }, + { + "epoch": 2.843658567612263, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.775161743164062e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.043435096740723, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.9038056135177612, + "num_tokens": 868006852.0, + "step": 22354 + }, + { + "epoch": 2.8437857778908535, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.775161743164062e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.921404838562012, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8991087079048157, + "num_tokens": 868050619.0, + "step": 22355 + }, + { + "epoch": 2.843912988169444, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.775161743164062e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.071059226989746, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8957201242446899, + "num_tokens": 868084757.0, + "step": 22356 + }, + { + "epoch": 2.8440401984480346, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.775161743164062e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.044229507446289, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8945753574371338, + "num_tokens": 868119803.0, + "step": 22357 + }, + { + "epoch": 2.844167408726625, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.775161743164062e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.880939483642578, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8956995010375977, + "num_tokens": 868167725.0, + "step": 22358 + }, + { + "epoch": 2.8442946190052156, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.775161743164062e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.973052978515625, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8940484523773193, + "num_tokens": 868209767.0, + "step": 22359 + }, + { + "epoch": 2.844421829283806, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.988224983215332, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8952523469924927, + "num_tokens": 868252914.0, + "step": 22360 + }, + { + "epoch": 2.8445490395623967, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.72747802734375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.072997093200684, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.895560622215271, + "num_tokens": 868291372.0, + "step": 22361 + }, + { + "epoch": 2.844676249840987, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.775161743164062e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.046211242675781, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.900550365447998, + "num_tokens": 868330585.0, + "step": 22362 + }, + { + "epoch": 2.8448034601195777, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.775161743164062e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.97899341583252, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.900886058807373, + "num_tokens": 868370146.0, + "step": 22363 + }, + { + "epoch": 2.8449306703981683, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.775161743164062e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.027632713317871, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.9031001329421997, + "num_tokens": 868408477.0, + "step": 22364 + }, + { + "epoch": 2.845057880676759, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.775161743164062e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.001137733459473, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.9029579162597656, + "num_tokens": 868446792.0, + "step": 22365 + }, + { + "epoch": 2.8451850909553493, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.775161743164062e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.979708671569824, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.9030570983886719, + "num_tokens": 868485133.0, + "step": 22366 + }, + { + "epoch": 2.84531230123394, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.775161743164062e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.028975486755371, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8972344398498535, + "num_tokens": 868528229.0, + "step": 22367 + }, + { + "epoch": 2.84543951151253, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.822845458984375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.942912101745605, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.90108722448349, + "num_tokens": 868574390.0, + "step": 22368 + }, + { + "epoch": 2.845566721791121, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.822845458984375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.051664352416992, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8942680358886719, + "num_tokens": 868613673.0, + "step": 22369 + }, + { + "epoch": 2.845693932069711, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.822845458984375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.0917387008667, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8927253484725952, + "num_tokens": 868654935.0, + "step": 22370 + }, + { + "epoch": 2.845821142348302, + "ewc_loss": 0.0220947265625, + "ewc_loss_diag": 9.822845458984375e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.026281356811523, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8972958326339722, + "num_tokens": 868686874.0, + "step": 22371 + }, + { + "epoch": 2.845948352626892, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.822845458984375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.07299518585205, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8978153467178345, + "num_tokens": 868730627.0, + "step": 22372 + }, + { + "epoch": 2.846075562905483, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.822845458984375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.103378295898438, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8869296312332153, + "num_tokens": 868769367.0, + "step": 22373 + }, + { + "epoch": 2.846202773184073, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.822845458984375e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.050722122192383, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.9024224877357483, + "num_tokens": 868808644.0, + "step": 22374 + }, + { + "epoch": 2.8463299834626636, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.822845458984375e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.113131523132324, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8767837285995483, + "num_tokens": 868850684.0, + "step": 22375 + }, + { + "epoch": 2.846457193741254, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.01840877532959, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8854515552520752, + "num_tokens": 868890667.0, + "step": 22376 + }, + { + "epoch": 2.8465844040198447, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.04033088684082, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9129656553268433, + "num_tokens": 868924541.0, + "step": 22377 + }, + { + "epoch": 2.8467116142984352, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.150823593139648, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8901221752166748, + "num_tokens": 868958124.0, + "step": 22378 + }, + { + "epoch": 2.8468388245770258, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.907206535339355, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.9057263135910034, + "num_tokens": 868996217.0, + "step": 22379 + }, + { + "epoch": 2.8469660348556163, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.032731056213379, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8834570646286011, + "num_tokens": 869038648.0, + "step": 22380 + }, + { + "epoch": 2.847093245134207, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.06950855255127, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8896615505218506, + "num_tokens": 869079344.0, + "step": 22381 + }, + { + "epoch": 2.8472204554127973, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.912260055541992, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9039205312728882, + "num_tokens": 869122483.0, + "step": 22382 + }, + { + "epoch": 2.847347665691388, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.029989242553711, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9103487730026245, + "num_tokens": 869160371.0, + "step": 22383 + }, + { + "epoch": 2.8474748759699784, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.121682167053223, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8882201313972473, + "num_tokens": 869195052.0, + "step": 22384 + }, + { + "epoch": 2.847602086248569, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.006488800048828, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8943632245063782, + "num_tokens": 869236076.0, + "step": 22385 + }, + { + "epoch": 2.8477292965271594, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.954742431640625, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8992480039596558, + "num_tokens": 869275561.0, + "step": 22386 + }, + { + "epoch": 2.84785650680575, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.021438598632812, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9101147651672363, + "num_tokens": 869312484.0, + "step": 22387 + }, + { + "epoch": 2.8479837170843405, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.990239143371582, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.896357536315918, + "num_tokens": 869348217.0, + "step": 22388 + }, + { + "epoch": 2.848110927362931, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.070158004760742, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8994503617286682, + "num_tokens": 869383784.0, + "step": 22389 + }, + { + "epoch": 2.8482381376415216, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.938009262084961, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.9060466885566711, + "num_tokens": 869421055.0, + "step": 22390 + }, + { + "epoch": 2.848365347920112, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.118595123291016, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8914307951927185, + "num_tokens": 869454510.0, + "step": 22391 + }, + { + "epoch": 2.8484925581987026, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.0503511428833, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.9019287824630737, + "num_tokens": 869488923.0, + "step": 22392 + }, + { + "epoch": 2.8486197684772927, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.010993957519531, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8944481015205383, + "num_tokens": 869524706.0, + "step": 22393 + }, + { + "epoch": 2.8487469787558837, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.034385681152344, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8829296827316284, + "num_tokens": 869566826.0, + "step": 22394 + }, + { + "epoch": 2.8488741890344738, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.99160385131836, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.898361325263977, + "num_tokens": 869603535.0, + "step": 22395 + }, + { + "epoch": 2.8490013993130647, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.914798736572266, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.9021145105361938, + "num_tokens": 869646091.0, + "step": 22396 + }, + { + "epoch": 2.849128609591655, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.92209529876709, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8989100456237793, + "num_tokens": 869685104.0, + "step": 22397 + }, + { + "epoch": 2.849255819870246, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.945430755615234, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9064760804176331, + "num_tokens": 869725774.0, + "step": 22398 + }, + { + "epoch": 2.849383030148836, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.991329193115234, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9053070545196533, + "num_tokens": 869763978.0, + "step": 22399 + }, + { + "epoch": 2.8495102404274264, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.069921493530273, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8965867757797241, + "num_tokens": 869801538.0, + "step": 22400 + }, + { + "epoch": 2.849637450706017, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.958253860473633, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8981678485870361, + "num_tokens": 869843648.0, + "step": 22401 + }, + { + "epoch": 2.8497646609846075, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.00277328491211, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8961677551269531, + "num_tokens": 869883643.0, + "step": 22402 + }, + { + "epoch": 2.849891871263198, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.102437019348145, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.89006108045578, + "num_tokens": 869917162.0, + "step": 22403 + }, + { + "epoch": 2.8500190815417885, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.946334838867188, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8926876783370972, + "num_tokens": 869956151.0, + "step": 22404 + }, + { + "epoch": 2.850146291820379, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.083187103271484, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8875703811645508, + "num_tokens": 869990937.0, + "step": 22405 + }, + { + "epoch": 2.8502735020989696, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.984762191772461, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8905680179595947, + "num_tokens": 870034112.0, + "step": 22406 + }, + { + "epoch": 2.85040071237756, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.953328132629395, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.9012954235076904, + "num_tokens": 870070246.0, + "step": 22407 + }, + { + "epoch": 2.8505279226561506, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.023552894592285, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8949072360992432, + "num_tokens": 870108673.0, + "step": 22408 + }, + { + "epoch": 2.850655132934741, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.024324417114258, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.9042640924453735, + "num_tokens": 870149054.0, + "step": 22409 + }, + { + "epoch": 2.8507823432133317, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.063215255737305, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.898266077041626, + "num_tokens": 870183764.0, + "step": 22410 + }, + { + "epoch": 2.850909553491922, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.969559669494629, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8991860747337341, + "num_tokens": 870222593.0, + "step": 22411 + }, + { + "epoch": 2.8510367637705127, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.91721248626709, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.900587797164917, + "num_tokens": 870256060.0, + "step": 22412 + }, + { + "epoch": 2.8511639740491033, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.043700218200684, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8987159729003906, + "num_tokens": 870289718.0, + "step": 22413 + }, + { + "epoch": 2.851291184327694, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.013114929199219, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8985915780067444, + "num_tokens": 870328654.0, + "step": 22414 + }, + { + "epoch": 2.8514183946062843, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.057269096374512, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8894941210746765, + "num_tokens": 870364464.0, + "step": 22415 + }, + { + "epoch": 2.851545604884875, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.932217597961426, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9037419557571411, + "num_tokens": 870401162.0, + "step": 22416 + }, + { + "epoch": 2.8516728151634654, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.998680114746094, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8843249678611755, + "num_tokens": 870441656.0, + "step": 22417 + }, + { + "epoch": 2.8518000254420555, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.98367977142334, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8921574354171753, + "num_tokens": 870485227.0, + "step": 22418 + }, + { + "epoch": 2.8519272357206464, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.9124755859375, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.895378589630127, + "num_tokens": 870525016.0, + "step": 22419 + }, + { + "epoch": 2.8520544459992365, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.081988334655762, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8851180672645569, + "num_tokens": 870563775.0, + "step": 22420 + }, + { + "epoch": 2.8521816562778275, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.975621223449707, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8958473205566406, + "num_tokens": 870600866.0, + "step": 22421 + }, + { + "epoch": 2.8523088665564176, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.11202621459961, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8929237723350525, + "num_tokens": 870635164.0, + "step": 22422 + }, + { + "epoch": 2.852436076835008, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.924935340881348, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8890789747238159, + "num_tokens": 870676984.0, + "step": 22423 + }, + { + "epoch": 2.8525632871135986, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.004437446594238, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8984788060188293, + "num_tokens": 870717052.0, + "step": 22424 + }, + { + "epoch": 2.852690497392189, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.138675689697266, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.893574059009552, + "num_tokens": 870750623.0, + "step": 22425 + }, + { + "epoch": 2.8528177076707797, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.072919845581055, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8857977390289307, + "num_tokens": 870792349.0, + "step": 22426 + }, + { + "epoch": 2.85294491794937, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.988668441772461, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.9042695760726929, + "num_tokens": 870828947.0, + "step": 22427 + }, + { + "epoch": 2.8530721282279607, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.062555313110352, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8965593576431274, + "num_tokens": 870865409.0, + "step": 22428 + }, + { + "epoch": 2.8531993385065513, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.010601997375488, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9089534282684326, + "num_tokens": 870898757.0, + "step": 22429 + }, + { + "epoch": 2.853326548785142, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.065374374389648, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.9054299592971802, + "num_tokens": 870937284.0, + "step": 22430 + }, + { + "epoch": 2.8534537590637323, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.027207374572754, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8948488235473633, + "num_tokens": 870978105.0, + "step": 22431 + }, + { + "epoch": 2.853580969342323, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.081795692443848, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8917503356933594, + "num_tokens": 871016330.0, + "step": 22432 + }, + { + "epoch": 2.8537081796209134, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.963736534118652, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8976852893829346, + "num_tokens": 871055050.0, + "step": 22433 + }, + { + "epoch": 2.853835389899504, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.005879402160645, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.9019342660903931, + "num_tokens": 871097403.0, + "step": 22434 + }, + { + "epoch": 2.8539626001780944, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.084256172180176, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8944969773292542, + "num_tokens": 871133976.0, + "step": 22435 + }, + { + "epoch": 2.854089810456685, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.034439086914062, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.891278088092804, + "num_tokens": 871175270.0, + "step": 22436 + }, + { + "epoch": 2.8542170207352755, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.982826232910156, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9049397110939026, + "num_tokens": 871212384.0, + "step": 22437 + }, + { + "epoch": 2.854344231013866, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.10607624053955, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.902387261390686, + "num_tokens": 871244510.0, + "step": 22438 + }, + { + "epoch": 2.8544714412924566, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.065771102905273, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8863441944122314, + "num_tokens": 871284035.0, + "step": 22439 + }, + { + "epoch": 2.854598651571047, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.943879127502441, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.906222939491272, + "num_tokens": 871323218.0, + "step": 22440 + }, + { + "epoch": 2.8547258618496376, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.070887565612793, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8857561349868774, + "num_tokens": 871364811.0, + "step": 22441 + }, + { + "epoch": 2.854853072128228, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.126859664916992, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8970564603805542, + "num_tokens": 871401524.0, + "step": 22442 + }, + { + "epoch": 2.8549802824068182, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.918767929077148, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.9006062150001526, + "num_tokens": 871445230.0, + "step": 22443 + }, + { + "epoch": 2.855107492685409, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.092521667480469, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9110948443412781, + "num_tokens": 871479115.0, + "step": 22444 + }, + { + "epoch": 2.8552347029639993, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.01596736907959, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8895464539527893, + "num_tokens": 871520559.0, + "step": 22445 + }, + { + "epoch": 2.8553619132425903, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.174600601196289, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.9026449918746948, + "num_tokens": 871553310.0, + "step": 22446 + }, + { + "epoch": 2.8554891235211803, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.079185485839844, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8909103870391846, + "num_tokens": 871591971.0, + "step": 22447 + }, + { + "epoch": 2.855616333799771, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.98962688446045, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8963852524757385, + "num_tokens": 871637110.0, + "step": 22448 + }, + { + "epoch": 2.8557435440783614, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.017182350158691, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8985630869865417, + "num_tokens": 871676698.0, + "step": 22449 + }, + { + "epoch": 2.855870754356952, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.018221855163574, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.9010980129241943, + "num_tokens": 871713611.0, + "step": 22450 + }, + { + "epoch": 2.8559979646355425, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.948742866516113, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9018423557281494, + "num_tokens": 871754083.0, + "step": 22451 + }, + { + "epoch": 2.856125174914133, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.024578094482422, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8894846439361572, + "num_tokens": 871795934.0, + "step": 22452 + }, + { + "epoch": 2.8562523851927235, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.028976440429688, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.9032084941864014, + "num_tokens": 871832977.0, + "step": 22453 + }, + { + "epoch": 2.856379595471314, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.019929885864258, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8949779272079468, + "num_tokens": 871872645.0, + "step": 22454 + }, + { + "epoch": 2.8565068057499046, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.990403175354004, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8962212800979614, + "num_tokens": 871907411.0, + "step": 22455 + }, + { + "epoch": 2.856634016028495, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.01541519165039, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8971762657165527, + "num_tokens": 871944335.0, + "step": 22456 + }, + { + "epoch": 2.8567612263070856, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.021427154541016, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8898757696151733, + "num_tokens": 871984295.0, + "step": 22457 + }, + { + "epoch": 2.856888436585676, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.091231346130371, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8771103024482727, + "num_tokens": 872021686.0, + "step": 22458 + }, + { + "epoch": 2.8570156468642667, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.074212074279785, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.888209342956543, + "num_tokens": 872056866.0, + "step": 22459 + }, + { + "epoch": 2.857142857142857, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.050714492797852, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8981167078018188, + "num_tokens": 872091797.0, + "step": 22460 + }, + { + "epoch": 2.8572700674214477, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.019179344177246, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8895508050918579, + "num_tokens": 872132776.0, + "step": 22461 + }, + { + "epoch": 2.8573972777000383, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.985333442687988, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.9010748267173767, + "num_tokens": 872174504.0, + "step": 22462 + }, + { + "epoch": 2.857524487978629, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.017227172851562, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8972338438034058, + "num_tokens": 872213841.0, + "step": 22463 + }, + { + "epoch": 2.8576516982572193, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.1849946975708, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.891877293586731, + "num_tokens": 872248792.0, + "step": 22464 + }, + { + "epoch": 2.85777890853581, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.992522239685059, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8813539743423462, + "num_tokens": 872289017.0, + "step": 22465 + }, + { + "epoch": 2.8579061188144, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.070577621459961, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8830405473709106, + "num_tokens": 872336737.0, + "step": 22466 + }, + { + "epoch": 2.858033329092991, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.985821723937988, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8920868039131165, + "num_tokens": 872379985.0, + "step": 22467 + }, + { + "epoch": 2.858160539371581, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.9777193069458, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8959473371505737, + "num_tokens": 872419387.0, + "step": 22468 + }, + { + "epoch": 2.858287749650172, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.027456283569336, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.9039257168769836, + "num_tokens": 872461756.0, + "step": 22469 + }, + { + "epoch": 2.858414959928762, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.029022216796875, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9019408226013184, + "num_tokens": 872496754.0, + "step": 22470 + }, + { + "epoch": 2.858542170207353, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.173523902893066, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8960376977920532, + "num_tokens": 872530630.0, + "step": 22471 + }, + { + "epoch": 2.858669380485943, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.066868782043457, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8991908431053162, + "num_tokens": 872567909.0, + "step": 22472 + }, + { + "epoch": 2.8587965907645336, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.980826377868652, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8969868421554565, + "num_tokens": 872610128.0, + "step": 22473 + }, + { + "epoch": 2.858923801043124, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.98800277709961, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9149587750434875, + "num_tokens": 872650326.0, + "step": 22474 + }, + { + "epoch": 2.8590510113217147, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.998756408691406, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.899077296257019, + "num_tokens": 872690479.0, + "step": 22475 + }, + { + "epoch": 2.859178221600305, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.109286308288574, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.9018164277076721, + "num_tokens": 872726423.0, + "step": 22476 + }, + { + "epoch": 2.8593054318788957, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.949535369873047, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9134327173233032, + "num_tokens": 872763063.0, + "step": 22477 + }, + { + "epoch": 2.8594326421574863, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.065582275390625, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9075893759727478, + "num_tokens": 872796675.0, + "step": 22478 + }, + { + "epoch": 2.859559852436077, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.056774139404297, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.886894941329956, + "num_tokens": 872835303.0, + "step": 22479 + }, + { + "epoch": 2.8596870627146673, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.018054008483887, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8955923318862915, + "num_tokens": 872875881.0, + "step": 22480 + }, + { + "epoch": 2.859814272993258, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.001485824584961, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9039005637168884, + "num_tokens": 872915175.0, + "step": 22481 + }, + { + "epoch": 2.8599414832718484, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.012504577636719, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9070624113082886, + "num_tokens": 872952539.0, + "step": 22482 + }, + { + "epoch": 2.860068693550439, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.091888427734375, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8926824927330017, + "num_tokens": 872995733.0, + "step": 22483 + }, + { + "epoch": 2.8601959038290294, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.043724060058594, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.9063203930854797, + "num_tokens": 873029928.0, + "step": 22484 + }, + { + "epoch": 2.86032311410762, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.208038330078125, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8858230710029602, + "num_tokens": 873061020.0, + "step": 22485 + }, + { + "epoch": 2.8604503243862105, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.047426223754883, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.9009490609169006, + "num_tokens": 873099646.0, + "step": 22486 + }, + { + "epoch": 2.860577534664801, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.093716621398926, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.890507698059082, + "num_tokens": 873145566.0, + "step": 22487 + }, + { + "epoch": 2.8607047449433916, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.925700187683105, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9141048789024353, + "num_tokens": 873186763.0, + "step": 22488 + }, + { + "epoch": 2.860831955221982, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.018341064453125, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.905301570892334, + "num_tokens": 873221516.0, + "step": 22489 + }, + { + "epoch": 2.8609591655005726, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.967083930969238, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9025843143463135, + "num_tokens": 873262181.0, + "step": 22490 + }, + { + "epoch": 2.8610863757791627, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.008880615234375, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9140180349349976, + "num_tokens": 873298221.0, + "step": 22491 + }, + { + "epoch": 2.8612135860577537, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.938542366027832, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.893487811088562, + "num_tokens": 873338894.0, + "step": 22492 + }, + { + "epoch": 2.8613407963363438, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.022130966186523, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.905022144317627, + "num_tokens": 873380743.0, + "step": 22493 + }, + { + "epoch": 2.8614680066149347, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.115832328796387, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.9042685031890869, + "num_tokens": 873414312.0, + "step": 22494 + }, + { + "epoch": 2.861595216893525, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.965758323669434, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8980357646942139, + "num_tokens": 873455682.0, + "step": 22495 + }, + { + "epoch": 2.861722427172116, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.047124862670898, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.902265191078186, + "num_tokens": 873495104.0, + "step": 22496 + }, + { + "epoch": 2.861849637450706, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.082058906555176, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8969199657440186, + "num_tokens": 873534238.0, + "step": 22497 + }, + { + "epoch": 2.8619768477292964, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.003015518188477, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8970219492912292, + "num_tokens": 873578057.0, + "step": 22498 + }, + { + "epoch": 2.862104058007887, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.991543769836426, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9089490175247192, + "num_tokens": 873618305.0, + "step": 22499 + }, + { + "epoch": 2.8622312682864774, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.1202974319458, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.9007298350334167, + "num_tokens": 873661985.0, + "step": 22500 + }, + { + "epoch": 2.862358478565068, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.004672050476074, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.9030382037162781, + "num_tokens": 873702421.0, + "step": 22501 + }, + { + "epoch": 2.8624856888436585, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.999756813049316, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8865301609039307, + "num_tokens": 873742756.0, + "step": 22502 + }, + { + "epoch": 2.862612899122249, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.062076568603516, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9131984710693359, + "num_tokens": 873778330.0, + "step": 22503 + }, + { + "epoch": 2.8627401094008396, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.97579574584961, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9038324952125549, + "num_tokens": 873816800.0, + "step": 22504 + }, + { + "epoch": 2.86286731967943, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.216367721557617, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9101642370223999, + "num_tokens": 873845790.0, + "step": 22505 + }, + { + "epoch": 2.8629945299580206, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.995830535888672, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8828904628753662, + "num_tokens": 873887851.0, + "step": 22506 + }, + { + "epoch": 2.863121740236611, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.998976707458496, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8944876194000244, + "num_tokens": 873933607.0, + "step": 22507 + }, + { + "epoch": 2.8632489505152017, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.100817680358887, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.904797375202179, + "num_tokens": 873969156.0, + "step": 22508 + }, + { + "epoch": 2.863376160793792, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.965188980102539, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.9000021815299988, + "num_tokens": 874012681.0, + "step": 22509 + }, + { + "epoch": 2.8635033710723827, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.087998390197754, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8935664892196655, + "num_tokens": 874049498.0, + "step": 22510 + }, + { + "epoch": 2.8636305813509733, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.094218254089355, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.9018713235855103, + "num_tokens": 874082668.0, + "step": 22511 + }, + { + "epoch": 2.863757791629564, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.04108715057373, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8967043161392212, + "num_tokens": 874123768.0, + "step": 22512 + }, + { + "epoch": 2.8638850019081543, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.075170516967773, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8942446708679199, + "num_tokens": 874160191.0, + "step": 22513 + }, + { + "epoch": 2.864012212186745, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.036847114562988, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9056086540222168, + "num_tokens": 874200305.0, + "step": 22514 + }, + { + "epoch": 2.8641394224653354, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 9.92314338684082, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.905659556388855, + "num_tokens": 874241913.0, + "step": 22515 + }, + { + "epoch": 2.8642666327439255, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.041683197021484, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8988798260688782, + "num_tokens": 874278564.0, + "step": 22516 + }, + { + "epoch": 2.8643938430225164, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.004170417785645, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8959636688232422, + "num_tokens": 874320339.0, + "step": 22517 + }, + { + "epoch": 2.8645210533011065, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.050978660583496, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.9029528498649597, + "num_tokens": 874357122.0, + "step": 22518 + }, + { + "epoch": 2.8646482635796975, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.13568115234375, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8881514072418213, + "num_tokens": 874392112.0, + "step": 22519 + }, + { + "epoch": 2.8647754738582876, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.991048812866211, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8906798958778381, + "num_tokens": 874432614.0, + "step": 22520 + }, + { + "epoch": 2.864902684136878, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.071589469909668, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8955916166305542, + "num_tokens": 874468038.0, + "step": 22521 + }, + { + "epoch": 2.8650298944154686, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.04183292388916, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.9065622091293335, + "num_tokens": 874506226.0, + "step": 22522 + }, + { + "epoch": 2.865157104694059, + "ewc_loss": 0.022216796875, + "ewc_loss_diag": 9.918212890625e-05, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.094425201416016, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8971011638641357, + "num_tokens": 874541851.0, + "step": 22523 + }, + { + "epoch": 2.8652843149726497, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010013580322265625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.979228019714355, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8960624933242798, + "num_tokens": 874588709.0, + "step": 22524 + }, + { + "epoch": 2.86541152525124, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010013580322265625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.110154151916504, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.900388777256012, + "num_tokens": 874621815.0, + "step": 22525 + }, + { + "epoch": 2.8655387355298307, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010013580322265625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.039918899536133, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9053128957748413, + "num_tokens": 874656953.0, + "step": 22526 + }, + { + "epoch": 2.8656659458084213, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010013580322265625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.00275993347168, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.906543493270874, + "num_tokens": 874692879.0, + "step": 22527 + }, + { + "epoch": 2.865793156087012, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010013580322265625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.154669761657715, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8988359570503235, + "num_tokens": 874727526.0, + "step": 22528 + }, + { + "epoch": 2.8659203663656023, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010013580322265625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.085359573364258, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8889107704162598, + "num_tokens": 874763402.0, + "step": 22529 + }, + { + "epoch": 2.866047576644193, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010013580322265625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.96239185333252, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.910049557685852, + "num_tokens": 874807021.0, + "step": 22530 + }, + { + "epoch": 2.8661747869227834, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010013580322265625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.09131145477295, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8892185688018799, + "num_tokens": 874843811.0, + "step": 22531 + }, + { + "epoch": 2.866301997201374, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010013580322265625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.102319717407227, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8991006016731262, + "num_tokens": 874876227.0, + "step": 22532 + }, + { + "epoch": 2.8664292074799644, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010013580322265625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.064671516418457, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9080528020858765, + "num_tokens": 874908276.0, + "step": 22533 + }, + { + "epoch": 2.866556417758555, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010013580322265625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.00477409362793, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8950339555740356, + "num_tokens": 874953033.0, + "step": 22534 + }, + { + "epoch": 2.8666836280371455, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.98033618927002, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8971473574638367, + "num_tokens": 874996289.0, + "step": 22535 + }, + { + "epoch": 2.866810838315736, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010013580322265625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.05142879486084, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8894741535186768, + "num_tokens": 875038522.0, + "step": 22536 + }, + { + "epoch": 2.8669380485943265, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.899348258972168, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.910894513130188, + "num_tokens": 875082113.0, + "step": 22537 + }, + { + "epoch": 2.867065258872917, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.159242630004883, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8989285230636597, + "num_tokens": 875113564.0, + "step": 22538 + }, + { + "epoch": 2.8671924691515076, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.133135795593262, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8956900835037231, + "num_tokens": 875146181.0, + "step": 22539 + }, + { + "epoch": 2.867319679430098, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.922738075256348, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9033921360969543, + "num_tokens": 875186879.0, + "step": 22540 + }, + { + "epoch": 2.867446889708688, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.055041313171387, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8846527338027954, + "num_tokens": 875221390.0, + "step": 22541 + }, + { + "epoch": 2.867574099987279, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.11055850982666, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8962212204933167, + "num_tokens": 875254927.0, + "step": 22542 + }, + { + "epoch": 2.8677013102658693, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.963350296020508, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.9087363481521606, + "num_tokens": 875294183.0, + "step": 22543 + }, + { + "epoch": 2.8678285205444602, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.160037994384766, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8893958330154419, + "num_tokens": 875331304.0, + "step": 22544 + }, + { + "epoch": 2.8679557308230503, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.000051498413086, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8991621136665344, + "num_tokens": 875367116.0, + "step": 22545 + }, + { + "epoch": 2.868082941101641, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.875199317932129, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9114949703216553, + "num_tokens": 875408921.0, + "step": 22546 + }, + { + "epoch": 2.8682101513802314, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.977450370788574, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.9007723331451416, + "num_tokens": 875450714.0, + "step": 22547 + }, + { + "epoch": 2.868337361658822, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.077690124511719, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8862811923027039, + "num_tokens": 875491221.0, + "step": 22548 + }, + { + "epoch": 2.8684645719374124, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.137155532836914, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.892250657081604, + "num_tokens": 875524827.0, + "step": 22549 + }, + { + "epoch": 2.868591782216003, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.043886184692383, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.895864725112915, + "num_tokens": 875565533.0, + "step": 22550 + }, + { + "epoch": 2.8687189924945935, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.936951637268066, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9047226309776306, + "num_tokens": 875606484.0, + "step": 22551 + }, + { + "epoch": 2.868846202773184, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.027134895324707, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8897961378097534, + "num_tokens": 875648785.0, + "step": 22552 + }, + { + "epoch": 2.8689734130517746, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.179451942443848, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8908963203430176, + "num_tokens": 875676839.0, + "step": 22553 + }, + { + "epoch": 2.869100623330365, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.916375160217285, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.908548891544342, + "num_tokens": 875721146.0, + "step": 22554 + }, + { + "epoch": 2.8692278336089556, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.056380271911621, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9034520387649536, + "num_tokens": 875753652.0, + "step": 22555 + }, + { + "epoch": 2.869355043887546, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.015857696533203, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8936271667480469, + "num_tokens": 875794926.0, + "step": 22556 + }, + { + "epoch": 2.8694822541661367, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.94400691986084, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8837408423423767, + "num_tokens": 875843336.0, + "step": 22557 + }, + { + "epoch": 2.869609464444727, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.012042999267578, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.899232804775238, + "num_tokens": 875880368.0, + "step": 22558 + }, + { + "epoch": 2.8697366747233177, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.98075008392334, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.9019012451171875, + "num_tokens": 875923812.0, + "step": 22559 + }, + { + "epoch": 2.8698638850019083, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.101693153381348, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8904637098312378, + "num_tokens": 875959242.0, + "step": 22560 + }, + { + "epoch": 2.869991095280499, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.058125495910645, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8889896273612976, + "num_tokens": 875994575.0, + "step": 22561 + }, + { + "epoch": 2.8701183055590893, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.004654884338379, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9064452648162842, + "num_tokens": 876034961.0, + "step": 22562 + }, + { + "epoch": 2.87024551583768, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.067224502563477, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8941159844398499, + "num_tokens": 876075213.0, + "step": 22563 + }, + { + "epoch": 2.87037272611627, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.99169921875, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.9017879962921143, + "num_tokens": 876114383.0, + "step": 22564 + }, + { + "epoch": 2.870499936394861, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.013129234313965, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.893839955329895, + "num_tokens": 876153875.0, + "step": 22565 + }, + { + "epoch": 2.870627146673451, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.959043502807617, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8980467319488525, + "num_tokens": 876197219.0, + "step": 22566 + }, + { + "epoch": 2.870754356952042, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.078194618225098, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.9016852974891663, + "num_tokens": 876235201.0, + "step": 22567 + }, + { + "epoch": 2.870881567230632, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.969301223754883, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8806548118591309, + "num_tokens": 876277014.0, + "step": 22568 + }, + { + "epoch": 2.871008777509223, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.117989540100098, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9047756791114807, + "num_tokens": 876305336.0, + "step": 22569 + }, + { + "epoch": 2.871135987787813, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.988252639770508, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.90755295753479, + "num_tokens": 876339108.0, + "step": 22570 + }, + { + "epoch": 2.8712631980664036, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.010520935058594, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8926339149475098, + "num_tokens": 876377247.0, + "step": 22571 + }, + { + "epoch": 2.871390408344994, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.009425163269043, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8965572118759155, + "num_tokens": 876418248.0, + "step": 22572 + }, + { + "epoch": 2.8715176186235847, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.986052513122559, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8980804681777954, + "num_tokens": 876458931.0, + "step": 22573 + }, + { + "epoch": 2.871644828902175, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.975184440612793, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.9017782211303711, + "num_tokens": 876500968.0, + "step": 22574 + }, + { + "epoch": 2.8717720391807657, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.009575843811035, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9025877714157104, + "num_tokens": 876541421.0, + "step": 22575 + }, + { + "epoch": 2.8718992494593563, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.016435623168945, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.898544192314148, + "num_tokens": 876577467.0, + "step": 22576 + }, + { + "epoch": 2.872026459737947, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.976570129394531, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9110879898071289, + "num_tokens": 876613007.0, + "step": 22577 + }, + { + "epoch": 2.8721536700165373, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.13598346710205, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8891416788101196, + "num_tokens": 876653260.0, + "step": 22578 + }, + { + "epoch": 2.872280880295128, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.025200843811035, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.886238157749176, + "num_tokens": 876691123.0, + "step": 22579 + }, + { + "epoch": 2.8724080905737184, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.069683074951172, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.9025019407272339, + "num_tokens": 876730816.0, + "step": 22580 + }, + { + "epoch": 2.872535300852309, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.082186698913574, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8958945870399475, + "num_tokens": 876772987.0, + "step": 22581 + }, + { + "epoch": 2.8726625111308994, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.061285972595215, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8996599912643433, + "num_tokens": 876814014.0, + "step": 22582 + }, + { + "epoch": 2.87278972140949, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.004783630371094, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.90083909034729, + "num_tokens": 876853842.0, + "step": 22583 + }, + { + "epoch": 2.8729169316880805, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.03092098236084, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.9135982990264893, + "num_tokens": 876890525.0, + "step": 22584 + }, + { + "epoch": 2.873044141966671, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.978902816772461, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8944566249847412, + "num_tokens": 876934990.0, + "step": 22585 + }, + { + "epoch": 2.8731713522452615, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.084799766540527, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.9008193016052246, + "num_tokens": 876971682.0, + "step": 22586 + }, + { + "epoch": 2.873298562523852, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.096273422241211, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8840864896774292, + "num_tokens": 877006459.0, + "step": 22587 + }, + { + "epoch": 2.8734257728024426, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.976536750793457, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8938658833503723, + "num_tokens": 877047320.0, + "step": 22588 + }, + { + "epoch": 2.8735529830810327, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.996307373046875, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8891251683235168, + "num_tokens": 877084905.0, + "step": 22589 + }, + { + "epoch": 2.8736801933596237, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.087096214294434, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8923314213752747, + "num_tokens": 877122073.0, + "step": 22590 + }, + { + "epoch": 2.8738074036382137, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.061403274536133, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8897027969360352, + "num_tokens": 877161946.0, + "step": 22591 + }, + { + "epoch": 2.8739346139168047, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.898381233215332, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.900394856929779, + "num_tokens": 877206409.0, + "step": 22592 + }, + { + "epoch": 2.874061824195395, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.951360702514648, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.9013229608535767, + "num_tokens": 877250597.0, + "step": 22593 + }, + { + "epoch": 2.8741890344739858, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.982739448547363, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9041933417320251, + "num_tokens": 877286044.0, + "step": 22594 + }, + { + "epoch": 2.874316244752576, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.153807640075684, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8926253318786621, + "num_tokens": 877321367.0, + "step": 22595 + }, + { + "epoch": 2.8744434550311664, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.983129501342773, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.9043214917182922, + "num_tokens": 877358903.0, + "step": 22596 + }, + { + "epoch": 2.874570665309757, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.017171859741211, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8865535259246826, + "num_tokens": 877396399.0, + "step": 22597 + }, + { + "epoch": 2.8746978755883474, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.944585800170898, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9093190431594849, + "num_tokens": 877434447.0, + "step": 22598 + }, + { + "epoch": 2.874825085866938, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.024538040161133, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8955024480819702, + "num_tokens": 877476784.0, + "step": 22599 + }, + { + "epoch": 2.8749522961455285, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.97491455078125, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8946922421455383, + "num_tokens": 877519350.0, + "step": 22600 + }, + { + "epoch": 2.875079506424119, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.030983924865723, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8933895826339722, + "num_tokens": 877560975.0, + "step": 22601 + }, + { + "epoch": 2.8752067167027096, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.057506561279297, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8955744504928589, + "num_tokens": 877595235.0, + "step": 22602 + }, + { + "epoch": 2.8753339269813, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.96069622039795, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8982800841331482, + "num_tokens": 877637944.0, + "step": 22603 + }, + { + "epoch": 2.8754611372598906, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.061064720153809, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.9067031741142273, + "num_tokens": 877673264.0, + "step": 22604 + }, + { + "epoch": 2.875588347538481, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.17496395111084, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8934502601623535, + "num_tokens": 877716781.0, + "step": 22605 + }, + { + "epoch": 2.8757155578170717, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.030098915100098, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9048987627029419, + "num_tokens": 877754191.0, + "step": 22606 + }, + { + "epoch": 2.875842768095662, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.104464530944824, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8951981067657471, + "num_tokens": 877788337.0, + "step": 22607 + }, + { + "epoch": 2.8759699783742527, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.035408020019531, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.893203854560852, + "num_tokens": 877831582.0, + "step": 22608 + }, + { + "epoch": 2.8760971886528433, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.954033851623535, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.9011952877044678, + "num_tokens": 877869380.0, + "step": 22609 + }, + { + "epoch": 2.876224398931434, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.998503684997559, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9062693119049072, + "num_tokens": 877910380.0, + "step": 22610 + }, + { + "epoch": 2.8763516092100243, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.896771430969238, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.909437894821167, + "num_tokens": 877948415.0, + "step": 22611 + }, + { + "epoch": 2.876478819488615, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.077868461608887, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8919841647148132, + "num_tokens": 877987965.0, + "step": 22612 + }, + { + "epoch": 2.8766060297672054, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.024852752685547, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8982546329498291, + "num_tokens": 878023597.0, + "step": 22613 + }, + { + "epoch": 2.8767332400457954, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.034123420715332, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.9034372568130493, + "num_tokens": 878057971.0, + "step": 22614 + }, + { + "epoch": 2.8768604503243864, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.08696174621582, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9045263528823853, + "num_tokens": 878095569.0, + "step": 22615 + }, + { + "epoch": 2.8769876606029765, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.954105377197266, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9042985439300537, + "num_tokens": 878135125.0, + "step": 22616 + }, + { + "epoch": 2.8771148708815675, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.016667366027832, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.9061427116394043, + "num_tokens": 878173112.0, + "step": 22617 + }, + { + "epoch": 2.8772420811601576, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.008652687072754, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8978848457336426, + "num_tokens": 878214415.0, + "step": 22618 + }, + { + "epoch": 2.877369291438748, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.983577728271484, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.900539219379425, + "num_tokens": 878253069.0, + "step": 22619 + }, + { + "epoch": 2.8774965017173386, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.076705932617188, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8955467939376831, + "num_tokens": 878292695.0, + "step": 22620 + }, + { + "epoch": 2.877623711995929, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.018180847167969, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9041758179664612, + "num_tokens": 878330961.0, + "step": 22621 + }, + { + "epoch": 2.8777509222745197, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.979121208190918, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.9084490537643433, + "num_tokens": 878372503.0, + "step": 22622 + }, + { + "epoch": 2.87787813255311, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.016409873962402, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8951932787895203, + "num_tokens": 878409288.0, + "step": 22623 + }, + { + "epoch": 2.8780053428317007, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.04379940032959, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8956084251403809, + "num_tokens": 878444668.0, + "step": 22624 + }, + { + "epoch": 2.8781325531102913, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.078750610351562, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8935800790786743, + "num_tokens": 878479957.0, + "step": 22625 + }, + { + "epoch": 2.878259763388882, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.920162200927734, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9071998596191406, + "num_tokens": 878520354.0, + "step": 22626 + }, + { + "epoch": 2.8783869736674723, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.1026029586792, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8932496905326843, + "num_tokens": 878557148.0, + "step": 22627 + }, + { + "epoch": 2.878514183946063, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.015127182006836, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8972607851028442, + "num_tokens": 878595978.0, + "step": 22628 + }, + { + "epoch": 2.8786413942246534, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.99012565612793, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.9042016267776489, + "num_tokens": 878630856.0, + "step": 22629 + }, + { + "epoch": 2.878768604503244, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.944684028625488, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9088023900985718, + "num_tokens": 878668997.0, + "step": 22630 + }, + { + "epoch": 2.8788958147818344, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.973413467407227, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8977832794189453, + "num_tokens": 878713850.0, + "step": 22631 + }, + { + "epoch": 2.879023025060425, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.001832962036133, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8808094263076782, + "num_tokens": 878754495.0, + "step": 22632 + }, + { + "epoch": 2.8791502353390155, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.182541847229004, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8831800818443298, + "num_tokens": 878788925.0, + "step": 22633 + }, + { + "epoch": 2.879277445617606, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.029067039489746, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.902429461479187, + "num_tokens": 878821051.0, + "step": 22634 + }, + { + "epoch": 2.8794046558961965, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.033650398254395, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8931889533996582, + "num_tokens": 878864162.0, + "step": 22635 + }, + { + "epoch": 2.879531866174787, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.922306060791016, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9028729200363159, + "num_tokens": 878903237.0, + "step": 22636 + }, + { + "epoch": 2.8796590764533776, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.107696533203125, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8895721435546875, + "num_tokens": 878939002.0, + "step": 22637 + }, + { + "epoch": 2.879786286731968, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.926125526428223, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9104669094085693, + "num_tokens": 878980825.0, + "step": 22638 + }, + { + "epoch": 2.879913497010558, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.03407096862793, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.898158073425293, + "num_tokens": 879020125.0, + "step": 22639 + }, + { + "epoch": 2.880040707289149, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.970693588256836, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9054369330406189, + "num_tokens": 879057650.0, + "step": 22640 + }, + { + "epoch": 2.8801679175677393, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.982917785644531, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.9014875292778015, + "num_tokens": 879095662.0, + "step": 22641 + }, + { + "epoch": 2.8802951278463302, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.984061241149902, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.9009147882461548, + "num_tokens": 879139518.0, + "step": 22642 + }, + { + "epoch": 2.8804223381249203, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.988615989685059, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.9016469717025757, + "num_tokens": 879177733.0, + "step": 22643 + }, + { + "epoch": 2.880549548403511, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.040045738220215, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8874585628509521, + "num_tokens": 879214014.0, + "step": 22644 + }, + { + "epoch": 2.8806767586821014, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.015323638916016, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8920694589614868, + "num_tokens": 879253080.0, + "step": 22645 + }, + { + "epoch": 2.880803968960692, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.930493354797363, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.9025570154190063, + "num_tokens": 879294143.0, + "step": 22646 + }, + { + "epoch": 2.8809311792392824, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.030537605285645, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9076921343803406, + "num_tokens": 879331120.0, + "step": 22647 + }, + { + "epoch": 2.881058389517873, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.060425758361816, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.9015370607376099, + "num_tokens": 879364525.0, + "step": 22648 + }, + { + "epoch": 2.8811855997964635, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.902742385864258, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8997824192047119, + "num_tokens": 879409350.0, + "step": 22649 + }, + { + "epoch": 2.881312810075054, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.096406936645508, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.903355598449707, + "num_tokens": 879441682.0, + "step": 22650 + }, + { + "epoch": 2.8814400203536445, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.050421714782715, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8857927322387695, + "num_tokens": 879483750.0, + "step": 22651 + }, + { + "epoch": 2.881567230632235, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.97551441192627, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8899703025817871, + "num_tokens": 879519520.0, + "step": 22652 + }, + { + "epoch": 2.8816944409108256, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.981648445129395, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.9035879373550415, + "num_tokens": 879557244.0, + "step": 22653 + }, + { + "epoch": 2.881821651189416, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.001646041870117, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8855428695678711, + "num_tokens": 879594804.0, + "step": 22654 + }, + { + "epoch": 2.8819488614680067, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.979578971862793, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8960468173027039, + "num_tokens": 879632900.0, + "step": 22655 + }, + { + "epoch": 2.882076071746597, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.092965126037598, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8929046392440796, + "num_tokens": 879669580.0, + "step": 22656 + }, + { + "epoch": 2.8822032820251877, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.913777351379395, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8998302221298218, + "num_tokens": 879708956.0, + "step": 22657 + }, + { + "epoch": 2.8823304923037782, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.006895065307617, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8950934410095215, + "num_tokens": 879749869.0, + "step": 22658 + }, + { + "epoch": 2.8824577025823688, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.922591209411621, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8944246768951416, + "num_tokens": 879794998.0, + "step": 22659 + }, + { + "epoch": 2.8825849128609593, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.08059310913086, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8871840238571167, + "num_tokens": 879829177.0, + "step": 22660 + }, + { + "epoch": 2.88271212313955, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.940875053405762, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9055778980255127, + "num_tokens": 879868824.0, + "step": 22661 + }, + { + "epoch": 2.88283933341814, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.083477973937988, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8964279890060425, + "num_tokens": 879901950.0, + "step": 22662 + }, + { + "epoch": 2.882966543696731, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.150420188903809, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.905221700668335, + "num_tokens": 879938296.0, + "step": 22663 + }, + { + "epoch": 2.883093753975321, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 9.960200309753418, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8950164318084717, + "num_tokens": 879983931.0, + "step": 22664 + }, + { + "epoch": 2.883220964253912, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.999188423156738, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8896220326423645, + "num_tokens": 880025667.0, + "step": 22665 + }, + { + "epoch": 2.883348174532502, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.021164894104004, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.9052464962005615, + "num_tokens": 880063317.0, + "step": 22666 + }, + { + "epoch": 2.883475384811093, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.997674942016602, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8991358876228333, + "num_tokens": 880101645.0, + "step": 22667 + }, + { + "epoch": 2.883602595089683, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.00793743133545, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8938320279121399, + "num_tokens": 880143252.0, + "step": 22668 + }, + { + "epoch": 2.8837298053682736, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.910558700561523, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9089927673339844, + "num_tokens": 880183086.0, + "step": 22669 + }, + { + "epoch": 2.883857015646864, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.100210189819336, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8834351897239685, + "num_tokens": 880220293.0, + "step": 22670 + }, + { + "epoch": 2.8839842259254547, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.042337417602539, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8901435136795044, + "num_tokens": 880258069.0, + "step": 22671 + }, + { + "epoch": 2.884111436204045, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.02115249633789, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8978092670440674, + "num_tokens": 880294569.0, + "step": 22672 + }, + { + "epoch": 2.8842386464826357, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.901749610900879, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8972475528717041, + "num_tokens": 880337203.0, + "step": 22673 + }, + { + "epoch": 2.8843658567612263, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.016589164733887, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9107643365859985, + "num_tokens": 880376510.0, + "step": 22674 + }, + { + "epoch": 2.884493067039817, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.03023624420166, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8950769305229187, + "num_tokens": 880415769.0, + "step": 22675 + }, + { + "epoch": 2.8846202773184073, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.117643356323242, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8946303725242615, + "num_tokens": 880452137.0, + "step": 22676 + }, + { + "epoch": 2.884747487596998, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.959383964538574, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8995270729064941, + "num_tokens": 880492619.0, + "step": 22677 + }, + { + "epoch": 2.8848746978755884, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.96082592010498, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.9026919603347778, + "num_tokens": 880529032.0, + "step": 22678 + }, + { + "epoch": 2.885001908154179, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.984004974365234, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.9039626717567444, + "num_tokens": 880570615.0, + "step": 22679 + }, + { + "epoch": 2.8851291184327694, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.065629959106445, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.9002969264984131, + "num_tokens": 880610305.0, + "step": 22680 + }, + { + "epoch": 2.88525632871136, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.009264945983887, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8980639576911926, + "num_tokens": 880649342.0, + "step": 22681 + }, + { + "epoch": 2.8853835389899505, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.064178466796875, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8981783390045166, + "num_tokens": 880688013.0, + "step": 22682 + }, + { + "epoch": 2.885510749268541, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.97661304473877, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.9026580452919006, + "num_tokens": 880725400.0, + "step": 22683 + }, + { + "epoch": 2.8856379595471315, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.06822395324707, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.890852689743042, + "num_tokens": 880762085.0, + "step": 22684 + }, + { + "epoch": 2.885765169825722, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.947484016418457, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.91325443983078, + "num_tokens": 880794648.0, + "step": 22685 + }, + { + "epoch": 2.8858923801043126, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.064424514770508, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8967676162719727, + "num_tokens": 880832009.0, + "step": 22686 + }, + { + "epoch": 2.8860195903829027, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.018291473388672, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8882012367248535, + "num_tokens": 880870864.0, + "step": 22687 + }, + { + "epoch": 2.8861468006614936, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.030438423156738, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8952203392982483, + "num_tokens": 880904265.0, + "step": 22688 + }, + { + "epoch": 2.8862740109400837, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.097214698791504, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.895950436592102, + "num_tokens": 880938603.0, + "step": 22689 + }, + { + "epoch": 2.8864012212186747, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.94969654083252, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8940826058387756, + "num_tokens": 880979612.0, + "step": 22690 + }, + { + "epoch": 2.886528431497265, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.02660846710205, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.9018995761871338, + "num_tokens": 881014144.0, + "step": 22691 + }, + { + "epoch": 2.8866556417758558, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.968321800231934, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.9009109735488892, + "num_tokens": 881050765.0, + "step": 22692 + }, + { + "epoch": 2.886782852054446, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.920120239257812, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.904160737991333, + "num_tokens": 881095867.0, + "step": 22693 + }, + { + "epoch": 2.8869100623330364, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.037821769714355, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8916624784469604, + "num_tokens": 881134248.0, + "step": 22694 + }, + { + "epoch": 2.887037272611627, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.961633682250977, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9084218740463257, + "num_tokens": 881174069.0, + "step": 22695 + }, + { + "epoch": 2.8871644828902174, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.012527465820312, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8858771920204163, + "num_tokens": 881213292.0, + "step": 22696 + }, + { + "epoch": 2.887291693168808, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.027260780334473, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8952662348747253, + "num_tokens": 881253082.0, + "step": 22697 + }, + { + "epoch": 2.8874189034473985, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.947002410888672, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8954219818115234, + "num_tokens": 881291314.0, + "step": 22698 + }, + { + "epoch": 2.887546113725989, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.095664024353027, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9069176912307739, + "num_tokens": 881330850.0, + "step": 22699 + }, + { + "epoch": 2.8876733240045795, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.134836196899414, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8876302242279053, + "num_tokens": 881365165.0, + "step": 22700 + }, + { + "epoch": 2.88780053428317, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.038564682006836, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8903117179870605, + "num_tokens": 881404625.0, + "step": 22701 + }, + { + "epoch": 2.8879277445617606, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.0001010894775390625, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.087141990661621, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8886146545410156, + "num_tokens": 881443697.0, + "step": 22702 + }, + { + "epoch": 2.888054954840351, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.131732940673828, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8954079151153564, + "num_tokens": 881473221.0, + "step": 22703 + }, + { + "epoch": 2.8881821651189417, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.96129322052002, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8881871700286865, + "num_tokens": 881522727.0, + "step": 22704 + }, + { + "epoch": 2.888309375397532, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.010696411132812, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8890615105628967, + "num_tokens": 881564449.0, + "step": 22705 + }, + { + "epoch": 2.8884365856761227, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.077916145324707, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8984888792037964, + "num_tokens": 881605716.0, + "step": 22706 + }, + { + "epoch": 2.8885637959547132, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.944400787353516, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.9006031155586243, + "num_tokens": 881645767.0, + "step": 22707 + }, + { + "epoch": 2.8886910062333038, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.085604667663574, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8946815729141235, + "num_tokens": 881681802.0, + "step": 22708 + }, + { + "epoch": 2.8888182165118943, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.001593589782715, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8915411233901978, + "num_tokens": 881726818.0, + "step": 22709 + }, + { + "epoch": 2.888945426790485, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.062273025512695, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8922810554504395, + "num_tokens": 881770529.0, + "step": 22710 + }, + { + "epoch": 2.8890726370690754, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.028648376464844, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8997301459312439, + "num_tokens": 881808863.0, + "step": 22711 + }, + { + "epoch": 2.8891998473476654, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.031649589538574, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8921553492546082, + "num_tokens": 881848025.0, + "step": 22712 + }, + { + "epoch": 2.8893270576262564, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.076033592224121, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.9015437364578247, + "num_tokens": 881890896.0, + "step": 22713 + }, + { + "epoch": 2.8894542679048465, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.052499771118164, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8923598527908325, + "num_tokens": 881927337.0, + "step": 22714 + }, + { + "epoch": 2.8895814781834375, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.101030349731445, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8893966674804688, + "num_tokens": 881962998.0, + "step": 22715 + }, + { + "epoch": 2.8897086884620276, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.10737133026123, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8966677188873291, + "num_tokens": 882001569.0, + "step": 22716 + }, + { + "epoch": 2.889835898740618, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.079195976257324, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8853840827941895, + "num_tokens": 882036868.0, + "step": 22717 + }, + { + "epoch": 2.8899631090192086, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.009827613830566, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.908207356929779, + "num_tokens": 882073737.0, + "step": 22718 + }, + { + "epoch": 2.890090319297799, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.000503540039062, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8900840878486633, + "num_tokens": 882116345.0, + "step": 22719 + }, + { + "epoch": 2.8902175295763897, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.0482759475708, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8864721059799194, + "num_tokens": 882161727.0, + "step": 22720 + }, + { + "epoch": 2.89034473985498, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.074446678161621, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8990195989608765, + "num_tokens": 882202558.0, + "step": 22721 + }, + { + "epoch": 2.8904719501335707, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.091073036193848, + "learning_rate": 1e-06, + "loss": 0.3654, + "mean_token_accuracy": 0.873069167137146, + "num_tokens": 882244614.0, + "step": 22722 + }, + { + "epoch": 2.8905991604121613, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.160296440124512, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8880234956741333, + "num_tokens": 882280028.0, + "step": 22723 + }, + { + "epoch": 2.890726370690752, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.068679809570312, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8827563524246216, + "num_tokens": 882323980.0, + "step": 22724 + }, + { + "epoch": 2.8908535809693423, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.153097152709961, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.9016358256340027, + "num_tokens": 882356356.0, + "step": 22725 + }, + { + "epoch": 2.890980791247933, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.068328857421875, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9038382768630981, + "num_tokens": 882389030.0, + "step": 22726 + }, + { + "epoch": 2.8911080015265234, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.979998588562012, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.895706832408905, + "num_tokens": 882430443.0, + "step": 22727 + }, + { + "epoch": 2.891235211805114, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.014863014221191, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8980153799057007, + "num_tokens": 882465624.0, + "step": 22728 + }, + { + "epoch": 2.8913624220837044, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.005732536315918, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9098781943321228, + "num_tokens": 882505902.0, + "step": 22729 + }, + { + "epoch": 2.891489632362295, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.020215034484863, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8929144144058228, + "num_tokens": 882542328.0, + "step": 22730 + }, + { + "epoch": 2.8916168426408855, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.12527084350586, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.9015955328941345, + "num_tokens": 882578326.0, + "step": 22731 + }, + { + "epoch": 2.891744052919476, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.113232612609863, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8882999420166016, + "num_tokens": 882612972.0, + "step": 22732 + }, + { + "epoch": 2.8918712631980665, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.979622840881348, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9077681303024292, + "num_tokens": 882650467.0, + "step": 22733 + }, + { + "epoch": 2.891998473476657, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.048101425170898, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9040387272834778, + "num_tokens": 882690671.0, + "step": 22734 + }, + { + "epoch": 2.8921256837552476, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.121881484985352, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8915741443634033, + "num_tokens": 882726039.0, + "step": 22735 + }, + { + "epoch": 2.892252894033838, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.040806770324707, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8973997235298157, + "num_tokens": 882765152.0, + "step": 22736 + }, + { + "epoch": 2.892380104312428, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.141389846801758, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8939611911773682, + "num_tokens": 882800250.0, + "step": 22737 + }, + { + "epoch": 2.892507314591019, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.074688911437988, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8920036554336548, + "num_tokens": 882841183.0, + "step": 22738 + }, + { + "epoch": 2.8926345248696093, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.06783676147461, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8967068195343018, + "num_tokens": 882877079.0, + "step": 22739 + }, + { + "epoch": 2.8927617351482002, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.000442504882812, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9086660146713257, + "num_tokens": 882913260.0, + "step": 22740 + }, + { + "epoch": 2.8928889454267903, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.061741828918457, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8935216665267944, + "num_tokens": 882952234.0, + "step": 22741 + }, + { + "epoch": 2.893016155705381, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.033334732055664, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8991784453392029, + "num_tokens": 882992437.0, + "step": 22742 + }, + { + "epoch": 2.8931433659839714, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.147783279418945, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8888384103775024, + "num_tokens": 883025806.0, + "step": 22743 + }, + { + "epoch": 2.893270576262562, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.035780906677246, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8894524574279785, + "num_tokens": 883068242.0, + "step": 22744 + }, + { + "epoch": 2.8933977865411524, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.971787452697754, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.890407919883728, + "num_tokens": 883111965.0, + "step": 22745 + }, + { + "epoch": 2.893524996819743, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.926642417907715, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8953657150268555, + "num_tokens": 883153376.0, + "step": 22746 + }, + { + "epoch": 2.8936522070983335, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.023599624633789, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9050692319869995, + "num_tokens": 883189202.0, + "step": 22747 + }, + { + "epoch": 2.893779417376924, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.996567726135254, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9044426679611206, + "num_tokens": 883227019.0, + "step": 22748 + }, + { + "epoch": 2.8939066276555145, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.006741523742676, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9016337990760803, + "num_tokens": 883264729.0, + "step": 22749 + }, + { + "epoch": 2.894033837934105, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.132719993591309, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9008557796478271, + "num_tokens": 883298944.0, + "step": 22750 + }, + { + "epoch": 2.8941610482126956, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.922808647155762, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.9011569023132324, + "num_tokens": 883340591.0, + "step": 22751 + }, + { + "epoch": 2.894288258491286, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.094992637634277, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.9016571044921875, + "num_tokens": 883376529.0, + "step": 22752 + }, + { + "epoch": 2.8944154687698767, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.075984001159668, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8832416534423828, + "num_tokens": 883418312.0, + "step": 22753 + }, + { + "epoch": 2.894542679048467, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.020387649536133, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8893482685089111, + "num_tokens": 883458091.0, + "step": 22754 + }, + { + "epoch": 2.8946698893270577, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.006227493286133, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8935039043426514, + "num_tokens": 883498860.0, + "step": 22755 + }, + { + "epoch": 2.8947970996056482, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.119830131530762, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8913066983222961, + "num_tokens": 883534306.0, + "step": 22756 + }, + { + "epoch": 2.8949243098842388, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.108595848083496, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8992763161659241, + "num_tokens": 883569316.0, + "step": 22757 + }, + { + "epoch": 2.8950515201628293, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.05668830871582, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9093388319015503, + "num_tokens": 883600509.0, + "step": 22758 + }, + { + "epoch": 2.89517873044142, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.945311546325684, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8961889743804932, + "num_tokens": 883637547.0, + "step": 22759 + }, + { + "epoch": 2.89530594072001, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.168707847595215, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8822978734970093, + "num_tokens": 883676786.0, + "step": 22760 + }, + { + "epoch": 2.895433150998601, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.059711456298828, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8965365886688232, + "num_tokens": 883719667.0, + "step": 22761 + }, + { + "epoch": 2.895560361277191, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.07767391204834, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8940409421920776, + "num_tokens": 883753625.0, + "step": 22762 + }, + { + "epoch": 2.895687571555782, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.987239837646484, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8987901210784912, + "num_tokens": 883796994.0, + "step": 22763 + }, + { + "epoch": 2.895814781834372, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.98733901977539, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8905253410339355, + "num_tokens": 883840503.0, + "step": 22764 + }, + { + "epoch": 2.895941992112963, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.990046501159668, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.9064048528671265, + "num_tokens": 883882882.0, + "step": 22765 + }, + { + "epoch": 2.896069202391553, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.9389009475708, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.9061763286590576, + "num_tokens": 883922040.0, + "step": 22766 + }, + { + "epoch": 2.8961964126701436, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.996435165405273, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8816046714782715, + "num_tokens": 883966046.0, + "step": 22767 + }, + { + "epoch": 2.896323622948734, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.021041870117188, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.9042106866836548, + "num_tokens": 884003962.0, + "step": 22768 + }, + { + "epoch": 2.8964508332273247, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.92912769317627, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9153666496276855, + "num_tokens": 884041556.0, + "step": 22769 + }, + { + "epoch": 2.896578043505915, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.098654747009277, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8822721242904663, + "num_tokens": 884080075.0, + "step": 22770 + }, + { + "epoch": 2.8967052537845057, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.960529327392578, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.902123212814331, + "num_tokens": 884126205.0, + "step": 22771 + }, + { + "epoch": 2.8968324640630962, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.051360130310059, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8919245004653931, + "num_tokens": 884160878.0, + "step": 22772 + }, + { + "epoch": 2.8969596743416868, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.143916130065918, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.899125337600708, + "num_tokens": 884198880.0, + "step": 22773 + }, + { + "epoch": 2.8970868846202773, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.946866035461426, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9022200107574463, + "num_tokens": 884236996.0, + "step": 22774 + }, + { + "epoch": 2.897214094898868, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.07932186126709, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8942883610725403, + "num_tokens": 884274492.0, + "step": 22775 + }, + { + "epoch": 2.8973413051774584, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.93962287902832, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9055932760238647, + "num_tokens": 884317469.0, + "step": 22776 + }, + { + "epoch": 2.897468515456049, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.141029357910156, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8872577548027039, + "num_tokens": 884355104.0, + "step": 22777 + }, + { + "epoch": 2.8975957257346394, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.994446754455566, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.9033524990081787, + "num_tokens": 884398523.0, + "step": 22778 + }, + { + "epoch": 2.89772293601323, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.125879287719727, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9079755544662476, + "num_tokens": 884432668.0, + "step": 22779 + }, + { + "epoch": 2.8978501462918205, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.060697555541992, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8921799659729004, + "num_tokens": 884472790.0, + "step": 22780 + }, + { + "epoch": 2.897977356570411, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.042765617370605, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.9052473902702332, + "num_tokens": 884508978.0, + "step": 22781 + }, + { + "epoch": 2.8981045668490015, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.017151832580566, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9156641960144043, + "num_tokens": 884549792.0, + "step": 22782 + }, + { + "epoch": 2.898231777127592, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.072504043579102, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8930060863494873, + "num_tokens": 884591181.0, + "step": 22783 + }, + { + "epoch": 2.8983589874061826, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.092818260192871, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.9042677879333496, + "num_tokens": 884627559.0, + "step": 22784 + }, + { + "epoch": 2.8984861976847727, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.986720085144043, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8912374973297119, + "num_tokens": 884670283.0, + "step": 22785 + }, + { + "epoch": 2.8986134079633636, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.037459373474121, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9042443633079529, + "num_tokens": 884707709.0, + "step": 22786 + }, + { + "epoch": 2.8987406182419537, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.03911018371582, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.9003939628601074, + "num_tokens": 884744117.0, + "step": 22787 + }, + { + "epoch": 2.8988678285205447, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.02038860321045, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8986037969589233, + "num_tokens": 884784040.0, + "step": 22788 + }, + { + "epoch": 2.898995038799135, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.126813888549805, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9037386775016785, + "num_tokens": 884815487.0, + "step": 22789 + }, + { + "epoch": 2.8991222490777258, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.033950805664062, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8924094438552856, + "num_tokens": 884854187.0, + "step": 22790 + }, + { + "epoch": 2.899249459356316, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.999124526977539, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8968465328216553, + "num_tokens": 884891585.0, + "step": 22791 + }, + { + "epoch": 2.8993766696349064, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010156631469726562, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.056859016418457, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8971692323684692, + "num_tokens": 884928729.0, + "step": 22792 + }, + { + "epoch": 2.899503879913497, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.961814880371094, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8946554660797119, + "num_tokens": 884973120.0, + "step": 22793 + }, + { + "epoch": 2.8996310901920874, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.027376174926758, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8997283577919006, + "num_tokens": 885013222.0, + "step": 22794 + }, + { + "epoch": 2.899758300470678, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.025835990905762, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.9022612571716309, + "num_tokens": 885049658.0, + "step": 22795 + }, + { + "epoch": 2.8998855107492685, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.05306625366211, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8944203853607178, + "num_tokens": 885088200.0, + "step": 22796 + }, + { + "epoch": 2.900012721027859, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.993998527526855, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9166725277900696, + "num_tokens": 885124102.0, + "step": 22797 + }, + { + "epoch": 2.9001399313064495, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.175106048583984, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8929128050804138, + "num_tokens": 885156206.0, + "step": 22798 + }, + { + "epoch": 2.90026714158504, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.01370620727539, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8999539613723755, + "num_tokens": 885195869.0, + "step": 22799 + }, + { + "epoch": 2.9003943518636306, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.005438804626465, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8875791430473328, + "num_tokens": 885238091.0, + "step": 22800 + }, + { + "epoch": 2.900521562142221, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.105996131896973, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8888159990310669, + "num_tokens": 885275816.0, + "step": 22801 + }, + { + "epoch": 2.9006487724208116, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.019999504089355, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.9008360505104065, + "num_tokens": 885313390.0, + "step": 22802 + }, + { + "epoch": 2.900775982699402, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.000876426696777, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.9067273139953613, + "num_tokens": 885349860.0, + "step": 22803 + }, + { + "epoch": 2.9009031929779927, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.069140434265137, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9019747972488403, + "num_tokens": 885389491.0, + "step": 22804 + }, + { + "epoch": 2.9010304032565832, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.030769348144531, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8983316421508789, + "num_tokens": 885431591.0, + "step": 22805 + }, + { + "epoch": 2.9011576135351738, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.08946418762207, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.899995744228363, + "num_tokens": 885464914.0, + "step": 22806 + }, + { + "epoch": 2.9012848238137643, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.084681510925293, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9039338231086731, + "num_tokens": 885502669.0, + "step": 22807 + }, + { + "epoch": 2.901412034092355, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.159377098083496, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.9017618894577026, + "num_tokens": 885541829.0, + "step": 22808 + }, + { + "epoch": 2.9015392443709453, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.047196388244629, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.887757420539856, + "num_tokens": 885581988.0, + "step": 22809 + }, + { + "epoch": 2.9016664546495354, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.022316932678223, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8941543102264404, + "num_tokens": 885621605.0, + "step": 22810 + }, + { + "epoch": 2.9017936649281264, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.130951881408691, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.9051916599273682, + "num_tokens": 885657133.0, + "step": 22811 + }, + { + "epoch": 2.9019208752067165, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.031442642211914, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9017428159713745, + "num_tokens": 885692288.0, + "step": 22812 + }, + { + "epoch": 2.9020480854853075, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.00056266784668, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.9034319519996643, + "num_tokens": 885734345.0, + "step": 22813 + }, + { + "epoch": 2.9021752957638975, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.023816108703613, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9097820520401001, + "num_tokens": 885774629.0, + "step": 22814 + }, + { + "epoch": 2.902302506042488, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.055025100708008, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8970868587493896, + "num_tokens": 885816652.0, + "step": 22815 + }, + { + "epoch": 2.9024297163210786, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.126439094543457, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8916242718696594, + "num_tokens": 885852034.0, + "step": 22816 + }, + { + "epoch": 2.902556926599669, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.033058166503906, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9142709970474243, + "num_tokens": 885888038.0, + "step": 22817 + }, + { + "epoch": 2.9026841368782597, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.978718757629395, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.9045235514640808, + "num_tokens": 885931085.0, + "step": 22818 + }, + { + "epoch": 2.90281134715685, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.031557083129883, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9112807512283325, + "num_tokens": 885969598.0, + "step": 22819 + }, + { + "epoch": 2.9029385574354407, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.070341110229492, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8923909068107605, + "num_tokens": 886007343.0, + "step": 22820 + }, + { + "epoch": 2.9030657677140312, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.967061996459961, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9097851514816284, + "num_tokens": 886046665.0, + "step": 22821 + }, + { + "epoch": 2.9031929779926218, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.049639701843262, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8991737365722656, + "num_tokens": 886085338.0, + "step": 22822 + }, + { + "epoch": 2.9033201882712123, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.962048530578613, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9098794460296631, + "num_tokens": 886124950.0, + "step": 22823 + }, + { + "epoch": 2.903447398549803, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.09688949584961, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8969959020614624, + "num_tokens": 886160823.0, + "step": 22824 + }, + { + "epoch": 2.9035746088283934, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.234992980957031, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8957698345184326, + "num_tokens": 886197193.0, + "step": 22825 + }, + { + "epoch": 2.903701819106984, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.053427696228027, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8980056047439575, + "num_tokens": 886236256.0, + "step": 22826 + }, + { + "epoch": 2.9038290293855744, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2278556823730469e-05, + "grad_norm": 10.004620552062988, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8982371687889099, + "num_tokens": 886271123.0, + "step": 22827 + }, + { + "epoch": 2.903956239664165, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.009848594665527, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.891867995262146, + "num_tokens": 886310775.0, + "step": 22828 + }, + { + "epoch": 2.9040834499427555, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.018901824951172, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9017211198806763, + "num_tokens": 886347433.0, + "step": 22829 + }, + { + "epoch": 2.904210660221346, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.96152114868164, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8870847821235657, + "num_tokens": 886391779.0, + "step": 22830 + }, + { + "epoch": 2.9043378704999365, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.106593132019043, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8959476947784424, + "num_tokens": 886425695.0, + "step": 22831 + }, + { + "epoch": 2.904465080778527, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.019689559936523, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.897216260433197, + "num_tokens": 886461617.0, + "step": 22832 + }, + { + "epoch": 2.9045922910571176, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010204315185546875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.113763809204102, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.9011814594268799, + "num_tokens": 886493481.0, + "step": 22833 + }, + { + "epoch": 2.904719501335708, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.000565528869629, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8996173143386841, + "num_tokens": 886532236.0, + "step": 22834 + }, + { + "epoch": 2.904846711614298, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.073775291442871, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8978683352470398, + "num_tokens": 886573039.0, + "step": 22835 + }, + { + "epoch": 2.904973921892889, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.939641952514648, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9129208326339722, + "num_tokens": 886617495.0, + "step": 22836 + }, + { + "epoch": 2.9051011321714793, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.108132362365723, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8860394954681396, + "num_tokens": 886652981.0, + "step": 22837 + }, + { + "epoch": 2.9052283424500702, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.091034889221191, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8793249726295471, + "num_tokens": 886695818.0, + "step": 22838 + }, + { + "epoch": 2.9053555527286603, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.977965354919434, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.892886757850647, + "num_tokens": 886738336.0, + "step": 22839 + }, + { + "epoch": 2.905482763007251, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.9599027633667, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.9027751088142395, + "num_tokens": 886779941.0, + "step": 22840 + }, + { + "epoch": 2.9056099732858414, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.095932960510254, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8979348540306091, + "num_tokens": 886821767.0, + "step": 22841 + }, + { + "epoch": 2.905737183564432, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.958642959594727, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.9016380310058594, + "num_tokens": 886862877.0, + "step": 22842 + }, + { + "epoch": 2.9058643938430224, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.073837280273438, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.9012680053710938, + "num_tokens": 886898922.0, + "step": 22843 + }, + { + "epoch": 2.905991604121613, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.961892127990723, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.9017248749732971, + "num_tokens": 886940668.0, + "step": 22844 + }, + { + "epoch": 2.9061188144002035, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.088973999023438, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8922614455223083, + "num_tokens": 886980451.0, + "step": 22845 + }, + { + "epoch": 2.906246024678794, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.02020263671875, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8918374180793762, + "num_tokens": 887021264.0, + "step": 22846 + }, + { + "epoch": 2.9063732349573845, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.060441970825195, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.9017899036407471, + "num_tokens": 887057632.0, + "step": 22847 + }, + { + "epoch": 2.906500445235975, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.032401084899902, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8971248865127563, + "num_tokens": 887101277.0, + "step": 22848 + }, + { + "epoch": 2.9066276555145656, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.14187240600586, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9018076658248901, + "num_tokens": 887135773.0, + "step": 22849 + }, + { + "epoch": 2.906754865793156, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.036545753479004, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8907582759857178, + "num_tokens": 887175169.0, + "step": 22850 + }, + { + "epoch": 2.9068820760717466, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.040763854980469, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8999191522598267, + "num_tokens": 887214639.0, + "step": 22851 + }, + { + "epoch": 2.907009286350337, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.001936912536621, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8953686356544495, + "num_tokens": 887252088.0, + "step": 22852 + }, + { + "epoch": 2.9071364966289277, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.004956245422363, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8949642181396484, + "num_tokens": 887291137.0, + "step": 22853 + }, + { + "epoch": 2.9072637069075182, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.051339149475098, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.9002236723899841, + "num_tokens": 887328779.0, + "step": 22854 + }, + { + "epoch": 2.9073909171861088, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.01053237915039, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.8981999158859253, + "num_tokens": 887361873.0, + "step": 22855 + }, + { + "epoch": 2.9075181274646993, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.014172554016113, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.9044762849807739, + "num_tokens": 887400860.0, + "step": 22856 + }, + { + "epoch": 2.90764533774329, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.08244514465332, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8938754200935364, + "num_tokens": 887439180.0, + "step": 22857 + }, + { + "epoch": 2.90777254802188, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.065716743469238, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8857837915420532, + "num_tokens": 887472481.0, + "step": 22858 + }, + { + "epoch": 2.907899758300471, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.041274070739746, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.9043710231781006, + "num_tokens": 887505660.0, + "step": 22859 + }, + { + "epoch": 2.908026968579061, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.155284881591797, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8777081966400146, + "num_tokens": 887543021.0, + "step": 22860 + }, + { + "epoch": 2.908154178857652, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.964273452758789, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8976852297782898, + "num_tokens": 887584194.0, + "step": 22861 + }, + { + "epoch": 2.908281389136242, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.191995620727539, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8962672352790833, + "num_tokens": 887616042.0, + "step": 22862 + }, + { + "epoch": 2.908408599414833, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.022711753845215, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.9001151323318481, + "num_tokens": 887654953.0, + "step": 22863 + }, + { + "epoch": 2.908535809693423, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.948858261108398, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.9005800485610962, + "num_tokens": 887697312.0, + "step": 22864 + }, + { + "epoch": 2.9086630199720136, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.006373405456543, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.889872133731842, + "num_tokens": 887743916.0, + "step": 22865 + }, + { + "epoch": 2.908790230250604, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.03990650177002, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9071668982505798, + "num_tokens": 887776260.0, + "step": 22866 + }, + { + "epoch": 2.9089174405291947, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.0914888381958, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.896950364112854, + "num_tokens": 887813626.0, + "step": 22867 + }, + { + "epoch": 2.909044650807785, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.013434410095215, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.886784017086029, + "num_tokens": 887850870.0, + "step": 22868 + }, + { + "epoch": 2.9091718610863757, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.031779289245605, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.900787353515625, + "num_tokens": 887894766.0, + "step": 22869 + }, + { + "epoch": 2.9092990713649662, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.097356796264648, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.896900475025177, + "num_tokens": 887934512.0, + "step": 22870 + }, + { + "epoch": 2.9094262816435568, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.131766319274902, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.889777421951294, + "num_tokens": 887971902.0, + "step": 22871 + }, + { + "epoch": 2.9095534919221473, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.03685474395752, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9021046161651611, + "num_tokens": 888011073.0, + "step": 22872 + }, + { + "epoch": 2.909680702200738, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.0338134765625, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.894088625907898, + "num_tokens": 888057397.0, + "step": 22873 + }, + { + "epoch": 2.9098079124793284, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.057215690612793, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8936671018600464, + "num_tokens": 888092770.0, + "step": 22874 + }, + { + "epoch": 2.909935122757919, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.08453369140625, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8853160738945007, + "num_tokens": 888134708.0, + "step": 22875 + }, + { + "epoch": 2.9100623330365094, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.953323364257812, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.9049885869026184, + "num_tokens": 888178350.0, + "step": 22876 + }, + { + "epoch": 2.9101895433151, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.06124496459961, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.9046515226364136, + "num_tokens": 888211830.0, + "step": 22877 + }, + { + "epoch": 2.9103167535936905, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.041725158691406, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8937445282936096, + "num_tokens": 888249551.0, + "step": 22878 + }, + { + "epoch": 2.910443963872281, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.95993709564209, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9097933173179626, + "num_tokens": 888287142.0, + "step": 22879 + }, + { + "epoch": 2.9105711741508715, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.045564651489258, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8952051401138306, + "num_tokens": 888327621.0, + "step": 22880 + }, + { + "epoch": 2.910698384429462, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.003539085388184, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.906613826751709, + "num_tokens": 888368038.0, + "step": 22881 + }, + { + "epoch": 2.9108255947080526, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.998945236206055, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.9002135992050171, + "num_tokens": 888415129.0, + "step": 22882 + }, + { + "epoch": 2.9109528049866427, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.113530158996582, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.892890989780426, + "num_tokens": 888454237.0, + "step": 22883 + }, + { + "epoch": 2.9110800152652336, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.070836067199707, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8918671607971191, + "num_tokens": 888490960.0, + "step": 22884 + }, + { + "epoch": 2.9112072255438237, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.12734603881836, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8947861194610596, + "num_tokens": 888530899.0, + "step": 22885 + }, + { + "epoch": 2.9113344358224147, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.01688289642334, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8997434973716736, + "num_tokens": 888571438.0, + "step": 22886 + }, + { + "epoch": 2.9114616461010048, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.990049362182617, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8989269137382507, + "num_tokens": 888611027.0, + "step": 22887 + }, + { + "epoch": 2.9115888563795957, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.179906845092773, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8893367052078247, + "num_tokens": 888646819.0, + "step": 22888 + }, + { + "epoch": 2.911716066658186, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.121352195739746, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.9001973271369934, + "num_tokens": 888681779.0, + "step": 22889 + }, + { + "epoch": 2.9118432769367764, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.082978248596191, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.9010732173919678, + "num_tokens": 888716637.0, + "step": 22890 + }, + { + "epoch": 2.911970487215367, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.111112594604492, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8932676315307617, + "num_tokens": 888749726.0, + "step": 22891 + }, + { + "epoch": 2.9120976974939574, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.986323356628418, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.91173255443573, + "num_tokens": 888791347.0, + "step": 22892 + }, + { + "epoch": 2.912224907772548, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.020392417907715, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8980724811553955, + "num_tokens": 888829992.0, + "step": 22893 + }, + { + "epoch": 2.9123521180511385, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.963608741760254, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9091010093688965, + "num_tokens": 888874810.0, + "step": 22894 + }, + { + "epoch": 2.912479328329729, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.10496711730957, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8857917785644531, + "num_tokens": 888911106.0, + "step": 22895 + }, + { + "epoch": 2.9126065386083195, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.030866622924805, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.890133261680603, + "num_tokens": 888957613.0, + "step": 22896 + }, + { + "epoch": 2.91273374888691, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.036662101745605, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.9009349942207336, + "num_tokens": 888996735.0, + "step": 22897 + }, + { + "epoch": 2.9128609591655006, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.025299072265625, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8967970013618469, + "num_tokens": 889035676.0, + "step": 22898 + }, + { + "epoch": 2.912988169444091, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.062506675720215, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.9021040201187134, + "num_tokens": 889073729.0, + "step": 22899 + }, + { + "epoch": 2.9131153797226816, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.016853332519531, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8923057317733765, + "num_tokens": 889117986.0, + "step": 22900 + }, + { + "epoch": 2.913242590001272, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.134424209594727, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.890195369720459, + "num_tokens": 889157547.0, + "step": 22901 + }, + { + "epoch": 2.9133698002798627, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.998055458068848, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.9007450938224792, + "num_tokens": 889197544.0, + "step": 22902 + }, + { + "epoch": 2.9134970105584532, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.105549812316895, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8939544558525085, + "num_tokens": 889236660.0, + "step": 22903 + }, + { + "epoch": 2.9136242208370438, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 12.473424911499023, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8899455070495605, + "num_tokens": 889278437.0, + "step": 22904 + }, + { + "epoch": 2.9137514311156343, + "ewc_loss": 0.023681640625, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.341104507446289e-05, + "grad_norm": 10.08526611328125, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9067797064781189, + "num_tokens": 889320529.0, + "step": 22905 + }, + { + "epoch": 2.913878641394225, + "ewc_loss": 0.0223388671875, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2099742889404297e-05, + "grad_norm": 9.878437042236328, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.883661150932312, + "num_tokens": 889363633.0, + "step": 22906 + }, + { + "epoch": 2.9140058516728153, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.166030883789062, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9100228548049927, + "num_tokens": 889401417.0, + "step": 22907 + }, + { + "epoch": 2.9141330619514054, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.992563247680664, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.9044514894485474, + "num_tokens": 889440341.0, + "step": 22908 + }, + { + "epoch": 2.9142602722299964, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.094512939453125, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8980810046195984, + "num_tokens": 889479367.0, + "step": 22909 + }, + { + "epoch": 2.9143874825085865, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.199280738830566, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8967717885971069, + "num_tokens": 889516368.0, + "step": 22910 + }, + { + "epoch": 2.9145146927871775, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.908064842224121, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.9004641771316528, + "num_tokens": 889560797.0, + "step": 22911 + }, + { + "epoch": 2.9146419030657675, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.1309175491333, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8979638814926147, + "num_tokens": 889600141.0, + "step": 22912 + }, + { + "epoch": 2.914769113344358, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.089455604553223, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8956958651542664, + "num_tokens": 889642446.0, + "step": 22913 + }, + { + "epoch": 2.9148963236229486, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.03718376159668, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9034727811813354, + "num_tokens": 889683470.0, + "step": 22914 + }, + { + "epoch": 2.915023533901539, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.089689254760742, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8964711427688599, + "num_tokens": 889723007.0, + "step": 22915 + }, + { + "epoch": 2.9151507441801296, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.98829174041748, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8991348743438721, + "num_tokens": 889760684.0, + "step": 22916 + }, + { + "epoch": 2.91527795445872, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.109980583190918, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8961485624313354, + "num_tokens": 889798118.0, + "step": 22917 + }, + { + "epoch": 2.9154051647373107, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.019716262817383, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8953968286514282, + "num_tokens": 889845658.0, + "step": 22918 + }, + { + "epoch": 2.9155323750159012, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.06663990020752, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8997301459312439, + "num_tokens": 889885792.0, + "step": 22919 + }, + { + "epoch": 2.9156595852944918, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.981027603149414, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8911183476448059, + "num_tokens": 889930469.0, + "step": 22920 + }, + { + "epoch": 2.9157867955730823, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.98194408416748, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.9017089605331421, + "num_tokens": 889975518.0, + "step": 22921 + }, + { + "epoch": 2.915914005851673, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.03858757019043, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.9046603441238403, + "num_tokens": 890014351.0, + "step": 22922 + }, + { + "epoch": 2.9160412161302633, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.052337646484375, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.9048694372177124, + "num_tokens": 890052495.0, + "step": 22923 + }, + { + "epoch": 2.916168426408854, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.066633224487305, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8979086875915527, + "num_tokens": 890091445.0, + "step": 22924 + }, + { + "epoch": 2.9162956366874444, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.092751502990723, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8966737985610962, + "num_tokens": 890128141.0, + "step": 22925 + }, + { + "epoch": 2.916422846966035, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.009428977966309, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8998247385025024, + "num_tokens": 890168428.0, + "step": 22926 + }, + { + "epoch": 2.9165500572446255, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.077154159545898, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.897144079208374, + "num_tokens": 890202509.0, + "step": 22927 + }, + { + "epoch": 2.916677267523216, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.117321014404297, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.9007651805877686, + "num_tokens": 890241051.0, + "step": 22928 + }, + { + "epoch": 2.9168044778018065, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.144233703613281, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8813693523406982, + "num_tokens": 890273464.0, + "step": 22929 + }, + { + "epoch": 2.916931688080397, + "ewc_loss": 0.0225830078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.106754302978516, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8929518461227417, + "num_tokens": 890308407.0, + "step": 22930 + }, + { + "epoch": 2.917058898358987, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.018087387084961, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.89677894115448, + "num_tokens": 890347912.0, + "step": 22931 + }, + { + "epoch": 2.917186108637578, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.035171508789062, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9056781530380249, + "num_tokens": 890386734.0, + "step": 22932 + }, + { + "epoch": 2.917313318916168, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.051178932189941, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.9070661067962646, + "num_tokens": 890420860.0, + "step": 22933 + }, + { + "epoch": 2.917440529194759, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.189624786376953, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.892103374004364, + "num_tokens": 890452644.0, + "step": 22934 + }, + { + "epoch": 2.9175677394733492, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.040589332580566, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8958837985992432, + "num_tokens": 890489509.0, + "step": 22935 + }, + { + "epoch": 2.91769494975194, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.931133270263672, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.904039204120636, + "num_tokens": 890537500.0, + "step": 22936 + }, + { + "epoch": 2.9178221600305303, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.107807159423828, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.9027731418609619, + "num_tokens": 890573198.0, + "step": 22937 + }, + { + "epoch": 2.917949370309121, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.092286109924316, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9038485288619995, + "num_tokens": 890606174.0, + "step": 22938 + }, + { + "epoch": 2.9180765805877114, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.004150390625, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.900336742401123, + "num_tokens": 890644565.0, + "step": 22939 + }, + { + "epoch": 2.918203790866302, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.012351989746094, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8979027271270752, + "num_tokens": 890683736.0, + "step": 22940 + }, + { + "epoch": 2.9183310011448924, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.096782684326172, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8945583701133728, + "num_tokens": 890722775.0, + "step": 22941 + }, + { + "epoch": 2.918458211423483, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.947919845581055, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8880177736282349, + "num_tokens": 890765387.0, + "step": 22942 + }, + { + "epoch": 2.9185854217020735, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.123744010925293, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8904067277908325, + "num_tokens": 890808138.0, + "step": 22943 + }, + { + "epoch": 2.918712631980664, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.946495056152344, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8953045010566711, + "num_tokens": 890851372.0, + "step": 22944 + }, + { + "epoch": 2.9188398422592545, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.08473014831543, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8988639712333679, + "num_tokens": 890888463.0, + "step": 22945 + }, + { + "epoch": 2.918967052537845, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.00239086151123, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8939571976661682, + "num_tokens": 890931231.0, + "step": 22946 + }, + { + "epoch": 2.9190942628164356, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.968685150146484, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.9071853160858154, + "num_tokens": 890967358.0, + "step": 22947 + }, + { + "epoch": 2.919221473095026, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.01686954498291, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8996078968048096, + "num_tokens": 891002760.0, + "step": 22948 + }, + { + "epoch": 2.9193486833736166, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.991766929626465, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.9091760516166687, + "num_tokens": 891039461.0, + "step": 22949 + }, + { + "epoch": 2.919475893652207, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.030057907104492, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8900288343429565, + "num_tokens": 891076610.0, + "step": 22950 + }, + { + "epoch": 2.9196031039307977, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.079642295837402, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8981638550758362, + "num_tokens": 891111468.0, + "step": 22951 + }, + { + "epoch": 2.9197303142093882, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.872126579284668, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.9060393571853638, + "num_tokens": 891155797.0, + "step": 22952 + }, + { + "epoch": 2.9198575244879788, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.112051963806152, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.9023897647857666, + "num_tokens": 891190538.0, + "step": 22953 + }, + { + "epoch": 2.9199847347665693, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.013361930847168, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8972286581993103, + "num_tokens": 891228982.0, + "step": 22954 + }, + { + "epoch": 2.92011194504516, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.139803886413574, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.9011015892028809, + "num_tokens": 891262022.0, + "step": 22955 + }, + { + "epoch": 2.92023915532375, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.07131290435791, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8982025384902954, + "num_tokens": 891297415.0, + "step": 22956 + }, + { + "epoch": 2.920366365602341, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.140802383422852, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8928651213645935, + "num_tokens": 891335655.0, + "step": 22957 + }, + { + "epoch": 2.920493575880931, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.93863296508789, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.9052124619483948, + "num_tokens": 891382228.0, + "step": 22958 + }, + { + "epoch": 2.920620786159522, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010251998901367188, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.1638765335083, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8919283151626587, + "num_tokens": 891421306.0, + "step": 22959 + }, + { + "epoch": 2.920747996438112, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.174574851989746, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8832899332046509, + "num_tokens": 891460184.0, + "step": 22960 + }, + { + "epoch": 2.920875206716703, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.110821723937988, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8864918947219849, + "num_tokens": 891499801.0, + "step": 22961 + }, + { + "epoch": 2.921002416995293, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.070627212524414, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8869037628173828, + "num_tokens": 891537556.0, + "step": 22962 + }, + { + "epoch": 2.9211296272738836, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.003558158874512, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.9059512615203857, + "num_tokens": 891577077.0, + "step": 22963 + }, + { + "epoch": 2.921256837552474, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.020681381225586, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.9007526636123657, + "num_tokens": 891615087.0, + "step": 22964 + }, + { + "epoch": 2.9213840478310646, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.102510452270508, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.9013649821281433, + "num_tokens": 891656884.0, + "step": 22965 + }, + { + "epoch": 2.921511258109655, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.01147747039795, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8996115922927856, + "num_tokens": 891694993.0, + "step": 22966 + }, + { + "epoch": 2.9216384683882457, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.090860366821289, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.900078535079956, + "num_tokens": 891728984.0, + "step": 22967 + }, + { + "epoch": 2.9217656786668362, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.031323432922363, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9008741974830627, + "num_tokens": 891769835.0, + "step": 22968 + }, + { + "epoch": 2.9218928889454268, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.215394973754883, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8917005658149719, + "num_tokens": 891807946.0, + "step": 22969 + }, + { + "epoch": 2.9220200992240173, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.205506324768066, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8932613134384155, + "num_tokens": 891837738.0, + "step": 22970 + }, + { + "epoch": 2.922147309502608, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.106953620910645, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8989410400390625, + "num_tokens": 891874003.0, + "step": 22971 + }, + { + "epoch": 2.9222745197811983, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 9.997084617614746, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.9029960632324219, + "num_tokens": 891906668.0, + "step": 22972 + }, + { + "epoch": 2.922401730059789, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.071065902709961, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.9055487513542175, + "num_tokens": 891946721.0, + "step": 22973 + }, + { + "epoch": 2.9225289403383794, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.98115062713623, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8969902396202087, + "num_tokens": 891985619.0, + "step": 22974 + }, + { + "epoch": 2.92265615061697, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.036895751953125, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8943071961402893, + "num_tokens": 892029891.0, + "step": 22975 + }, + { + "epoch": 2.9227833608955605, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.99573040008545, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8921629786491394, + "num_tokens": 892073482.0, + "step": 22976 + }, + { + "epoch": 2.922910571174151, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.14168643951416, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9074315428733826, + "num_tokens": 892112941.0, + "step": 22977 + }, + { + "epoch": 2.9230377814527415, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.126494407653809, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8984167575836182, + "num_tokens": 892147057.0, + "step": 22978 + }, + { + "epoch": 2.923164991731332, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.112565994262695, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.897167444229126, + "num_tokens": 892180741.0, + "step": 22979 + }, + { + "epoch": 2.9232922020099226, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.076981544494629, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.9013947248458862, + "num_tokens": 892215392.0, + "step": 22980 + }, + { + "epoch": 2.9234194122885127, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.055020332336426, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.9019783139228821, + "num_tokens": 892253000.0, + "step": 22981 + }, + { + "epoch": 2.9235466225671036, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.124796867370605, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.896152675151825, + "num_tokens": 892284839.0, + "step": 22982 + }, + { + "epoch": 2.9236738328456937, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.173827171325684, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8908241987228394, + "num_tokens": 892318015.0, + "step": 22983 + }, + { + "epoch": 2.9238010431242847, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 9.980753898620605, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.895350992679596, + "num_tokens": 892355863.0, + "step": 22984 + }, + { + "epoch": 2.9239282534028748, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.037195205688477, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8961371183395386, + "num_tokens": 892395248.0, + "step": 22985 + }, + { + "epoch": 2.9240554636814657, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.099443435668945, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.9036848545074463, + "num_tokens": 892427459.0, + "step": 22986 + }, + { + "epoch": 2.924182673960056, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.039682388305664, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8905304670333862, + "num_tokens": 892468592.0, + "step": 22987 + }, + { + "epoch": 2.9243098842386464, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.034764289855957, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8883661031723022, + "num_tokens": 892510421.0, + "step": 22988 + }, + { + "epoch": 2.924437094517237, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.012094497680664, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8995007276535034, + "num_tokens": 892552392.0, + "step": 22989 + }, + { + "epoch": 2.9245643047958274, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.285467147827148, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8992536067962646, + "num_tokens": 892589649.0, + "step": 22990 + }, + { + "epoch": 2.924691515074418, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.135499000549316, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.9019367694854736, + "num_tokens": 892620998.0, + "step": 22991 + }, + { + "epoch": 2.9248187253530085, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.11733341217041, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8944224715232849, + "num_tokens": 892655603.0, + "step": 22992 + }, + { + "epoch": 2.924945935631599, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.02429485321045, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8900905847549438, + "num_tokens": 892697167.0, + "step": 22993 + }, + { + "epoch": 2.9250731459101895, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.041410446166992, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8907341361045837, + "num_tokens": 892734505.0, + "step": 22994 + }, + { + "epoch": 2.92520035618878, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.023296356201172, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9042614102363586, + "num_tokens": 892771921.0, + "step": 22995 + }, + { + "epoch": 2.9253275664673706, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.046789169311523, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.88686203956604, + "num_tokens": 892812254.0, + "step": 22996 + }, + { + "epoch": 2.925454776745961, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.996262550354004, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8971676826477051, + "num_tokens": 892851288.0, + "step": 22997 + }, + { + "epoch": 2.9255819870245516, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.117738723754883, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.9018017649650574, + "num_tokens": 892886332.0, + "step": 22998 + }, + { + "epoch": 2.925709197303142, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.01683521270752, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9048508405685425, + "num_tokens": 892925004.0, + "step": 22999 + }, + { + "epoch": 2.9258364075817327, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.06442928314209, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.9055230617523193, + "num_tokens": 892957847.0, + "step": 23000 + }, + { + "epoch": 2.925963617860323, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 11.024640083312988, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8765784502029419, + "num_tokens": 892997439.0, + "step": 23001 + }, + { + "epoch": 2.9260908281389137, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2934207916259766e-05, + "grad_norm": 10.059807777404785, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8980098962783813, + "num_tokens": 893031333.0, + "step": 23002 + }, + { + "epoch": 2.9262180384175043, + "ewc_loss": 0.0224609375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2218952178955078e-05, + "grad_norm": 9.873418807983398, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.908169150352478, + "num_tokens": 893065107.0, + "step": 23003 + }, + { + "epoch": 2.926345248696095, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.06688117980957, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.900693953037262, + "num_tokens": 893102474.0, + "step": 23004 + }, + { + "epoch": 2.9264724589746853, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2814998626708984e-05, + "grad_norm": 9.99764347076416, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.9065196514129639, + "num_tokens": 893144326.0, + "step": 23005 + }, + { + "epoch": 2.9265996692532754, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.050174713134766, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.900916576385498, + "num_tokens": 893176943.0, + "step": 23006 + }, + { + "epoch": 2.9267268795318664, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.06989574432373, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9048836827278137, + "num_tokens": 893215222.0, + "step": 23007 + }, + { + "epoch": 2.9268540898104565, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.940851211547852, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8986977338790894, + "num_tokens": 893259738.0, + "step": 23008 + }, + { + "epoch": 2.9269813000890474, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.117615699768066, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8818238973617554, + "num_tokens": 893299772.0, + "step": 23009 + }, + { + "epoch": 2.9271085103676375, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.963029861450195, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8917117118835449, + "num_tokens": 893343918.0, + "step": 23010 + }, + { + "epoch": 2.927235720646228, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.031124114990234, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8965824842453003, + "num_tokens": 893384367.0, + "step": 23011 + }, + { + "epoch": 2.9273629309248186, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.09183406829834, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8833352327346802, + "num_tokens": 893426203.0, + "step": 23012 + }, + { + "epoch": 2.927490141203409, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.966897010803223, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.9044309854507446, + "num_tokens": 893467674.0, + "step": 23013 + }, + { + "epoch": 2.9276173514819996, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.025577545166016, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8925020694732666, + "num_tokens": 893510599.0, + "step": 23014 + }, + { + "epoch": 2.92774456176059, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.185968399047852, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8940373659133911, + "num_tokens": 893547615.0, + "step": 23015 + }, + { + "epoch": 2.9278717720391807, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.114365577697754, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.9027194380760193, + "num_tokens": 893580074.0, + "step": 23016 + }, + { + "epoch": 2.9279989823177712, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.091460227966309, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8808077573776245, + "num_tokens": 893625642.0, + "step": 23017 + }, + { + "epoch": 2.9281261925963618, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.063874244689941, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8896984457969666, + "num_tokens": 893667044.0, + "step": 23018 + }, + { + "epoch": 2.9282534028749523, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.097046852111816, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8970962762832642, + "num_tokens": 893700135.0, + "step": 23019 + }, + { + "epoch": 2.928380613153543, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.968278884887695, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.9037368893623352, + "num_tokens": 893741902.0, + "step": 23020 + }, + { + "epoch": 2.9285078234321333, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.092309951782227, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9068683385848999, + "num_tokens": 893773226.0, + "step": 23021 + }, + { + "epoch": 2.928635033710724, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.959458351135254, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.9039647579193115, + "num_tokens": 893816900.0, + "step": 23022 + }, + { + "epoch": 2.9287622439893144, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.978928565979004, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.9053278565406799, + "num_tokens": 893861250.0, + "step": 23023 + }, + { + "epoch": 2.928889454267905, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.996313095092773, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.9024441242218018, + "num_tokens": 893903091.0, + "step": 23024 + }, + { + "epoch": 2.9290166645464955, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.132856369018555, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.9030070304870605, + "num_tokens": 893936895.0, + "step": 23025 + }, + { + "epoch": 2.929143874825086, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.102032661437988, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8926110863685608, + "num_tokens": 893978546.0, + "step": 23026 + }, + { + "epoch": 2.9292710851036765, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.152734756469727, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8874857425689697, + "num_tokens": 894014991.0, + "step": 23027 + }, + { + "epoch": 2.929398295382267, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.081327438354492, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.904401421546936, + "num_tokens": 894057745.0, + "step": 23028 + }, + { + "epoch": 2.929525505660857, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.929183959960938, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8998152613639832, + "num_tokens": 894099737.0, + "step": 23029 + }, + { + "epoch": 2.929652715939448, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.21794605255127, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8826316595077515, + "num_tokens": 894137302.0, + "step": 23030 + }, + { + "epoch": 2.929779926218038, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.116662979125977, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9010343551635742, + "num_tokens": 894176390.0, + "step": 23031 + }, + { + "epoch": 2.929907136496629, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.14233684539795, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.9012744426727295, + "num_tokens": 894214622.0, + "step": 23032 + }, + { + "epoch": 2.9300343467752192, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.980619430541992, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8937009572982788, + "num_tokens": 894258587.0, + "step": 23033 + }, + { + "epoch": 2.93016155705381, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.15904712677002, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.890555739402771, + "num_tokens": 894296697.0, + "step": 23034 + }, + { + "epoch": 2.9302887673324003, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.086750984191895, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8905053734779358, + "num_tokens": 894333788.0, + "step": 23035 + }, + { + "epoch": 2.930415977610991, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.083823204040527, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8962420225143433, + "num_tokens": 894373690.0, + "step": 23036 + }, + { + "epoch": 2.9305431878895813, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.109562873840332, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.890605092048645, + "num_tokens": 894412383.0, + "step": 23037 + }, + { + "epoch": 2.930670398168172, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.015296936035156, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.9070952534675598, + "num_tokens": 894451387.0, + "step": 23038 + }, + { + "epoch": 2.9307976084467624, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.2755126953125, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.9034985303878784, + "num_tokens": 894482895.0, + "step": 23039 + }, + { + "epoch": 2.930924818725353, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.129810333251953, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.896212100982666, + "num_tokens": 894520399.0, + "step": 23040 + }, + { + "epoch": 2.9310520290039435, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.142621994018555, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8925119638442993, + "num_tokens": 894556833.0, + "step": 23041 + }, + { + "epoch": 2.931179239282534, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.08165168762207, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8910274505615234, + "num_tokens": 894593815.0, + "step": 23042 + }, + { + "epoch": 2.9313064495611245, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.983455657958984, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8971369862556458, + "num_tokens": 894631912.0, + "step": 23043 + }, + { + "epoch": 2.931433659839715, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.196131706237793, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8966047763824463, + "num_tokens": 894663403.0, + "step": 23044 + }, + { + "epoch": 2.9315608701183056, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.99106502532959, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9081986546516418, + "num_tokens": 894701047.0, + "step": 23045 + }, + { + "epoch": 2.931688080396896, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.054154396057129, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8958235383033752, + "num_tokens": 894740440.0, + "step": 23046 + }, + { + "epoch": 2.9318152906754866, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.057904243469238, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8970763087272644, + "num_tokens": 894781474.0, + "step": 23047 + }, + { + "epoch": 2.931942500954077, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.055444717407227, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8995475769042969, + "num_tokens": 894822383.0, + "step": 23048 + }, + { + "epoch": 2.9320697112326677, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.046346664428711, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8847298622131348, + "num_tokens": 894865340.0, + "step": 23049 + }, + { + "epoch": 2.932196921511258, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.96173095703125, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.9047353267669678, + "num_tokens": 894906432.0, + "step": 23050 + }, + { + "epoch": 2.9323241317898487, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.101791381835938, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8990533351898193, + "num_tokens": 894942832.0, + "step": 23051 + }, + { + "epoch": 2.9324513420684393, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.163655281066895, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8907228112220764, + "num_tokens": 894980631.0, + "step": 23052 + }, + { + "epoch": 2.93257855234703, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.962563514709473, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9033469557762146, + "num_tokens": 895022073.0, + "step": 23053 + }, + { + "epoch": 2.93270576262562, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.079424858093262, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.892883837223053, + "num_tokens": 895056689.0, + "step": 23054 + }, + { + "epoch": 2.932832972904211, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.103047370910645, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.9050306081771851, + "num_tokens": 895093263.0, + "step": 23055 + }, + { + "epoch": 2.932960183182801, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.035499572753906, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.9010720252990723, + "num_tokens": 895137522.0, + "step": 23056 + }, + { + "epoch": 2.933087393461392, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.166648864746094, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.9008277654647827, + "num_tokens": 895169515.0, + "step": 23057 + }, + { + "epoch": 2.933214603739982, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.091327667236328, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.9035326242446899, + "num_tokens": 895208436.0, + "step": 23058 + }, + { + "epoch": 2.933341814018573, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.95536994934082, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8992232084274292, + "num_tokens": 895247929.0, + "step": 23059 + }, + { + "epoch": 2.933469024297163, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.232665061950684, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8974698781967163, + "num_tokens": 895282590.0, + "step": 23060 + }, + { + "epoch": 2.9335962345757536, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 9.985902786254883, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8965915441513062, + "num_tokens": 895323122.0, + "step": 23061 + }, + { + "epoch": 2.933723444854344, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.148953437805176, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9055077433586121, + "num_tokens": 895357021.0, + "step": 23062 + }, + { + "epoch": 2.9338506551329346, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.067361831665039, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8925586938858032, + "num_tokens": 895398230.0, + "step": 23063 + }, + { + "epoch": 2.933977865411525, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.14873218536377, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8826301097869873, + "num_tokens": 895432761.0, + "step": 23064 + }, + { + "epoch": 2.9341050756901157, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.080799102783203, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8901772499084473, + "num_tokens": 895472603.0, + "step": 23065 + }, + { + "epoch": 2.934232285968706, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.075421333312988, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9035068154335022, + "num_tokens": 895510145.0, + "step": 23066 + }, + { + "epoch": 2.9343594962472968, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.345087051391602, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8813186287879944, + "num_tokens": 895539240.0, + "step": 23067 + }, + { + "epoch": 2.9344867065258873, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.077018737792969, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8963624238967896, + "num_tokens": 895573086.0, + "step": 23068 + }, + { + "epoch": 2.934613916804478, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 9.988228797912598, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8989614248275757, + "num_tokens": 895608684.0, + "step": 23069 + }, + { + "epoch": 2.9347411270830683, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.04499626159668, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.903473436832428, + "num_tokens": 895646729.0, + "step": 23070 + }, + { + "epoch": 2.934868337361659, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.094090461730957, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8915436863899231, + "num_tokens": 895685846.0, + "step": 23071 + }, + { + "epoch": 2.9349955476402494, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.018725395202637, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.898078441619873, + "num_tokens": 895724893.0, + "step": 23072 + }, + { + "epoch": 2.93512275791884, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.108942985534668, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8878620862960815, + "num_tokens": 895763381.0, + "step": 23073 + }, + { + "epoch": 2.9352499681974304, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.10888957977295, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.9040656089782715, + "num_tokens": 895798201.0, + "step": 23074 + }, + { + "epoch": 2.935377178476021, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.001812934875488, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8957492709159851, + "num_tokens": 895841018.0, + "step": 23075 + }, + { + "epoch": 2.9355043887546115, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2814998626708984e-05, + "grad_norm": 10.115656852722168, + "learning_rate": 1e-06, + "loss": 0.2631, + "mean_token_accuracy": 0.9094395041465759, + "num_tokens": 895875461.0, + "step": 23076 + }, + { + "epoch": 2.935631599033202, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.95974063873291, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.9059996008872986, + "num_tokens": 895914459.0, + "step": 23077 + }, + { + "epoch": 2.9357588093117926, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.137425422668457, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8926418423652649, + "num_tokens": 895951044.0, + "step": 23078 + }, + { + "epoch": 2.9358860195903826, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.165349006652832, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.886060893535614, + "num_tokens": 895989451.0, + "step": 23079 + }, + { + "epoch": 2.9360132298689736, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.061842918395996, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.891459584236145, + "num_tokens": 896028998.0, + "step": 23080 + }, + { + "epoch": 2.9361404401475637, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.082245826721191, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8913859128952026, + "num_tokens": 896067728.0, + "step": 23081 + }, + { + "epoch": 2.9362676504261547, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.063201904296875, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8982031941413879, + "num_tokens": 896102494.0, + "step": 23082 + }, + { + "epoch": 2.9363948607047448, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.024084091186523, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.9001864194869995, + "num_tokens": 896145220.0, + "step": 23083 + }, + { + "epoch": 2.9365220709833357, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.987245559692383, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9081201553344727, + "num_tokens": 896182757.0, + "step": 23084 + }, + { + "epoch": 2.936649281261926, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.157011985778809, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8867795467376709, + "num_tokens": 896218749.0, + "step": 23085 + }, + { + "epoch": 2.9367764915405163, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.99907398223877, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.904753565788269, + "num_tokens": 896257935.0, + "step": 23086 + }, + { + "epoch": 2.936903701819107, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.972708702087402, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8965533971786499, + "num_tokens": 896303694.0, + "step": 23087 + }, + { + "epoch": 2.9370309120976974, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.06065559387207, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8947545289993286, + "num_tokens": 896341096.0, + "step": 23088 + }, + { + "epoch": 2.937158122376288, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.025975227355957, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.9050149917602539, + "num_tokens": 896381086.0, + "step": 23089 + }, + { + "epoch": 2.9372853326548785, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.918238639831543, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.9069392681121826, + "num_tokens": 896423930.0, + "step": 23090 + }, + { + "epoch": 2.937412542933469, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.102282524108887, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8998103737831116, + "num_tokens": 896457206.0, + "step": 23091 + }, + { + "epoch": 2.9375397532120595, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.058094024658203, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8832637667655945, + "num_tokens": 896496725.0, + "step": 23092 + }, + { + "epoch": 2.93766696349065, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.029315948486328, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8997589349746704, + "num_tokens": 896535014.0, + "step": 23093 + }, + { + "epoch": 2.9377941737692406, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.030526161193848, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.909692645072937, + "num_tokens": 896574092.0, + "step": 23094 + }, + { + "epoch": 2.937921384047831, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.11620044708252, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8985515236854553, + "num_tokens": 896611874.0, + "step": 23095 + }, + { + "epoch": 2.9380485943264216, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.029760360717773, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.9034909009933472, + "num_tokens": 896649693.0, + "step": 23096 + }, + { + "epoch": 2.938175804605012, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.017319679260254, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8935686349868774, + "num_tokens": 896688891.0, + "step": 23097 + }, + { + "epoch": 2.9383030148836027, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.131321907043457, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8990094065666199, + "num_tokens": 896723102.0, + "step": 23098 + }, + { + "epoch": 2.938430225162193, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.107120513916016, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8857163190841675, + "num_tokens": 896765825.0, + "step": 23099 + }, + { + "epoch": 2.9385574354407837, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.993623733520508, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8925011157989502, + "num_tokens": 896806957.0, + "step": 23100 + }, + { + "epoch": 2.9386846457193743, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.11578369140625, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8901553750038147, + "num_tokens": 896843773.0, + "step": 23101 + }, + { + "epoch": 2.938811855997965, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.09380054473877, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8952103853225708, + "num_tokens": 896877460.0, + "step": 23102 + }, + { + "epoch": 2.9389390662765553, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.02632999420166, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8944019079208374, + "num_tokens": 896917094.0, + "step": 23103 + }, + { + "epoch": 2.9390662765551454, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.035273551940918, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8935003876686096, + "num_tokens": 896958773.0, + "step": 23104 + }, + { + "epoch": 2.9391934868337364, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.052974700927734, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8962428569793701, + "num_tokens": 896995368.0, + "step": 23105 + }, + { + "epoch": 2.9393206971123265, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.025211334228516, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8982773423194885, + "num_tokens": 897034966.0, + "step": 23106 + }, + { + "epoch": 2.9394479073909174, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.998196601867676, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9096101522445679, + "num_tokens": 897072216.0, + "step": 23107 + }, + { + "epoch": 2.9395751176695075, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.088973045349121, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8925040364265442, + "num_tokens": 897111902.0, + "step": 23108 + }, + { + "epoch": 2.939702327948098, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.104899406433105, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8937790393829346, + "num_tokens": 897150564.0, + "step": 23109 + }, + { + "epoch": 2.9398295382266886, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.087285041809082, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8985877633094788, + "num_tokens": 897188697.0, + "step": 23110 + }, + { + "epoch": 2.939956748505279, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.1837158203125, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8858433961868286, + "num_tokens": 897223547.0, + "step": 23111 + }, + { + "epoch": 2.9400839587838696, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.198657035827637, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.9074001312255859, + "num_tokens": 897257117.0, + "step": 23112 + }, + { + "epoch": 2.94021116906246, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.087316513061523, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8936376571655273, + "num_tokens": 897296136.0, + "step": 23113 + }, + { + "epoch": 2.9403383793410507, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.022751808166504, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.893182098865509, + "num_tokens": 897336248.0, + "step": 23114 + }, + { + "epoch": 2.940465589619641, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.137354850769043, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8922009468078613, + "num_tokens": 897373159.0, + "step": 23115 + }, + { + "epoch": 2.9405927998982317, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.034727096557617, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.901038646697998, + "num_tokens": 897414004.0, + "step": 23116 + }, + { + "epoch": 2.9407200101768223, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.040239334106445, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9117797613143921, + "num_tokens": 897447503.0, + "step": 23117 + }, + { + "epoch": 2.940847220455413, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 9.989413261413574, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9061053991317749, + "num_tokens": 897485580.0, + "step": 23118 + }, + { + "epoch": 2.9409744307340033, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.00937271118164, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.9006419777870178, + "num_tokens": 897532352.0, + "step": 23119 + }, + { + "epoch": 2.941101641012594, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.399369239807129, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.9007039666175842, + "num_tokens": 897565754.0, + "step": 23120 + }, + { + "epoch": 2.9412288512911844, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.973082542419434, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.9035593271255493, + "num_tokens": 897602587.0, + "step": 23121 + }, + { + "epoch": 2.941356061569775, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.07190227508545, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8901002407073975, + "num_tokens": 897639897.0, + "step": 23122 + }, + { + "epoch": 2.9414832718483654, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.048233032226562, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8986444473266602, + "num_tokens": 897677549.0, + "step": 23123 + }, + { + "epoch": 2.941610482126956, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.13913345336914, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8893750905990601, + "num_tokens": 897719757.0, + "step": 23124 + }, + { + "epoch": 2.9417376924055465, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.07712173461914, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8821683526039124, + "num_tokens": 897755193.0, + "step": 23125 + }, + { + "epoch": 2.941864902684137, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.047679901123047, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8942031860351562, + "num_tokens": 897796748.0, + "step": 23126 + }, + { + "epoch": 2.941992112962727, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.909928321838379, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9133398532867432, + "num_tokens": 897840944.0, + "step": 23127 + }, + { + "epoch": 2.942119323241318, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.00793743133545, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.9030622243881226, + "num_tokens": 897880539.0, + "step": 23128 + }, + { + "epoch": 2.942246533519908, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.023090362548828, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8802604079246521, + "num_tokens": 897924938.0, + "step": 23129 + }, + { + "epoch": 2.942373743798499, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.072698593139648, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.9029762148857117, + "num_tokens": 897965037.0, + "step": 23130 + }, + { + "epoch": 2.9425009540770892, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.243167877197266, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.886267900466919, + "num_tokens": 898002515.0, + "step": 23131 + }, + { + "epoch": 2.94262816435568, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.069901466369629, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8980532884597778, + "num_tokens": 898041276.0, + "step": 23132 + }, + { + "epoch": 2.9427553746342703, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.118046760559082, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8903776407241821, + "num_tokens": 898080261.0, + "step": 23133 + }, + { + "epoch": 2.942882584912861, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.050867080688477, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8945811986923218, + "num_tokens": 898117807.0, + "step": 23134 + }, + { + "epoch": 2.9430097951914513, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.085684776306152, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.902211606502533, + "num_tokens": 898155409.0, + "step": 23135 + }, + { + "epoch": 2.943137005470042, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.049309730529785, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.9036176204681396, + "num_tokens": 898196102.0, + "step": 23136 + }, + { + "epoch": 2.9432642157486324, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.11755084991455, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8910261392593384, + "num_tokens": 898232002.0, + "step": 23137 + }, + { + "epoch": 2.943391426027223, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.06324577331543, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9087608456611633, + "num_tokens": 898265148.0, + "step": 23138 + }, + { + "epoch": 2.9435186363058135, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.081339836120605, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.884746789932251, + "num_tokens": 898302810.0, + "step": 23139 + }, + { + "epoch": 2.943645846584404, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.122797966003418, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8893071413040161, + "num_tokens": 898344365.0, + "step": 23140 + }, + { + "epoch": 2.9437730568629945, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.951239585876465, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.9069889783859253, + "num_tokens": 898385630.0, + "step": 23141 + }, + { + "epoch": 2.943900267141585, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.083573341369629, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9071646928787231, + "num_tokens": 898425699.0, + "step": 23142 + }, + { + "epoch": 2.9440274774201756, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.111506462097168, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9073580503463745, + "num_tokens": 898462880.0, + "step": 23143 + }, + { + "epoch": 2.944154687698766, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.921073913574219, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9115445017814636, + "num_tokens": 898508238.0, + "step": 23144 + }, + { + "epoch": 2.9442818979773566, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.111631393432617, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.906718373298645, + "num_tokens": 898546841.0, + "step": 23145 + }, + { + "epoch": 2.944409108255947, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.056262016296387, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8957741856575012, + "num_tokens": 898578968.0, + "step": 23146 + }, + { + "epoch": 2.9445363185345377, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.240397453308105, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8826548457145691, + "num_tokens": 898614178.0, + "step": 23147 + }, + { + "epoch": 2.944663528813128, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.014054298400879, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.892972469329834, + "num_tokens": 898654540.0, + "step": 23148 + }, + { + "epoch": 2.9447907390917187, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.02133560180664, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.9007593393325806, + "num_tokens": 898695434.0, + "step": 23149 + }, + { + "epoch": 2.9449179493703093, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.089374542236328, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8986890316009521, + "num_tokens": 898734830.0, + "step": 23150 + }, + { + "epoch": 2.9450451596489, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.027449607849121, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.9010634422302246, + "num_tokens": 898773397.0, + "step": 23151 + }, + { + "epoch": 2.94517236992749, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 9.963342666625977, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8920333385467529, + "num_tokens": 898818660.0, + "step": 23152 + }, + { + "epoch": 2.945299580206081, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.03547477722168, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.9012126326560974, + "num_tokens": 898854162.0, + "step": 23153 + }, + { + "epoch": 2.945426790484671, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.045333862304688, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.9020549654960632, + "num_tokens": 898890575.0, + "step": 23154 + }, + { + "epoch": 2.945554000763262, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.034314155578613, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9054703116416931, + "num_tokens": 898929680.0, + "step": 23155 + }, + { + "epoch": 2.945681211041852, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.204461097717285, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8858715891838074, + "num_tokens": 898964878.0, + "step": 23156 + }, + { + "epoch": 2.945808421320443, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.042659759521484, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8926165699958801, + "num_tokens": 899004349.0, + "step": 23157 + }, + { + "epoch": 2.945935631599033, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.084364891052246, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9054528474807739, + "num_tokens": 899041640.0, + "step": 23158 + }, + { + "epoch": 2.9460628418776236, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.085021018981934, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9071414470672607, + "num_tokens": 899078958.0, + "step": 23159 + }, + { + "epoch": 2.946190052156214, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.03396224975586, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8988990783691406, + "num_tokens": 899118247.0, + "step": 23160 + }, + { + "epoch": 2.9463172624348046, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.057999610900879, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8925400376319885, + "num_tokens": 899163777.0, + "step": 23161 + }, + { + "epoch": 2.946444472713395, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.184298515319824, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8811802864074707, + "num_tokens": 899196200.0, + "step": 23162 + }, + { + "epoch": 2.9465716829919857, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.122052192687988, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8969503045082092, + "num_tokens": 899236100.0, + "step": 23163 + }, + { + "epoch": 2.946698893270576, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.038679122924805, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8840059041976929, + "num_tokens": 899278929.0, + "step": 23164 + }, + { + "epoch": 2.9468261035491667, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.094614028930664, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8942766189575195, + "num_tokens": 899318248.0, + "step": 23165 + }, + { + "epoch": 2.9469533138277573, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.139001846313477, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8883066773414612, + "num_tokens": 899356367.0, + "step": 23166 + }, + { + "epoch": 2.947080524106348, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.10573673248291, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9067681431770325, + "num_tokens": 899391902.0, + "step": 23167 + }, + { + "epoch": 2.9472077343849383, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.050529479980469, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.9057732224464417, + "num_tokens": 899427224.0, + "step": 23168 + }, + { + "epoch": 2.947334944663529, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.029458999633789, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8919713497161865, + "num_tokens": 899474831.0, + "step": 23169 + }, + { + "epoch": 2.9474621549421194, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.106532096862793, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.9067527651786804, + "num_tokens": 899514265.0, + "step": 23170 + }, + { + "epoch": 2.94758936522071, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.043618202209473, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8900835514068604, + "num_tokens": 899555587.0, + "step": 23171 + }, + { + "epoch": 2.9477165754993004, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.079626083374023, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8976153135299683, + "num_tokens": 899593940.0, + "step": 23172 + }, + { + "epoch": 2.947843785777891, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.165474891662598, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.902887225151062, + "num_tokens": 899628098.0, + "step": 23173 + }, + { + "epoch": 2.9479709960564815, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.140149116516113, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8963679075241089, + "num_tokens": 899667489.0, + "step": 23174 + }, + { + "epoch": 2.948098206335072, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.164462089538574, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8997400999069214, + "num_tokens": 899700954.0, + "step": 23175 + }, + { + "epoch": 2.9482254166136626, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.02724552154541, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9134689569473267, + "num_tokens": 899732898.0, + "step": 23176 + }, + { + "epoch": 2.9483526268922526, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.976712226867676, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.895038902759552, + "num_tokens": 899776198.0, + "step": 23177 + }, + { + "epoch": 2.9484798371708436, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.020933151245117, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.904501736164093, + "num_tokens": 899818261.0, + "step": 23178 + }, + { + "epoch": 2.9486070474494337, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.142118453979492, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8865036964416504, + "num_tokens": 899855442.0, + "step": 23179 + }, + { + "epoch": 2.9487342577280247, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.111480712890625, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.9006011486053467, + "num_tokens": 899887881.0, + "step": 23180 + }, + { + "epoch": 2.9488614680066147, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.963890075683594, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9125382900238037, + "num_tokens": 899927258.0, + "step": 23181 + }, + { + "epoch": 2.9489886782852053, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.045763969421387, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8938027620315552, + "num_tokens": 899969052.0, + "step": 23182 + }, + { + "epoch": 2.949115888563796, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.123327255249023, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8995741605758667, + "num_tokens": 900004107.0, + "step": 23183 + }, + { + "epoch": 2.9492430988423863, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.061189651489258, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8975452780723572, + "num_tokens": 900043597.0, + "step": 23184 + }, + { + "epoch": 2.949370309120977, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.11453914642334, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8949180841445923, + "num_tokens": 900077090.0, + "step": 23185 + }, + { + "epoch": 2.9494975193995674, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.075544357299805, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8917478322982788, + "num_tokens": 900119473.0, + "step": 23186 + }, + { + "epoch": 2.949624729678158, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.179560661315918, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8933566212654114, + "num_tokens": 900155350.0, + "step": 23187 + }, + { + "epoch": 2.9497519399567484, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.126581192016602, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8947936296463013, + "num_tokens": 900193407.0, + "step": 23188 + }, + { + "epoch": 2.949879150235339, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.035055160522461, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8974426984786987, + "num_tokens": 900233022.0, + "step": 23189 + }, + { + "epoch": 2.9500063605139295, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.098525047302246, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.9021264910697937, + "num_tokens": 900269836.0, + "step": 23190 + }, + { + "epoch": 2.95013357079252, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.110755920410156, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8969318270683289, + "num_tokens": 900310050.0, + "step": 23191 + }, + { + "epoch": 2.9502607810711106, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.974438667297363, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9104063510894775, + "num_tokens": 900344376.0, + "step": 23192 + }, + { + "epoch": 2.950387991349701, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.190536499023438, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8899998664855957, + "num_tokens": 900382521.0, + "step": 23193 + }, + { + "epoch": 2.9505152016282916, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.128411293029785, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9067163467407227, + "num_tokens": 900415619.0, + "step": 23194 + }, + { + "epoch": 2.950642411906882, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.017144203186035, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.9012479782104492, + "num_tokens": 900457350.0, + "step": 23195 + }, + { + "epoch": 2.9507696221854727, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.057761192321777, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.9002962112426758, + "num_tokens": 900500273.0, + "step": 23196 + }, + { + "epoch": 2.950896832464063, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.091872215270996, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8972849249839783, + "num_tokens": 900541598.0, + "step": 23197 + }, + { + "epoch": 2.9510240427426537, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.025603294372559, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9140751957893372, + "num_tokens": 900579755.0, + "step": 23198 + }, + { + "epoch": 2.9511512530212443, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.188612937927246, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8851763010025024, + "num_tokens": 900613569.0, + "step": 23199 + }, + { + "epoch": 2.951278463299835, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.05655288696289, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8918474912643433, + "num_tokens": 900649825.0, + "step": 23200 + }, + { + "epoch": 2.9514056735784253, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.023399353027344, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8993387818336487, + "num_tokens": 900691576.0, + "step": 23201 + }, + { + "epoch": 2.9515328838570154, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.924471855163574, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9036020636558533, + "num_tokens": 900736316.0, + "step": 23202 + }, + { + "epoch": 2.9516600941356064, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.0109224319458, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9086613655090332, + "num_tokens": 900777189.0, + "step": 23203 + }, + { + "epoch": 2.9517873044141965, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.162556648254395, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.902073860168457, + "num_tokens": 900811922.0, + "step": 23204 + }, + { + "epoch": 2.9519145146927874, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.009561538696289, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9096792936325073, + "num_tokens": 900854332.0, + "step": 23205 + }, + { + "epoch": 2.9520417249713775, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.03874683380127, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9114886522293091, + "num_tokens": 900893685.0, + "step": 23206 + }, + { + "epoch": 2.952168935249968, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.145956039428711, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.9021272659301758, + "num_tokens": 900929655.0, + "step": 23207 + }, + { + "epoch": 2.9522961455285586, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.167928695678711, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8791899085044861, + "num_tokens": 900962669.0, + "step": 23208 + }, + { + "epoch": 2.952423355807149, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.046125411987305, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9090975522994995, + "num_tokens": 901002594.0, + "step": 23209 + }, + { + "epoch": 2.9525505660857396, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.143084526062012, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8992105722427368, + "num_tokens": 901033146.0, + "step": 23210 + }, + { + "epoch": 2.95267777636433, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.026297569274902, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8951388001441956, + "num_tokens": 901080631.0, + "step": 23211 + }, + { + "epoch": 2.9528049866429207, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.018268585205078, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.9010310173034668, + "num_tokens": 901124394.0, + "step": 23212 + }, + { + "epoch": 2.952932196921511, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.125059127807617, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.9007951617240906, + "num_tokens": 901164534.0, + "step": 23213 + }, + { + "epoch": 2.9530594072001017, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.110461235046387, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8850153684616089, + "num_tokens": 901202388.0, + "step": 23214 + }, + { + "epoch": 2.9531866174786923, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.01405143737793, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8984540700912476, + "num_tokens": 901243418.0, + "step": 23215 + }, + { + "epoch": 2.953313827757283, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.009329795837402, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.9017912149429321, + "num_tokens": 901284624.0, + "step": 23216 + }, + { + "epoch": 2.9534410380358733, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.998394012451172, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.9047638177871704, + "num_tokens": 901331571.0, + "step": 23217 + }, + { + "epoch": 2.953568248314464, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.073015213012695, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8949592113494873, + "num_tokens": 901372561.0, + "step": 23218 + }, + { + "epoch": 2.9536954585930544, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.037782669067383, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8986619710922241, + "num_tokens": 901415466.0, + "step": 23219 + }, + { + "epoch": 2.953822668871645, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.097681045532227, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8997610807418823, + "num_tokens": 901451413.0, + "step": 23220 + }, + { + "epoch": 2.9539498791502354, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.076277732849121, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.9007610082626343, + "num_tokens": 901493636.0, + "step": 23221 + }, + { + "epoch": 2.954077089428826, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.13502025604248, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9086682200431824, + "num_tokens": 901523490.0, + "step": 23222 + }, + { + "epoch": 2.9542042997074165, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.177183151245117, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8971020579338074, + "num_tokens": 901560915.0, + "step": 23223 + }, + { + "epoch": 2.954331509986007, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.008133888244629, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9100388288497925, + "num_tokens": 901598248.0, + "step": 23224 + }, + { + "epoch": 2.954458720264597, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.08273983001709, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9032734632492065, + "num_tokens": 901631889.0, + "step": 23225 + }, + { + "epoch": 2.954585930543188, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.156920433044434, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8948293924331665, + "num_tokens": 901665278.0, + "step": 23226 + }, + { + "epoch": 2.954713140821778, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.084922790527344, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8901129364967346, + "num_tokens": 901708904.0, + "step": 23227 + }, + { + "epoch": 2.954840351100369, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.05186939239502, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.9043385982513428, + "num_tokens": 901747734.0, + "step": 23228 + }, + { + "epoch": 2.954967561378959, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.096570014953613, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8999112844467163, + "num_tokens": 901789976.0, + "step": 23229 + }, + { + "epoch": 2.95509477165755, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.059220314025879, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9046738147735596, + "num_tokens": 901827275.0, + "step": 23230 + }, + { + "epoch": 2.9552219819361403, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.045685768127441, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8950838446617126, + "num_tokens": 901865738.0, + "step": 23231 + }, + { + "epoch": 2.955349192214731, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.000102996826171875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.050596237182617, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8994400501251221, + "num_tokens": 901904618.0, + "step": 23232 + }, + { + "epoch": 2.9554764024933213, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.012481689453125, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9030697345733643, + "num_tokens": 901942153.0, + "step": 23233 + }, + { + "epoch": 2.955603612771912, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.146384239196777, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9137479066848755, + "num_tokens": 901973954.0, + "step": 23234 + }, + { + "epoch": 2.9557308230505024, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.18647289276123, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8989865779876709, + "num_tokens": 902007274.0, + "step": 23235 + }, + { + "epoch": 2.955858033329093, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.134856224060059, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8989695310592651, + "num_tokens": 902043244.0, + "step": 23236 + }, + { + "epoch": 2.9559852436076834, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.05280590057373, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9004249572753906, + "num_tokens": 902082888.0, + "step": 23237 + }, + { + "epoch": 2.956112453886274, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.08163070678711, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8838129043579102, + "num_tokens": 902122146.0, + "step": 23238 + }, + { + "epoch": 2.9562396641648645, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.151288032531738, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8984952569007874, + "num_tokens": 902154285.0, + "step": 23239 + }, + { + "epoch": 2.956366874443455, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.055229187011719, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.9028912782669067, + "num_tokens": 902187914.0, + "step": 23240 + }, + { + "epoch": 2.9564940847220456, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.127778053283691, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8995867967605591, + "num_tokens": 902223256.0, + "step": 23241 + }, + { + "epoch": 2.956621295000636, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.034032821655273, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9131112098693848, + "num_tokens": 902257794.0, + "step": 23242 + }, + { + "epoch": 2.9567485052792266, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.159320831298828, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.9021939039230347, + "num_tokens": 902295788.0, + "step": 23243 + }, + { + "epoch": 2.956875715557817, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.062981605529785, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8967098593711853, + "num_tokens": 902337757.0, + "step": 23244 + }, + { + "epoch": 2.9570029258364077, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.114473342895508, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.898120641708374, + "num_tokens": 902372909.0, + "step": 23245 + }, + { + "epoch": 2.957130136114998, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.19223690032959, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8908459544181824, + "num_tokens": 902408862.0, + "step": 23246 + }, + { + "epoch": 2.9572573463935887, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.098065376281738, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.9027010798454285, + "num_tokens": 902445014.0, + "step": 23247 + }, + { + "epoch": 2.9573845566721793, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.167098999023438, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8839704990386963, + "num_tokens": 902483712.0, + "step": 23248 + }, + { + "epoch": 2.95751176695077, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2874603271484375e-05, + "grad_norm": 10.035934448242188, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8852269649505615, + "num_tokens": 902527734.0, + "step": 23249 + }, + { + "epoch": 2.95763897722936, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.080528259277344, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8936103582382202, + "num_tokens": 902568960.0, + "step": 23250 + }, + { + "epoch": 2.957766187507951, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.1279296875, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.9042953848838806, + "num_tokens": 902602838.0, + "step": 23251 + }, + { + "epoch": 2.957893397786541, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.199793815612793, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8925690054893494, + "num_tokens": 902633878.0, + "step": 23252 + }, + { + "epoch": 2.958020608065132, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.994075775146484, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.9006203413009644, + "num_tokens": 902679047.0, + "step": 23253 + }, + { + "epoch": 2.958147818343722, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.069913864135742, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9161117672920227, + "num_tokens": 902714055.0, + "step": 23254 + }, + { + "epoch": 2.958275028622313, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.19483470916748, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.9027811288833618, + "num_tokens": 902748690.0, + "step": 23255 + }, + { + "epoch": 2.958402238900903, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.033714294433594, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8948667049407959, + "num_tokens": 902791644.0, + "step": 23256 + }, + { + "epoch": 2.9585294491794936, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.114607810974121, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.9070379137992859, + "num_tokens": 902829058.0, + "step": 23257 + }, + { + "epoch": 2.958656659458084, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.172724723815918, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9104868173599243, + "num_tokens": 902859913.0, + "step": 23258 + }, + { + "epoch": 2.9587838697366746, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.041324615478516, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8936905264854431, + "num_tokens": 902903424.0, + "step": 23259 + }, + { + "epoch": 2.958911080015265, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.076416969299316, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8987627029418945, + "num_tokens": 902941042.0, + "step": 23260 + }, + { + "epoch": 2.9590382902938557, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.11214542388916, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.9013877511024475, + "num_tokens": 902974261.0, + "step": 23261 + }, + { + "epoch": 2.959165500572446, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.102237701416016, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.897621750831604, + "num_tokens": 903016084.0, + "step": 23262 + }, + { + "epoch": 2.9592927108510367, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.13624382019043, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8954495191574097, + "num_tokens": 903052333.0, + "step": 23263 + }, + { + "epoch": 2.9594199211296273, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.03687572479248, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.9005306959152222, + "num_tokens": 903092321.0, + "step": 23264 + }, + { + "epoch": 2.959547131408218, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.136597633361816, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.9066656231880188, + "num_tokens": 903126158.0, + "step": 23265 + }, + { + "epoch": 2.9596743416868083, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.130996704101562, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8974233865737915, + "num_tokens": 903158877.0, + "step": 23266 + }, + { + "epoch": 2.959801551965399, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.187173843383789, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8944776058197021, + "num_tokens": 903193187.0, + "step": 23267 + }, + { + "epoch": 2.9599287622439894, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.047968864440918, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8957999348640442, + "num_tokens": 903235486.0, + "step": 23268 + }, + { + "epoch": 2.96005597252258, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.086119651794434, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8989232182502747, + "num_tokens": 903274865.0, + "step": 23269 + }, + { + "epoch": 2.9601831828011704, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.00961685180664, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9025232791900635, + "num_tokens": 903313520.0, + "step": 23270 + }, + { + "epoch": 2.960310393079761, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.127306938171387, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8847838640213013, + "num_tokens": 903361843.0, + "step": 23271 + }, + { + "epoch": 2.9604376033583515, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.095560073852539, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8843389749526978, + "num_tokens": 903405314.0, + "step": 23272 + }, + { + "epoch": 2.960564813636942, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.074958801269531, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8957708477973938, + "num_tokens": 903444975.0, + "step": 23273 + }, + { + "epoch": 2.9606920239155325, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.048019409179688, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.893103837966919, + "num_tokens": 903490070.0, + "step": 23274 + }, + { + "epoch": 2.9608192341941226, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.141000747680664, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8953008055686951, + "num_tokens": 903525380.0, + "step": 23275 + }, + { + "epoch": 2.9609464444727136, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.165565490722656, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.9049620628356934, + "num_tokens": 903563318.0, + "step": 23276 + }, + { + "epoch": 2.9610736547513037, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.006762504577637, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8962894678115845, + "num_tokens": 903606741.0, + "step": 23277 + }, + { + "epoch": 2.9612008650298947, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.169617652893066, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8996741771697998, + "num_tokens": 903642969.0, + "step": 23278 + }, + { + "epoch": 2.9613280753084847, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.01108455657959, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.9048116207122803, + "num_tokens": 903683004.0, + "step": 23279 + }, + { + "epoch": 2.9614552855870753, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.095819473266602, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.9056863784790039, + "num_tokens": 903715536.0, + "step": 23280 + }, + { + "epoch": 2.961582495865666, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.004616737365723, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.9025883674621582, + "num_tokens": 903756286.0, + "step": 23281 + }, + { + "epoch": 2.9617097061442563, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.137996673583984, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8868515491485596, + "num_tokens": 903795657.0, + "step": 23282 + }, + { + "epoch": 2.961836916422847, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.177055358886719, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8842331171035767, + "num_tokens": 903830884.0, + "step": 23283 + }, + { + "epoch": 2.9619641267014374, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.179365158081055, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8938647508621216, + "num_tokens": 903863354.0, + "step": 23284 + }, + { + "epoch": 2.962091336980028, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.12440013885498, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8935558199882507, + "num_tokens": 903909101.0, + "step": 23285 + }, + { + "epoch": 2.9622185472586184, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.026708602905273, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8931524753570557, + "num_tokens": 903950277.0, + "step": 23286 + }, + { + "epoch": 2.962345757537209, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.033289909362793, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8924202919006348, + "num_tokens": 903992007.0, + "step": 23287 + }, + { + "epoch": 2.9624729678157995, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.084152221679688, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8976154327392578, + "num_tokens": 904029678.0, + "step": 23288 + }, + { + "epoch": 2.96260017809439, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.050189018249512, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.9022517800331116, + "num_tokens": 904073894.0, + "step": 23289 + }, + { + "epoch": 2.9627273883729806, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.050024032592773, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9126507043838501, + "num_tokens": 904110152.0, + "step": 23290 + }, + { + "epoch": 2.962854598651571, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.148198127746582, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.900542140007019, + "num_tokens": 904146954.0, + "step": 23291 + }, + { + "epoch": 2.9629818089301616, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.99915599822998, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.9057002067565918, + "num_tokens": 904186611.0, + "step": 23292 + }, + { + "epoch": 2.963109019208752, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.124675750732422, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.9039721488952637, + "num_tokens": 904228888.0, + "step": 23293 + }, + { + "epoch": 2.9632362294873427, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.04849910736084, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8970540761947632, + "num_tokens": 904271653.0, + "step": 23294 + }, + { + "epoch": 2.963363439765933, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.136383056640625, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8952170610427856, + "num_tokens": 904307354.0, + "step": 23295 + }, + { + "epoch": 2.9634906500445237, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.037827491760254, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8994677662849426, + "num_tokens": 904350481.0, + "step": 23296 + }, + { + "epoch": 2.9636178603231143, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.94408893585205, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.904036283493042, + "num_tokens": 904392786.0, + "step": 23297 + }, + { + "epoch": 2.963745070601705, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.088789939880371, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.881528377532959, + "num_tokens": 904439169.0, + "step": 23298 + }, + { + "epoch": 2.9638722808802953, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.126956939697266, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8995705842971802, + "num_tokens": 904478074.0, + "step": 23299 + }, + { + "epoch": 2.9639994911588854, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.12679672241211, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.9022877216339111, + "num_tokens": 904510889.0, + "step": 23300 + }, + { + "epoch": 2.9641267014374764, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.166024208068848, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.9001628160476685, + "num_tokens": 904547547.0, + "step": 23301 + }, + { + "epoch": 2.9642539117160664, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.057039260864258, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.9012289047241211, + "num_tokens": 904584569.0, + "step": 23302 + }, + { + "epoch": 2.9643811219946574, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.149383544921875, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9119312763214111, + "num_tokens": 904621443.0, + "step": 23303 + }, + { + "epoch": 2.9645083322732475, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.0554780960083, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9091449975967407, + "num_tokens": 904654100.0, + "step": 23304 + }, + { + "epoch": 2.964635542551838, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.080556869506836, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8913702368736267, + "num_tokens": 904695307.0, + "step": 23305 + }, + { + "epoch": 2.9647627528304286, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.056288719177246, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9064232707023621, + "num_tokens": 904733498.0, + "step": 23306 + }, + { + "epoch": 2.964889963109019, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.046575546264648, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8961034417152405, + "num_tokens": 904772364.0, + "step": 23307 + }, + { + "epoch": 2.9650171733876096, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.16072940826416, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8794252872467041, + "num_tokens": 904809232.0, + "step": 23308 + }, + { + "epoch": 2.9651443836662, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.044709205627441, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8991982936859131, + "num_tokens": 904851544.0, + "step": 23309 + }, + { + "epoch": 2.9652715939447907, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.994428634643555, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9078425168991089, + "num_tokens": 904891978.0, + "step": 23310 + }, + { + "epoch": 2.965398804223381, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.11630916595459, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8897631168365479, + "num_tokens": 904934452.0, + "step": 23311 + }, + { + "epoch": 2.9655260145019717, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.091751098632812, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8928680419921875, + "num_tokens": 904973319.0, + "step": 23312 + }, + { + "epoch": 2.9656532247805623, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.14874267578125, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.9054784774780273, + "num_tokens": 905012158.0, + "step": 23313 + }, + { + "epoch": 2.965780435059153, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.02501392364502, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.9040512442588806, + "num_tokens": 905050863.0, + "step": 23314 + }, + { + "epoch": 2.9659076453377433, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.121878623962402, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.9001452326774597, + "num_tokens": 905089502.0, + "step": 23315 + }, + { + "epoch": 2.966034855616334, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.215353012084961, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8832684755325317, + "num_tokens": 905125136.0, + "step": 23316 + }, + { + "epoch": 2.9661620658949244, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.091249465942383, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8899480700492859, + "num_tokens": 905162039.0, + "step": 23317 + }, + { + "epoch": 2.966289276173515, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.180970191955566, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8804349899291992, + "num_tokens": 905199695.0, + "step": 23318 + }, + { + "epoch": 2.9664164864521054, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.145862579345703, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8964883089065552, + "num_tokens": 905231653.0, + "step": 23319 + }, + { + "epoch": 2.966543696730696, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.98612117767334, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9095028638839722, + "num_tokens": 905270834.0, + "step": 23320 + }, + { + "epoch": 2.9666709070092865, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.05284309387207, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.901561439037323, + "num_tokens": 905310639.0, + "step": 23321 + }, + { + "epoch": 2.966798117287877, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.10207462310791, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8954730033874512, + "num_tokens": 905347486.0, + "step": 23322 + }, + { + "epoch": 2.966925327566467, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.17660903930664, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.898536205291748, + "num_tokens": 905381034.0, + "step": 23323 + }, + { + "epoch": 2.967052537845058, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.136932373046875, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.9006146788597107, + "num_tokens": 905414915.0, + "step": 23324 + }, + { + "epoch": 2.967179748123648, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.0324068069458, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9106582403182983, + "num_tokens": 905449966.0, + "step": 23325 + }, + { + "epoch": 2.967306958402239, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.133722305297852, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8957645893096924, + "num_tokens": 905492457.0, + "step": 23326 + }, + { + "epoch": 2.967434168680829, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.017691612243652, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9132289886474609, + "num_tokens": 905534039.0, + "step": 23327 + }, + { + "epoch": 2.96756137895942, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.07027530670166, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8968765735626221, + "num_tokens": 905569537.0, + "step": 23328 + }, + { + "epoch": 2.9676885892380103, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.190226554870605, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8888369798660278, + "num_tokens": 905606912.0, + "step": 23329 + }, + { + "epoch": 2.967815799516601, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.99816608428955, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.9051806926727295, + "num_tokens": 905644883.0, + "step": 23330 + }, + { + "epoch": 2.9679430097951913, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.97684097290039, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9076710939407349, + "num_tokens": 905685654.0, + "step": 23331 + }, + { + "epoch": 2.968070220073782, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.000884056091309, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.9025370478630066, + "num_tokens": 905723574.0, + "step": 23332 + }, + { + "epoch": 2.9681974303523724, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.12268352508545, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.9051787257194519, + "num_tokens": 905755839.0, + "step": 23333 + }, + { + "epoch": 2.968324640630963, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.1339111328125, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.9062739610671997, + "num_tokens": 905789426.0, + "step": 23334 + }, + { + "epoch": 2.9684518509095534, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.098169326782227, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8977223634719849, + "num_tokens": 905824200.0, + "step": 23335 + }, + { + "epoch": 2.968579061188144, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.08779239654541, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8944534063339233, + "num_tokens": 905860047.0, + "step": 23336 + }, + { + "epoch": 2.9687062714667345, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.053074836730957, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9049910306930542, + "num_tokens": 905901067.0, + "step": 23337 + }, + { + "epoch": 2.968833481745325, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.020777702331543, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8950757384300232, + "num_tokens": 905938703.0, + "step": 23338 + }, + { + "epoch": 2.9689606920239155, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.05671215057373, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.9019861221313477, + "num_tokens": 905978425.0, + "step": 23339 + }, + { + "epoch": 2.969087902302506, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.111654281616211, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.89920973777771, + "num_tokens": 906015118.0, + "step": 23340 + }, + { + "epoch": 2.9692151125810966, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.253591537475586, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8849440813064575, + "num_tokens": 906048824.0, + "step": 23341 + }, + { + "epoch": 2.969342322859687, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.137240409851074, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8997236490249634, + "num_tokens": 906086242.0, + "step": 23342 + }, + { + "epoch": 2.9694695331382777, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.999298095703125, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8996498584747314, + "num_tokens": 906124715.0, + "step": 23343 + }, + { + "epoch": 2.969596743416868, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.074190139770508, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.9010316729545593, + "num_tokens": 906165619.0, + "step": 23344 + }, + { + "epoch": 2.9697239536954587, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.006752014160156, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.8985680341720581, + "num_tokens": 906205457.0, + "step": 23345 + }, + { + "epoch": 2.9698511639740492, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.101052284240723, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8987378478050232, + "num_tokens": 906246482.0, + "step": 23346 + }, + { + "epoch": 2.9699783742526398, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.120823860168457, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8914772272109985, + "num_tokens": 906285893.0, + "step": 23347 + }, + { + "epoch": 2.97010558453123, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.22368335723877, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8948055505752563, + "num_tokens": 906320785.0, + "step": 23348 + }, + { + "epoch": 2.970232794809821, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.074332237243652, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8975943922996521, + "num_tokens": 906363457.0, + "step": 23349 + }, + { + "epoch": 2.970360005088411, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.17431640625, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8832541704177856, + "num_tokens": 906405363.0, + "step": 23350 + }, + { + "epoch": 2.970487215367002, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.074862480163574, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.9101282954216003, + "num_tokens": 906440056.0, + "step": 23351 + }, + { + "epoch": 2.970614425645592, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.046113014221191, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8897296786308289, + "num_tokens": 906486085.0, + "step": 23352 + }, + { + "epoch": 2.970741635924183, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.285652160644531, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8904979228973389, + "num_tokens": 906523423.0, + "step": 23353 + }, + { + "epoch": 2.970868846202773, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.005799293518066, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8989266157150269, + "num_tokens": 906561720.0, + "step": 23354 + }, + { + "epoch": 2.9709960564813636, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.9912748336792, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.9059766530990601, + "num_tokens": 906598663.0, + "step": 23355 + }, + { + "epoch": 2.971123266759954, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.119219779968262, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9076387882232666, + "num_tokens": 906639560.0, + "step": 23356 + }, + { + "epoch": 2.9712504770385446, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.296906471252441, + "learning_rate": 1e-06, + "loss": 0.3969, + "mean_token_accuracy": 0.8742692470550537, + "num_tokens": 906673601.0, + "step": 23357 + }, + { + "epoch": 2.971377687317135, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.15856647491455, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.9034154415130615, + "num_tokens": 906708823.0, + "step": 23358 + }, + { + "epoch": 2.9715048975957257, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.980130195617676, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.9042317271232605, + "num_tokens": 906750085.0, + "step": 23359 + }, + { + "epoch": 2.971632107874316, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.097479820251465, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8933698534965515, + "num_tokens": 906790288.0, + "step": 23360 + }, + { + "epoch": 2.9717593181529067, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.055800437927246, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8888037800788879, + "num_tokens": 906830706.0, + "step": 23361 + }, + { + "epoch": 2.9718865284314973, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.039989471435547, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8888736367225647, + "num_tokens": 906864713.0, + "step": 23362 + }, + { + "epoch": 2.972013738710088, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.211381912231445, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.9001020193099976, + "num_tokens": 906898355.0, + "step": 23363 + }, + { + "epoch": 2.9721409489886783, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.16203498840332, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.88869309425354, + "num_tokens": 906929669.0, + "step": 23364 + }, + { + "epoch": 2.972268159267269, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.106947898864746, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8875724673271179, + "num_tokens": 906972565.0, + "step": 23365 + }, + { + "epoch": 2.9723953695458594, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.957077980041504, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8851086497306824, + "num_tokens": 907016631.0, + "step": 23366 + }, + { + "epoch": 2.97252257982445, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.979131698608398, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9065150022506714, + "num_tokens": 907056637.0, + "step": 23367 + }, + { + "epoch": 2.9726497901030404, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.021234512329102, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9080104231834412, + "num_tokens": 907094246.0, + "step": 23368 + }, + { + "epoch": 2.972777000381631, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.904666900634766, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9100313186645508, + "num_tokens": 907135837.0, + "step": 23369 + }, + { + "epoch": 2.9729042106602215, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.072209358215332, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9046576023101807, + "num_tokens": 907170864.0, + "step": 23370 + }, + { + "epoch": 2.973031420938812, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.101459503173828, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.9072321653366089, + "num_tokens": 907202250.0, + "step": 23371 + }, + { + "epoch": 2.9731586312174025, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.056801795959473, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.9063698053359985, + "num_tokens": 907241208.0, + "step": 23372 + }, + { + "epoch": 2.9732858414959926, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.075150489807129, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.903579831123352, + "num_tokens": 907276966.0, + "step": 23373 + }, + { + "epoch": 2.9734130517745836, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.142733573913574, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8925946950912476, + "num_tokens": 907314121.0, + "step": 23374 + }, + { + "epoch": 2.9735402620531737, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.055390357971191, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9085289239883423, + "num_tokens": 907348814.0, + "step": 23375 + }, + { + "epoch": 2.9736674723317646, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.101150512695312, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8968630433082581, + "num_tokens": 907383923.0, + "step": 23376 + }, + { + "epoch": 2.9737946826103547, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.113687515258789, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8936400413513184, + "num_tokens": 907422548.0, + "step": 23377 + }, + { + "epoch": 2.9739218928889453, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.099711418151855, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.9007526636123657, + "num_tokens": 907455527.0, + "step": 23378 + }, + { + "epoch": 2.974049103167536, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.05465030670166, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8985576629638672, + "num_tokens": 907489158.0, + "step": 23379 + }, + { + "epoch": 2.9741763134461263, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.06262493133545, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8932705521583557, + "num_tokens": 907527315.0, + "step": 23380 + }, + { + "epoch": 2.974303523724717, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010395050048828125, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 9.968528747558594, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8998535871505737, + "num_tokens": 907571410.0, + "step": 23381 + }, + { + "epoch": 2.9744307340033074, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.069328308105469, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9114437699317932, + "num_tokens": 907607596.0, + "step": 23382 + }, + { + "epoch": 2.974557944281898, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.16305923461914, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9046821594238281, + "num_tokens": 907640225.0, + "step": 23383 + }, + { + "epoch": 2.9746851545604884, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.109602928161621, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.875377357006073, + "num_tokens": 907681769.0, + "step": 23384 + }, + { + "epoch": 2.974812364839079, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.051352500915527, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8965615034103394, + "num_tokens": 907724478.0, + "step": 23385 + }, + { + "epoch": 2.9749395751176695, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.08005142211914, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8904558420181274, + "num_tokens": 907767667.0, + "step": 23386 + }, + { + "epoch": 2.97506678539626, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.044923782348633, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8868319988250732, + "num_tokens": 907810437.0, + "step": 23387 + }, + { + "epoch": 2.9751939956748505, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.045110702514648, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8913775086402893, + "num_tokens": 907849212.0, + "step": 23388 + }, + { + "epoch": 2.975321205953441, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.09762954711914, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8980027437210083, + "num_tokens": 907887122.0, + "step": 23389 + }, + { + "epoch": 2.9754484162320316, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.140844345092773, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8911884427070618, + "num_tokens": 907920736.0, + "step": 23390 + }, + { + "epoch": 2.975575626510622, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.132237434387207, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8901512622833252, + "num_tokens": 907958776.0, + "step": 23391 + }, + { + "epoch": 2.9757028367892127, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.109199523925781, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8900488615036011, + "num_tokens": 907994135.0, + "step": 23392 + }, + { + "epoch": 2.975830047067803, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.008403778076172, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8991496562957764, + "num_tokens": 908030859.0, + "step": 23393 + }, + { + "epoch": 2.9759572573463937, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.038456916809082, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8999767303466797, + "num_tokens": 908069497.0, + "step": 23394 + }, + { + "epoch": 2.9760844676249842, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.052741050720215, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8986611366271973, + "num_tokens": 908111139.0, + "step": 23395 + }, + { + "epoch": 2.9762116779035748, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.009931564331055, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.9030174016952515, + "num_tokens": 908147818.0, + "step": 23396 + }, + { + "epoch": 2.9763388881821653, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.067928314208984, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8999524116516113, + "num_tokens": 908184388.0, + "step": 23397 + }, + { + "epoch": 2.9764660984607554, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.132758140563965, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8888533115386963, + "num_tokens": 908219601.0, + "step": 23398 + }, + { + "epoch": 2.9765933087393464, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.049132347106934, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8996117115020752, + "num_tokens": 908260605.0, + "step": 23399 + }, + { + "epoch": 2.9767205190179364, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.984450340270996, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.9013333320617676, + "num_tokens": 908298734.0, + "step": 23400 + }, + { + "epoch": 2.9768477292965274, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.149465560913086, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8984853029251099, + "num_tokens": 908333375.0, + "step": 23401 + }, + { + "epoch": 2.9769749395751175, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.16784381866455, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8893672227859497, + "num_tokens": 908368637.0, + "step": 23402 + }, + { + "epoch": 2.977102149853708, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.111937522888184, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.9065832495689392, + "num_tokens": 908404004.0, + "step": 23403 + }, + { + "epoch": 2.9772293601322986, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.1121826171875, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8928699493408203, + "num_tokens": 908446010.0, + "step": 23404 + }, + { + "epoch": 2.977356570410889, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.008139610290527, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8979593515396118, + "num_tokens": 908484267.0, + "step": 23405 + }, + { + "epoch": 2.9774837806894796, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.116414070129395, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8884217739105225, + "num_tokens": 908527375.0, + "step": 23406 + }, + { + "epoch": 2.97761099096807, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.098082542419434, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9161908626556396, + "num_tokens": 908565250.0, + "step": 23407 + }, + { + "epoch": 2.9777382012466607, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.05689811706543, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.896723210811615, + "num_tokens": 908607733.0, + "step": 23408 + }, + { + "epoch": 2.977865411525251, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.068951606750488, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.9039896130561829, + "num_tokens": 908649737.0, + "step": 23409 + }, + { + "epoch": 2.9779926218038417, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.145548820495605, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8895450830459595, + "num_tokens": 908684162.0, + "step": 23410 + }, + { + "epoch": 2.9781198320824323, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.109467506408691, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8861484527587891, + "num_tokens": 908730349.0, + "step": 23411 + }, + { + "epoch": 2.978247042361023, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.104432106018066, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.9080556035041809, + "num_tokens": 908767349.0, + "step": 23412 + }, + { + "epoch": 2.9783742526396133, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.022198677062988, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.90746009349823, + "num_tokens": 908809417.0, + "step": 23413 + }, + { + "epoch": 2.978501462918204, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.006977081298828, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.9012267589569092, + "num_tokens": 908852451.0, + "step": 23414 + }, + { + "epoch": 2.9786286731967944, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.071859359741211, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.9019259214401245, + "num_tokens": 908892165.0, + "step": 23415 + }, + { + "epoch": 2.978755883475385, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.053173065185547, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8989410996437073, + "num_tokens": 908935897.0, + "step": 23416 + }, + { + "epoch": 2.9788830937539754, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.9966459274292, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.899412989616394, + "num_tokens": 908977750.0, + "step": 23417 + }, + { + "epoch": 2.979010304032566, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.132980346679688, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8847727179527283, + "num_tokens": 909023056.0, + "step": 23418 + }, + { + "epoch": 2.9791375143111565, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.076581001281738, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8927318453788757, + "num_tokens": 909061801.0, + "step": 23419 + }, + { + "epoch": 2.979264724589747, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.185405731201172, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8949325084686279, + "num_tokens": 909095932.0, + "step": 23420 + }, + { + "epoch": 2.979391934868337, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.020881652832031, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8973634839057922, + "num_tokens": 909135151.0, + "step": 23421 + }, + { + "epoch": 2.979519145146928, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.048454284667969, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8912245035171509, + "num_tokens": 909171394.0, + "step": 23422 + }, + { + "epoch": 2.979646355425518, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.11569595336914, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.897753894329071, + "num_tokens": 909214350.0, + "step": 23423 + }, + { + "epoch": 2.979773565704109, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.143153190612793, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8894239664077759, + "num_tokens": 909250826.0, + "step": 23424 + }, + { + "epoch": 2.979900775982699, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.14661693572998, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8908594846725464, + "num_tokens": 909294452.0, + "step": 23425 + }, + { + "epoch": 2.98002798626129, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.095070838928223, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8907594084739685, + "num_tokens": 909336404.0, + "step": 23426 + }, + { + "epoch": 2.9801551965398803, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.128280639648438, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.9005986452102661, + "num_tokens": 909372692.0, + "step": 23427 + }, + { + "epoch": 2.980282406818471, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 9.978975296020508, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9070001840591431, + "num_tokens": 909418165.0, + "step": 23428 + }, + { + "epoch": 2.9804096170970613, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.1069974899292, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8918387293815613, + "num_tokens": 909458752.0, + "step": 23429 + }, + { + "epoch": 2.980536827375652, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.122480392456055, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.9017061591148376, + "num_tokens": 909498745.0, + "step": 23430 + }, + { + "epoch": 2.9806640376542424, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.960286140441895, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9112793803215027, + "num_tokens": 909539970.0, + "step": 23431 + }, + { + "epoch": 2.980791247932833, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.951955795288086, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.9047696590423584, + "num_tokens": 909588758.0, + "step": 23432 + }, + { + "epoch": 2.9809184582114234, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.171375274658203, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.899351954460144, + "num_tokens": 909630314.0, + "step": 23433 + }, + { + "epoch": 2.981045668490014, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.148924827575684, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8944169282913208, + "num_tokens": 909668773.0, + "step": 23434 + }, + { + "epoch": 2.9811728787686045, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.119967460632324, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.9054856300354004, + "num_tokens": 909703993.0, + "step": 23435 + }, + { + "epoch": 2.981300089047195, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.04372787475586, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8972753882408142, + "num_tokens": 909745116.0, + "step": 23436 + }, + { + "epoch": 2.9814272993257855, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.037925720214844, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.905089259147644, + "num_tokens": 909782380.0, + "step": 23437 + }, + { + "epoch": 2.981554509604376, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.051979064941406, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.9000585675239563, + "num_tokens": 909824095.0, + "step": 23438 + }, + { + "epoch": 2.9816817198829666, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.230046272277832, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8805777430534363, + "num_tokens": 909858757.0, + "step": 23439 + }, + { + "epoch": 2.981808930161557, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.258894920349121, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8886590003967285, + "num_tokens": 909893847.0, + "step": 23440 + }, + { + "epoch": 2.9819361404401477, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.075318336486816, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8961537480354309, + "num_tokens": 909929863.0, + "step": 23441 + }, + { + "epoch": 2.982063350718738, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.99272632598877, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8980503082275391, + "num_tokens": 909973909.0, + "step": 23442 + }, + { + "epoch": 2.9821905609973287, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.052847862243652, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9126237630844116, + "num_tokens": 910009653.0, + "step": 23443 + }, + { + "epoch": 2.9823177712759192, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.10842514038086, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8999091386795044, + "num_tokens": 910048027.0, + "step": 23444 + }, + { + "epoch": 2.9824449815545098, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.145357131958008, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.9014467000961304, + "num_tokens": 910081643.0, + "step": 23445 + }, + { + "epoch": 2.9825721918331, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.116385459899902, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.9034652709960938, + "num_tokens": 910119937.0, + "step": 23446 + }, + { + "epoch": 2.982699402111691, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.067951202392578, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8945119380950928, + "num_tokens": 910161663.0, + "step": 23447 + }, + { + "epoch": 2.982826612390281, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.03691577911377, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8985191583633423, + "num_tokens": 910210015.0, + "step": 23448 + }, + { + "epoch": 2.982953822668872, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.105835914611816, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8972580432891846, + "num_tokens": 910246754.0, + "step": 23449 + }, + { + "epoch": 2.983081032947462, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.99610710144043, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8956246376037598, + "num_tokens": 910292763.0, + "step": 23450 + }, + { + "epoch": 2.983208243226053, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.036558151245117, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8951246738433838, + "num_tokens": 910335757.0, + "step": 23451 + }, + { + "epoch": 2.983335453504643, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.005586624145508, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.9059288501739502, + "num_tokens": 910377653.0, + "step": 23452 + }, + { + "epoch": 2.9834626637832335, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.03123664855957, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8971974849700928, + "num_tokens": 910419641.0, + "step": 23453 + }, + { + "epoch": 2.983589874061824, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.053742408752441, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9092236161231995, + "num_tokens": 910457068.0, + "step": 23454 + }, + { + "epoch": 2.9837170843404146, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.08834457397461, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.906771719455719, + "num_tokens": 910493918.0, + "step": 23455 + }, + { + "epoch": 2.983844294619005, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.023085594177246, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.9019980430603027, + "num_tokens": 910534962.0, + "step": 23456 + }, + { + "epoch": 2.9839715048975957, + "ewc_loss": 0.022705078125, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.089411735534668, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8939016461372375, + "num_tokens": 910571721.0, + "step": 23457 + }, + { + "epoch": 2.984098715176186, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.133869171142578, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8803904056549072, + "num_tokens": 910612078.0, + "step": 23458 + }, + { + "epoch": 2.9842259254547767, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.985579490661621, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.9013344049453735, + "num_tokens": 910657168.0, + "step": 23459 + }, + { + "epoch": 2.9843531357333672, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.041803359985352, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.9008270502090454, + "num_tokens": 910698406.0, + "step": 23460 + }, + { + "epoch": 2.9844803460119578, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.042915344238281, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.9018135070800781, + "num_tokens": 910740765.0, + "step": 23461 + }, + { + "epoch": 2.9846075562905483, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.132033348083496, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9044092893600464, + "num_tokens": 910772989.0, + "step": 23462 + }, + { + "epoch": 2.984734766569139, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.082063674926758, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8991457223892212, + "num_tokens": 910811519.0, + "step": 23463 + }, + { + "epoch": 2.9848619768477294, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.111631393432617, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8957967162132263, + "num_tokens": 910852737.0, + "step": 23464 + }, + { + "epoch": 2.98498918712632, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.091679573059082, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8936436772346497, + "num_tokens": 910890749.0, + "step": 23465 + }, + { + "epoch": 2.9851163974049104, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.02721881866455, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.9012613296508789, + "num_tokens": 910933895.0, + "step": 23466 + }, + { + "epoch": 2.985243607683501, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.159636497497559, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.9015500545501709, + "num_tokens": 910968021.0, + "step": 23467 + }, + { + "epoch": 2.9853708179620915, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.984277725219727, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9032002687454224, + "num_tokens": 911009892.0, + "step": 23468 + }, + { + "epoch": 2.985498028240682, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.125645637512207, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9096415042877197, + "num_tokens": 911045788.0, + "step": 23469 + }, + { + "epoch": 2.9856252385192725, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.030309677124023, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.9045501947402954, + "num_tokens": 911084812.0, + "step": 23470 + }, + { + "epoch": 2.9857524487978626, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.049577713012695, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.9100950360298157, + "num_tokens": 911122664.0, + "step": 23471 + }, + { + "epoch": 2.9858796590764536, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.113140106201172, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8910285234451294, + "num_tokens": 911164061.0, + "step": 23472 + }, + { + "epoch": 2.9860068693550437, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.099924087524414, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8965752124786377, + "num_tokens": 911201269.0, + "step": 23473 + }, + { + "epoch": 2.9861340796336346, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.00010442733764648438, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.22102165222168, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8882283568382263, + "num_tokens": 911238045.0, + "step": 23474 + }, + { + "epoch": 2.9862612899122247, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.001359939575195, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8913302421569824, + "num_tokens": 911279102.0, + "step": 23475 + }, + { + "epoch": 2.9863885001908153, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.045510292053223, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.9001344442367554, + "num_tokens": 911317881.0, + "step": 23476 + }, + { + "epoch": 2.986515710469406, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.109156608581543, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8935796022415161, + "num_tokens": 911357890.0, + "step": 23477 + }, + { + "epoch": 2.9866429207479963, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.991655349731445, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9106199145317078, + "num_tokens": 911399918.0, + "step": 23478 + }, + { + "epoch": 2.986770131026587, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.048907279968262, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8959344029426575, + "num_tokens": 911440451.0, + "step": 23479 + }, + { + "epoch": 2.9868973413051774, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.04811954498291, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.9023144245147705, + "num_tokens": 911480235.0, + "step": 23480 + }, + { + "epoch": 2.987024551583768, + "ewc_loss": 0.0228271484375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.233816146850586e-05, + "grad_norm": 10.184720039367676, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.9014102220535278, + "num_tokens": 911514962.0, + "step": 23481 + }, + { + "epoch": 2.9871517618623584, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.085165977478027, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8811094760894775, + "num_tokens": 911552401.0, + "step": 23482 + }, + { + "epoch": 2.987278972140949, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.11880111694336, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8999049663543701, + "num_tokens": 911589305.0, + "step": 23483 + }, + { + "epoch": 2.9874061824195395, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.991917610168457, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8919133543968201, + "num_tokens": 911629657.0, + "step": 23484 + }, + { + "epoch": 2.98753339269813, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.135061264038086, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.9058403968811035, + "num_tokens": 911664349.0, + "step": 23485 + }, + { + "epoch": 2.9876606029767205, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.996540069580078, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.905280590057373, + "num_tokens": 911700982.0, + "step": 23486 + }, + { + "epoch": 2.987787813255311, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.085638046264648, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.9021800756454468, + "num_tokens": 911740813.0, + "step": 23487 + }, + { + "epoch": 2.9879150235339016, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.110963821411133, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8869076371192932, + "num_tokens": 911778351.0, + "step": 23488 + }, + { + "epoch": 2.988042233812492, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.06340503692627, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.9046449661254883, + "num_tokens": 911815043.0, + "step": 23489 + }, + { + "epoch": 2.9881694440910826, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.063966751098633, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8955687284469604, + "num_tokens": 911853294.0, + "step": 23490 + }, + { + "epoch": 2.988296654369673, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.114394187927246, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.9082762002944946, + "num_tokens": 911886593.0, + "step": 23491 + }, + { + "epoch": 2.9884238646482637, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.042143821716309, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.9040356278419495, + "num_tokens": 911925780.0, + "step": 23492 + }, + { + "epoch": 2.9885510749268542, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.021929740905762, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8988431096076965, + "num_tokens": 911969196.0, + "step": 23493 + }, + { + "epoch": 2.9886782852054448, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.051573753356934, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.899455189704895, + "num_tokens": 912011163.0, + "step": 23494 + }, + { + "epoch": 2.9888054954840353, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.064774513244629, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8986415863037109, + "num_tokens": 912055327.0, + "step": 23495 + }, + { + "epoch": 2.9889327057626254, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.140267372131348, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8897373080253601, + "num_tokens": 912096518.0, + "step": 23496 + }, + { + "epoch": 2.9890599160412163, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.083878517150879, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8900632262229919, + "num_tokens": 912137500.0, + "step": 23497 + }, + { + "epoch": 2.9891871263198064, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.146004676818848, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.893674373626709, + "num_tokens": 912174044.0, + "step": 23498 + }, + { + "epoch": 2.9893143365983974, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.194836616516113, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9062848687171936, + "num_tokens": 912205857.0, + "step": 23499 + }, + { + "epoch": 2.9894415468769875, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.142715454101562, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.9003595113754272, + "num_tokens": 912241681.0, + "step": 23500 + }, + { + "epoch": 2.989568757155578, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.074687957763672, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8989335298538208, + "num_tokens": 912284803.0, + "step": 23501 + }, + { + "epoch": 2.9896959674341685, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.062416076660156, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.9046735763549805, + "num_tokens": 912325371.0, + "step": 23502 + }, + { + "epoch": 2.989823177712759, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.083457946777344, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.889771044254303, + "num_tokens": 912369730.0, + "step": 23503 + }, + { + "epoch": 2.9899503879913496, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.98798656463623, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9101278781890869, + "num_tokens": 912412413.0, + "step": 23504 + }, + { + "epoch": 2.99007759826994, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.102051734924316, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9031800031661987, + "num_tokens": 912445676.0, + "step": 23505 + }, + { + "epoch": 2.9902048085485307, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.09072494506836, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8959741592407227, + "num_tokens": 912489919.0, + "step": 23506 + }, + { + "epoch": 2.990332018827121, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.077239990234375, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8934627771377563, + "num_tokens": 912529808.0, + "step": 23507 + }, + { + "epoch": 2.9904592291057117, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.120712280273438, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8985664248466492, + "num_tokens": 912572573.0, + "step": 23508 + }, + { + "epoch": 2.9905864393843022, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.092185974121094, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8919982314109802, + "num_tokens": 912609886.0, + "step": 23509 + }, + { + "epoch": 2.9907136496628928, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.177666664123535, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8988726735115051, + "num_tokens": 912641863.0, + "step": 23510 + }, + { + "epoch": 2.9908408599414833, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.018614768981934, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8997167348861694, + "num_tokens": 912684302.0, + "step": 23511 + }, + { + "epoch": 2.990968070220074, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.016858100891113, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9121526479721069, + "num_tokens": 912725501.0, + "step": 23512 + }, + { + "epoch": 2.9910952804986644, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.057247161865234, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.896608829498291, + "num_tokens": 912770660.0, + "step": 23513 + }, + { + "epoch": 2.991222490777255, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.038168907165527, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.892898440361023, + "num_tokens": 912813955.0, + "step": 23514 + }, + { + "epoch": 2.9913497010558454, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.128207206726074, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8991079926490784, + "num_tokens": 912853430.0, + "step": 23515 + }, + { + "epoch": 2.991476911334436, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.026798248291016, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8973492383956909, + "num_tokens": 912894487.0, + "step": 23516 + }, + { + "epoch": 2.9916041216130265, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.075161933898926, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8913453817367554, + "num_tokens": 912935467.0, + "step": 23517 + }, + { + "epoch": 2.991731331891617, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 9.98855972290039, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.895474374294281, + "num_tokens": 912977988.0, + "step": 23518 + }, + { + "epoch": 2.991858542170207, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.082042694091797, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.904815673828125, + "num_tokens": 913013427.0, + "step": 23519 + }, + { + "epoch": 2.991985752448798, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.125600814819336, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8978538513183594, + "num_tokens": 913050185.0, + "step": 23520 + }, + { + "epoch": 2.992112962727388, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.047771453857422, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8990615606307983, + "num_tokens": 913089760.0, + "step": 23521 + }, + { + "epoch": 2.992240173005979, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.04156494140625, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9013354778289795, + "num_tokens": 913127548.0, + "step": 23522 + }, + { + "epoch": 2.992367383284569, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.002218246459961, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8950245976448059, + "num_tokens": 913176219.0, + "step": 23523 + }, + { + "epoch": 2.99249459356316, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.065999031066895, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8931680917739868, + "num_tokens": 913214180.0, + "step": 23524 + }, + { + "epoch": 2.9926218038417502, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.245737075805664e-05, + "grad_norm": 10.276556015014648, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8887313604354858, + "num_tokens": 913243683.0, + "step": 23525 + }, + { + "epoch": 2.992749014120341, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.974803924560547, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.9016584157943726, + "num_tokens": 913285075.0, + "step": 23526 + }, + { + "epoch": 2.9928762243989313, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.18805980682373, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8939805626869202, + "num_tokens": 913319784.0, + "step": 23527 + }, + { + "epoch": 2.993003434677522, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.004464149475098, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8971123099327087, + "num_tokens": 913361389.0, + "step": 23528 + }, + { + "epoch": 2.9931306449561124, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.23107624053955, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8915280699729919, + "num_tokens": 913393939.0, + "step": 23529 + }, + { + "epoch": 2.993257855234703, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.131292343139648, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8793759942054749, + "num_tokens": 913435025.0, + "step": 23530 + }, + { + "epoch": 2.9933850655132934, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.011946678161621, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.9019631147384644, + "num_tokens": 913475122.0, + "step": 23531 + }, + { + "epoch": 2.993512275791884, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.047865867614746, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.906852126121521, + "num_tokens": 913509426.0, + "step": 23532 + }, + { + "epoch": 2.9936394860704745, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.144951820373535, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9103041887283325, + "num_tokens": 913537991.0, + "step": 23533 + }, + { + "epoch": 2.993766696349065, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.025453567504883, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8947162628173828, + "num_tokens": 913581127.0, + "step": 23534 + }, + { + "epoch": 2.9938939066276555, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.013419151306152, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.9011234045028687, + "num_tokens": 913625397.0, + "step": 23535 + }, + { + "epoch": 2.994021116906246, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.068037033081055, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8992839455604553, + "num_tokens": 913670245.0, + "step": 23536 + }, + { + "epoch": 2.9941483271848366, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.097383499145508, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.897936224937439, + "num_tokens": 913706904.0, + "step": 23537 + }, + { + "epoch": 2.994275537463427, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.054374694824219, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.9042102694511414, + "num_tokens": 913744635.0, + "step": 23538 + }, + { + "epoch": 2.9944027477420176, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.1224946975708, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8959904909133911, + "num_tokens": 913784055.0, + "step": 23539 + }, + { + "epoch": 2.994529958020608, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.032906532287598, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9046696424484253, + "num_tokens": 913821893.0, + "step": 23540 + }, + { + "epoch": 2.9946571682991987, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.198333740234375, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8970454931259155, + "num_tokens": 913859468.0, + "step": 23541 + }, + { + "epoch": 2.9947843785777892, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.038025856018066, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.9007436037063599, + "num_tokens": 913897855.0, + "step": 23542 + }, + { + "epoch": 2.9949115888563798, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.013367652893066, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9141817688941956, + "num_tokens": 913938196.0, + "step": 23543 + }, + { + "epoch": 2.99503879913497, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.140113830566406, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.9033029675483704, + "num_tokens": 913973858.0, + "step": 23544 + }, + { + "epoch": 2.995166009413561, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.088671684265137, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8973419666290283, + "num_tokens": 914013956.0, + "step": 23545 + }, + { + "epoch": 2.995293219692151, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.091723442077637, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8868986964225769, + "num_tokens": 914055266.0, + "step": 23546 + }, + { + "epoch": 2.995420429970742, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.18989086151123, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8929728269577026, + "num_tokens": 914088812.0, + "step": 23547 + }, + { + "epoch": 2.995547640249332, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.077693939208984, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.9006768465042114, + "num_tokens": 914125319.0, + "step": 23548 + }, + { + "epoch": 2.995674850527923, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 9.97985553741455, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8979406952857971, + "num_tokens": 914174901.0, + "step": 23549 + }, + { + "epoch": 2.995802060806513, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.186627388000488, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9089046716690063, + "num_tokens": 914205605.0, + "step": 23550 + }, + { + "epoch": 2.9959292710851035, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 9.988641738891602, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.9011525511741638, + "num_tokens": 914248138.0, + "step": 23551 + }, + { + "epoch": 2.996056481363694, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.096015930175781, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8958659172058105, + "num_tokens": 914289099.0, + "step": 23552 + }, + { + "epoch": 2.9961836916422846, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.053021430969238, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8933183550834656, + "num_tokens": 914331034.0, + "step": 23553 + }, + { + "epoch": 2.996310901920875, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.106734275817871, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8992134928703308, + "num_tokens": 914366719.0, + "step": 23554 + }, + { + "epoch": 2.9964381121994657, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.047990798950195, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9047747850418091, + "num_tokens": 914403732.0, + "step": 23555 + }, + { + "epoch": 2.996565322478056, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.05592155456543, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8895965814590454, + "num_tokens": 914447350.0, + "step": 23556 + }, + { + "epoch": 2.9966925327566467, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.106778144836426, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8989883661270142, + "num_tokens": 914487387.0, + "step": 23557 + }, + { + "epoch": 2.9968197430352372, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.07193660736084, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8952611684799194, + "num_tokens": 914530606.0, + "step": 23558 + }, + { + "epoch": 2.9969469533138278, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.12504768371582, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.9076786041259766, + "num_tokens": 914565520.0, + "step": 23559 + }, + { + "epoch": 2.9970741635924183, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.098711967468262, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8894885182380676, + "num_tokens": 914606130.0, + "step": 23560 + }, + { + "epoch": 2.997201373871009, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.129913330078125, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.9025044441223145, + "num_tokens": 914642846.0, + "step": 23561 + }, + { + "epoch": 2.9973285841495994, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.072603225708008, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.892062783241272, + "num_tokens": 914682285.0, + "step": 23562 + }, + { + "epoch": 2.99745579442819, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.239776611328125e-05, + "grad_norm": 10.081365585327148, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.9040795564651489, + "num_tokens": 914716403.0, + "step": 23563 + }, + { + "epoch": 2.9975830047067804, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.140814781188965, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8950600624084473, + "num_tokens": 914755226.0, + "step": 23564 + }, + { + "epoch": 2.997710214985371, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.034354209899902, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9097540974617004, + "num_tokens": 914795943.0, + "step": 23565 + }, + { + "epoch": 2.9978374252639615, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.015095710754395, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8979644775390625, + "num_tokens": 914837108.0, + "step": 23566 + }, + { + "epoch": 2.997964635542552, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.13001823425293, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.898222804069519, + "num_tokens": 914872625.0, + "step": 23567 + }, + { + "epoch": 2.9980918458211425, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.02276611328125, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8994048237800598, + "num_tokens": 914910740.0, + "step": 23568 + }, + { + "epoch": 2.9982190560997326, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.108736038208008, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8963660001754761, + "num_tokens": 914952527.0, + "step": 23569 + }, + { + "epoch": 2.9983462663783236, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.092801094055176, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8967554569244385, + "num_tokens": 914986312.0, + "step": 23570 + }, + { + "epoch": 2.9984734766569137, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.186219215393066, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.9053424000740051, + "num_tokens": 915019484.0, + "step": 23571 + }, + { + "epoch": 2.9986006869355046, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.064983367919922, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.9015262722969055, + "num_tokens": 915055441.0, + "step": 23572 + }, + { + "epoch": 2.9987278972140947, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.135050773620605, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8936878442764282, + "num_tokens": 915091868.0, + "step": 23573 + }, + { + "epoch": 2.9988551074926852, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.14625358581543, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8972337245941162, + "num_tokens": 915123666.0, + "step": 23574 + }, + { + "epoch": 2.9989823177712758, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.082910537719727, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8854902982711792, + "num_tokens": 915160606.0, + "step": 23575 + }, + { + "epoch": 2.9991095280498663, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2755393981933594e-05, + "grad_norm": 10.056449890136719, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9112558364868164, + "num_tokens": 915196683.0, + "step": 23576 + }, + { + "epoch": 2.999236738328457, + "ewc_loss": 0.0230712890625, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2576580047607422e-05, + "grad_norm": 10.064361572265625, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8972872495651245, + "num_tokens": 915233509.0, + "step": 23577 + }, + { + "epoch": 2.9993639486070474, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.03980827331543, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8987445831298828, + "num_tokens": 915275205.0, + "step": 23578 + }, + { + "epoch": 2.999491158885638, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.03858757019043, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8984217643737793, + "num_tokens": 915315705.0, + "step": 23579 + }, + { + "epoch": 2.9996183691642284, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2695789337158203e-05, + "grad_norm": 10.0833158493042, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9091446995735168, + "num_tokens": 915357126.0, + "step": 23580 + }, + { + "epoch": 2.999745579442819, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.062841415405273, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8946083188056946, + "num_tokens": 915401645.0, + "step": 23581 + }, + { + "epoch": 2.9998727897214095, + "ewc_loss": 0.02294921875, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2516975402832031e-05, + "grad_norm": 10.111220359802246, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8945114612579346, + "num_tokens": 915444729.0, + "step": 23582 + }, + { + "epoch": 3.0, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "grad_norm": 10.113143920898438, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.9037442207336426, + "num_tokens": 915485630.0, + "step": 23583 + }, + { + "epoch": 3.0, + "ewc_loss": 0.023193359375, + "ewc_loss_diag": 0.0001049041748046875, + "ewc_loss_parallel": 1.2636184692382812e-05, + "step": 23583, + "total_flos": 5.751448627479105e+19, + "train_loss": 0.3408017477382914, + "train_runtime": 50265.1743, + "train_samples_per_second": 7.506, + "train_steps_per_second": 0.469 + } + ], + "logging_steps": 1, + "max_steps": 23583, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 11792, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 5.751448627479105e+19, + "train_batch_size": 4, + "trial_name": null, + "trial_params": null +} diff --git a/training_args.bin b/training_args.bin new file mode 100644 index 0000000..7b732c8 --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3d392e6e1a9aef2a68420e9fa8b727240403a0240e752bc0e591ff818bc3a8fa +size 12689